diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/last-checkpoint/trainer_state.json"
@@ -0,0 +1,38831 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999977398116359,
+  "eval_steps": 5000,
+  "global_step": 55305,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00018081506912786112,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.999971747693341e-06,
+      "loss": 17.52,
+      "step": 10
+    },
+    {
+      "epoch": 0.00036163013825572224,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.999943495386682e-06,
+      "loss": 17.8709,
+      "step": 20
+    },
+    {
+      "epoch": 0.0005424452073835833,
+      "grad_norm": 40.875,
+      "learning_rate": 9.999915243080021e-06,
+      "loss": 17.6916,
+      "step": 30
+    },
+    {
+      "epoch": 0.0007232602765114445,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.999886990773362e-06,
+      "loss": 17.9391,
+      "step": 40
+    },
+    {
+      "epoch": 0.0009040753456393055,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.999858738466704e-06,
+      "loss": 17.3349,
+      "step": 50
+    },
+    {
+      "epoch": 0.0010848904147671666,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.999830486160043e-06,
+      "loss": 17.7175,
+      "step": 60
+    },
+    {
+      "epoch": 0.0012657054838950278,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.999802233853383e-06,
+      "loss": 17.9345,
+      "step": 70
+    },
+    {
+      "epoch": 0.001446520553022889,
+      "grad_norm": 41.875,
+      "learning_rate": 9.999773981546724e-06,
+      "loss": 17.5544,
+      "step": 80
+    },
+    {
+      "epoch": 0.0016273356221507501,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.999745729240065e-06,
+      "loss": 17.255,
+      "step": 90
+    },
+    {
+      "epoch": 0.001808150691278611,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.999717476933405e-06,
+      "loss": 17.2562,
+      "step": 100
+    },
+    {
+      "epoch": 0.0019889657604064723,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.999689224626746e-06,
+      "loss": 17.4383,
+      "step": 110
+    },
+    {
+      "epoch": 0.002169780829534333,
+      "grad_norm": 41.75,
+      "learning_rate": 9.999660972320085e-06,
+      "loss": 17.9809,
+      "step": 120
+    },
+    {
+      "epoch": 0.0023505958986621946,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.999632720013426e-06,
+      "loss": 17.7502,
+      "step": 130
+    },
+    {
+      "epoch": 0.0025314109677900556,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.999604467706768e-06,
+      "loss": 17.4189,
+      "step": 140
+    },
+    {
+      "epoch": 0.002712226036917917,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.999576215400107e-06,
+      "loss": 17.8842,
+      "step": 150
+    },
+    {
+      "epoch": 0.002893041106045778,
+      "grad_norm": 40.75,
+      "learning_rate": 9.999547963093447e-06,
+      "loss": 17.6147,
+      "step": 160
+    },
+    {
+      "epoch": 0.003073856175173639,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.999519710786788e-06,
+      "loss": 17.6027,
+      "step": 170
+    },
+    {
+      "epoch": 0.0032546712443015002,
+      "grad_norm": 39.375,
+      "learning_rate": 9.999491458480129e-06,
+      "loss": 17.581,
+      "step": 180
+    },
+    {
+      "epoch": 0.003435486313429361,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.99946320617347e-06,
+      "loss": 17.5783,
+      "step": 190
+    },
+    {
+      "epoch": 0.003616301382557222,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.999434953866808e-06,
+      "loss": 17.3805,
+      "step": 200
+    },
+    {
+      "epoch": 0.0037971164516850836,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.999406701560149e-06,
+      "loss": 17.3353,
+      "step": 210
+    },
+    {
+      "epoch": 0.0039779315208129445,
+      "grad_norm": 40.25,
+      "learning_rate": 9.999378449253491e-06,
+      "loss": 17.6241,
+      "step": 220
+    },
+    {
+      "epoch": 0.004158746589940806,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.999350196946832e-06,
+      "loss": 17.5579,
+      "step": 230
+    },
+    {
+      "epoch": 0.004339561659068666,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.99932194464017e-06,
+      "loss": 17.7354,
+      "step": 240
+    },
+    {
+      "epoch": 0.004520376728196528,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.999293692333511e-06,
+      "loss": 17.3126,
+      "step": 250
+    },
+    {
+      "epoch": 0.004701191797324389,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.999265440026852e-06,
+      "loss": 17.5456,
+      "step": 260
+    },
+    {
+      "epoch": 0.00488200686645225,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.999237187720193e-06,
+      "loss": 17.7592,
+      "step": 270
+    },
+    {
+      "epoch": 0.005062821935580111,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.999208935413533e-06,
+      "loss": 17.3461,
+      "step": 280
+    },
+    {
+      "epoch": 0.0052436370047079725,
+      "grad_norm": 44.0,
+      "learning_rate": 9.999180683106872e-06,
+      "loss": 17.643,
+      "step": 290
+    },
+    {
+      "epoch": 0.005424452073835834,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.999152430800213e-06,
+      "loss": 17.3788,
+      "step": 300
+    },
+    {
+      "epoch": 0.005605267142963694,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.999124178493555e-06,
+      "loss": 17.5415,
+      "step": 310
+    },
+    {
+      "epoch": 0.005786082212091556,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.999095926186894e-06,
+      "loss": 17.9753,
+      "step": 320
+    },
+    {
+      "epoch": 0.005966897281219417,
+      "grad_norm": 38.375,
+      "learning_rate": 9.999067673880235e-06,
+      "loss": 17.3605,
+      "step": 330
+    },
+    {
+      "epoch": 0.006147712350347278,
+      "grad_norm": 39.875,
+      "learning_rate": 9.999039421573575e-06,
+      "loss": 17.8647,
+      "step": 340
+    },
+    {
+      "epoch": 0.006328527419475139,
+      "grad_norm": 40.0,
+      "learning_rate": 9.999011169266916e-06,
+      "loss": 17.6431,
+      "step": 350
+    },
+    {
+      "epoch": 0.0065093424886030005,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.998982916960256e-06,
+      "loss": 17.4284,
+      "step": 360
+    },
+    {
+      "epoch": 0.006690157557730861,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.998954664653595e-06,
+      "loss": 17.6202,
+      "step": 370
+    },
+    {
+      "epoch": 0.006870972626858722,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.998926412346936e-06,
+      "loss": 17.5108,
+      "step": 380
+    },
+    {
+      "epoch": 0.007051787695986584,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.998898160040277e-06,
+      "loss": 17.2915,
+      "step": 390
+    },
+    {
+      "epoch": 0.007232602765114444,
+      "grad_norm": 41.625,
+      "learning_rate": 9.998869907733619e-06,
+      "loss": 17.2113,
+      "step": 400
+    },
+    {
+      "epoch": 0.007413417834242306,
+      "grad_norm": 39.375,
+      "learning_rate": 9.998841655426958e-06,
+      "loss": 17.551,
+      "step": 410
+    },
+    {
+      "epoch": 0.007594232903370167,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.998813403120298e-06,
+      "loss": 17.6326,
+      "step": 420
+    },
+    {
+      "epoch": 0.007775047972498028,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.998785150813639e-06,
+      "loss": 17.1406,
+      "step": 430
+    },
+    {
+      "epoch": 0.007955863041625889,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.99875689850698e-06,
+      "loss": 17.8196,
+      "step": 440
+    },
+    {
+      "epoch": 0.00813667811075375,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.99872864620032e-06,
+      "loss": 17.2403,
+      "step": 450
+    },
+    {
+      "epoch": 0.008317493179881612,
+      "grad_norm": 40.375,
+      "learning_rate": 9.99870039389366e-06,
+      "loss": 17.9381,
+      "step": 460
+    },
+    {
+      "epoch": 0.008498308249009472,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.998672141587e-06,
+      "loss": 17.7649,
+      "step": 470
+    },
+    {
+      "epoch": 0.008679123318137333,
+      "grad_norm": 38.625,
+      "learning_rate": 9.99864388928034e-06,
+      "loss": 17.6915,
+      "step": 480
+    },
+    {
+      "epoch": 0.008859938387265195,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.998615636973681e-06,
+      "loss": 17.434,
+      "step": 490
+    },
+    {
+      "epoch": 0.009040753456393056,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.998587384667022e-06,
+      "loss": 17.4497,
+      "step": 500
+    },
+    {
+      "epoch": 0.009221568525520916,
+      "grad_norm": 39.0,
+      "learning_rate": 9.998559132360362e-06,
+      "loss": 17.5049,
+      "step": 510
+    },
+    {
+      "epoch": 0.009402383594648778,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.998530880053703e-06,
+      "loss": 17.8357,
+      "step": 520
+    },
+    {
+      "epoch": 0.009583198663776639,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.998502627747044e-06,
+      "loss": 17.334,
+      "step": 530
+    },
+    {
+      "epoch": 0.0097640137329045,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.998474375440384e-06,
+      "loss": 17.6756,
+      "step": 540
+    },
+    {
+      "epoch": 0.009944828802032362,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.998446123133723e-06,
+      "loss": 18.0823,
+      "step": 550
+    },
+    {
+      "epoch": 0.010125643871160222,
+      "grad_norm": 43.5,
+      "learning_rate": 9.998417870827064e-06,
+      "loss": 17.8315,
+      "step": 560
+    },
+    {
+      "epoch": 0.010306458940288083,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.998389618520406e-06,
+      "loss": 17.7426,
+      "step": 570
+    },
+    {
+      "epoch": 0.010487274009415945,
+      "grad_norm": 39.5,
+      "learning_rate": 9.998361366213745e-06,
+      "loss": 17.3527,
+      "step": 580
+    },
+    {
+      "epoch": 0.010668089078543806,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.998333113907086e-06,
+      "loss": 17.8237,
+      "step": 590
+    },
+    {
+      "epoch": 0.010848904147671668,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.998304861600426e-06,
+      "loss": 17.406,
+      "step": 600
+    },
+    {
+      "epoch": 0.011029719216799528,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.998276609293767e-06,
+      "loss": 17.7848,
+      "step": 610
+    },
+    {
+      "epoch": 0.011210534285927389,
+      "grad_norm": 40.5,
+      "learning_rate": 9.998248356987108e-06,
+      "loss": 17.4076,
+      "step": 620
+    },
+    {
+      "epoch": 0.011391349355055251,
+      "grad_norm": 42.5,
+      "learning_rate": 9.998220104680446e-06,
+      "loss": 17.6821,
+      "step": 630
+    },
+    {
+      "epoch": 0.011572164424183112,
+      "grad_norm": 40.125,
+      "learning_rate": 9.998191852373787e-06,
+      "loss": 17.5884,
+      "step": 640
+    },
+    {
+      "epoch": 0.011752979493310972,
+      "grad_norm": 37.4375,
+      "learning_rate": 9.998163600067128e-06,
+      "loss": 17.4682,
+      "step": 650
+    },
+    {
+      "epoch": 0.011933794562438834,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.99813534776047e-06,
+      "loss": 17.4463,
+      "step": 660
+    },
+    {
+      "epoch": 0.012114609631566695,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.998107095453809e-06,
+      "loss": 17.7127,
+      "step": 670
+    },
+    {
+      "epoch": 0.012295424700694555,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.99807884314715e-06,
+      "loss": 17.7266,
+      "step": 680
+    },
+    {
+      "epoch": 0.012476239769822418,
+      "grad_norm": 39.75,
+      "learning_rate": 9.99805059084049e-06,
+      "loss": 17.7013,
+      "step": 690
+    },
+    {
+      "epoch": 0.012657054838950278,
+      "grad_norm": 37.25,
+      "learning_rate": 9.99802233853383e-06,
+      "loss": 17.6106,
+      "step": 700
+    },
+    {
+      "epoch": 0.012837869908078139,
+      "grad_norm": 41.125,
+      "learning_rate": 9.997994086227171e-06,
+      "loss": 17.6656,
+      "step": 710
+    },
+    {
+      "epoch": 0.013018684977206001,
+      "grad_norm": 43.5,
+      "learning_rate": 9.99796583392051e-06,
+      "loss": 17.5118,
+      "step": 720
+    },
+    {
+      "epoch": 0.013199500046333862,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.997937581613851e-06,
+      "loss": 17.6598,
+      "step": 730
+    },
+    {
+      "epoch": 0.013380315115461722,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.997909329307192e-06,
+      "loss": 17.5925,
+      "step": 740
+    },
+    {
+      "epoch": 0.013561130184589584,
+      "grad_norm": 38.28125,
+      "learning_rate": 9.997881077000532e-06,
+      "loss": 17.5799,
+      "step": 750
+    },
+    {
+      "epoch": 0.013741945253717445,
+      "grad_norm": 38.5,
+      "learning_rate": 9.997852824693873e-06,
+      "loss": 17.5959,
+      "step": 760
+    },
+    {
+      "epoch": 0.013922760322845305,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.997824572387213e-06,
+      "loss": 17.2816,
+      "step": 770
+    },
+    {
+      "epoch": 0.014103575391973168,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.997796320080554e-06,
+      "loss": 17.6993,
+      "step": 780
+    },
+    {
+      "epoch": 0.014284390461101028,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.997768067773895e-06,
+      "loss": 17.8703,
+      "step": 790
+    },
+    {
+      "epoch": 0.014465205530228889,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.997739815467234e-06,
+      "loss": 17.6,
+      "step": 800
+    },
+    {
+      "epoch": 0.014646020599356751,
+      "grad_norm": 41.25,
+      "learning_rate": 9.997711563160574e-06,
+      "loss": 17.4995,
+      "step": 810
+    },
+    {
+      "epoch": 0.014826835668484611,
+      "grad_norm": 37.78125,
+      "learning_rate": 9.997683310853915e-06,
+      "loss": 17.3134,
+      "step": 820
+    },
+    {
+      "epoch": 0.015007650737612472,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.997655058547256e-06,
+      "loss": 17.6378,
+      "step": 830
+    },
+    {
+      "epoch": 0.015188465806740334,
+      "grad_norm": 41.625,
+      "learning_rate": 9.997626806240596e-06,
+      "loss": 18.0105,
+      "step": 840
+    },
+    {
+      "epoch": 0.015369280875868195,
+      "grad_norm": 40.625,
+      "learning_rate": 9.997598553933937e-06,
+      "loss": 17.5278,
+      "step": 850
+    },
+    {
+      "epoch": 0.015550095944996055,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.997570301627277e-06,
+      "loss": 17.677,
+      "step": 860
+    },
+    {
+      "epoch": 0.015730911014123918,
+      "grad_norm": 40.5,
+      "learning_rate": 9.997542049320618e-06,
+      "loss": 17.4019,
+      "step": 870
+    },
+    {
+      "epoch": 0.015911726083251778,
+      "grad_norm": 40.5,
+      "learning_rate": 9.997513797013959e-06,
+      "loss": 17.5114,
+      "step": 880
+    },
+    {
+      "epoch": 0.01609254115237964,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.997485544707298e-06,
+      "loss": 17.2865,
+      "step": 890
+    },
+    {
+      "epoch": 0.0162733562215075,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.997457292400638e-06,
+      "loss": 17.6203,
+      "step": 900
+    },
+    {
+      "epoch": 0.016454171290635363,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.997429040093979e-06,
+      "loss": 17.9291,
+      "step": 910
+    },
+    {
+      "epoch": 0.016634986359763224,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.99740078778732e-06,
+      "loss": 17.7519,
+      "step": 920
+    },
+    {
+      "epoch": 0.016815801428891084,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.99737253548066e-06,
+      "loss": 17.777,
+      "step": 930
+    },
+    {
+      "epoch": 0.016996616498018945,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.997344283174e-06,
+      "loss": 17.1429,
+      "step": 940
+    },
+    {
+      "epoch": 0.017177431567146805,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.997316030867341e-06,
+      "loss": 17.6653,
+      "step": 950
+    },
+    {
+      "epoch": 0.017358246636274666,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.997287778560682e-06,
+      "loss": 17.7769,
+      "step": 960
+    },
+    {
+      "epoch": 0.01753906170540253,
+      "grad_norm": 36.65625,
+      "learning_rate": 9.997259526254023e-06,
+      "loss": 17.4157,
+      "step": 970
+    },
+    {
+      "epoch": 0.01771987677453039,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.997231273947361e-06,
+      "loss": 17.7347,
+      "step": 980
+    },
+    {
+      "epoch": 0.01790069184365825,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.997203021640702e-06,
+      "loss": 17.3825,
+      "step": 990
+    },
+    {
+      "epoch": 0.01808150691278611,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.997174769334043e-06,
+      "loss": 17.6844,
+      "step": 1000
+    },
+    {
+      "epoch": 0.018262321981913972,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.997146517027383e-06,
+      "loss": 17.6222,
+      "step": 1010
+    },
+    {
+      "epoch": 0.018443137051041832,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.997118264720724e-06,
+      "loss": 17.6019,
+      "step": 1020
+    },
+    {
+      "epoch": 0.018623952120169696,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.997090012414065e-06,
+      "loss": 17.5789,
+      "step": 1030
+    },
+    {
+      "epoch": 0.018804767189297557,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.997061760107405e-06,
+      "loss": 17.1038,
+      "step": 1040
+    },
+    {
+      "epoch": 0.018985582258425417,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.997033507800746e-06,
+      "loss": 17.6586,
+      "step": 1050
+    },
+    {
+      "epoch": 0.019166397327553278,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.997005255494085e-06,
+      "loss": 17.4448,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01934721239668114,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.996977003187425e-06,
+      "loss": 16.9077,
+      "step": 1070
+    },
+    {
+      "epoch": 0.019528027465809,
+      "grad_norm": 36.1875,
+      "learning_rate": 9.996948750880766e-06,
+      "loss": 17.3949,
+      "step": 1080
+    },
+    {
+      "epoch": 0.019708842534936863,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.996920498574107e-06,
+      "loss": 18.0129,
+      "step": 1090
+    },
+    {
+      "epoch": 0.019889657604064723,
+      "grad_norm": 39.625,
+      "learning_rate": 9.996892246267447e-06,
+      "loss": 17.3197,
+      "step": 1100
+    },
+    {
+      "epoch": 0.020070472673192584,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.996863993960788e-06,
+      "loss": 17.4881,
+      "step": 1110
+    },
+    {
+      "epoch": 0.020251287742320444,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.996835741654128e-06,
+      "loss": 17.6111,
+      "step": 1120
+    },
+    {
+      "epoch": 0.020432102811448305,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.996807489347469e-06,
+      "loss": 17.5654,
+      "step": 1130
+    },
+    {
+      "epoch": 0.020612917880576166,
+      "grad_norm": 39.625,
+      "learning_rate": 9.99677923704081e-06,
+      "loss": 17.5175,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02079373294970403,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.996750984734149e-06,
+      "loss": 18.2359,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02097454801883189,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.99672273242749e-06,
+      "loss": 17.6192,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02115536308795975,
+      "grad_norm": 43.875,
+      "learning_rate": 9.99669448012083e-06,
+      "loss": 17.8928,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02133617815708761,
+      "grad_norm": 39.25,
+      "learning_rate": 9.99666622781417e-06,
+      "loss": 17.5018,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02151699322621547,
+      "grad_norm": 40.75,
+      "learning_rate": 9.996637975507511e-06,
+      "loss": 17.8533,
+      "step": 1190
+    },
+    {
+      "epoch": 0.021697808295343336,
+      "grad_norm": 39.5,
+      "learning_rate": 9.996609723200852e-06,
+      "loss": 17.491,
+      "step": 1200
+    },
+    {
+      "epoch": 0.021878623364471196,
+      "grad_norm": 38.75,
+      "learning_rate": 9.996581470894192e-06,
+      "loss": 17.6635,
+      "step": 1210
+    },
+    {
+      "epoch": 0.022059438433599057,
+      "grad_norm": 38.0625,
+      "learning_rate": 9.996553218587533e-06,
+      "loss": 17.7756,
+      "step": 1220
+    },
+    {
+      "epoch": 0.022240253502726917,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.996524966280872e-06,
+      "loss": 17.5359,
+      "step": 1230
+    },
+    {
+      "epoch": 0.022421068571854778,
+      "grad_norm": 38.53125,
+      "learning_rate": 9.996496713974213e-06,
+      "loss": 17.4919,
+      "step": 1240
+    },
+    {
+      "epoch": 0.022601883640982638,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.996468461667553e-06,
+      "loss": 17.6023,
+      "step": 1250
+    },
+    {
+      "epoch": 0.022782698710110502,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.996440209360894e-06,
+      "loss": 17.3854,
+      "step": 1260
+    },
+    {
+      "epoch": 0.022963513779238363,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.996411957054234e-06,
+      "loss": 17.7068,
+      "step": 1270
+    },
+    {
+      "epoch": 0.023144328848366223,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.996383704747575e-06,
+      "loss": 17.4914,
+      "step": 1280
+    },
+    {
+      "epoch": 0.023325143917494084,
+      "grad_norm": 38.875,
+      "learning_rate": 9.996355452440916e-06,
+      "loss": 18.045,
+      "step": 1290
+    },
+    {
+      "epoch": 0.023505958986621944,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.996327200134256e-06,
+      "loss": 17.5793,
+      "step": 1300
+    },
+    {
+      "epoch": 0.023686774055749805,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.996298947827597e-06,
+      "loss": 17.5581,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02386758912487767,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.996270695520936e-06,
+      "loss": 17.4817,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02404840419400553,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.996242443214276e-06,
+      "loss": 17.7449,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02422921926313339,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.996214190907617e-06,
+      "loss": 17.4616,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02441003433226125,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.996185938600958e-06,
+      "loss": 18.0316,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02459084940138911,
+      "grad_norm": 44.84375,
+      "learning_rate": 9.996157686294298e-06,
+      "loss": 17.4498,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02477166447051697,
+      "grad_norm": 36.1875,
+      "learning_rate": 9.996129433987639e-06,
+      "loss": 17.7884,
+      "step": 1370
+    },
+    {
+      "epoch": 0.024952479539644835,
+      "grad_norm": 37.9375,
+      "learning_rate": 9.99610118168098e-06,
+      "loss": 17.6003,
+      "step": 1380
+    },
+    {
+      "epoch": 0.025133294608772696,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.99607292937432e-06,
+      "loss": 17.8514,
+      "step": 1390
+    },
+    {
+      "epoch": 0.025314109677900556,
+      "grad_norm": 38.875,
+      "learning_rate": 9.99604467706766e-06,
+      "loss": 17.455,
+      "step": 1400
+    },
+    {
+      "epoch": 0.025494924747028417,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.996016424761e-06,
+      "loss": 17.4442,
+      "step": 1410
+    },
+    {
+      "epoch": 0.025675739816156277,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.99598817245434e-06,
+      "loss": 17.4482,
+      "step": 1420
+    },
+    {
+      "epoch": 0.025856554885284138,
+      "grad_norm": 38.0,
+      "learning_rate": 9.995959920147681e-06,
+      "loss": 17.3167,
+      "step": 1430
+    },
+    {
+      "epoch": 0.026037369954412002,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.995931667841022e-06,
+      "loss": 17.3963,
+      "step": 1440
+    },
+    {
+      "epoch": 0.026218185023539863,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.995903415534362e-06,
+      "loss": 17.8684,
+      "step": 1450
+    },
+    {
+      "epoch": 0.026399000092667723,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.995875163227703e-06,
+      "loss": 17.4063,
+      "step": 1460
+    },
+    {
+      "epoch": 0.026579815161795584,
+      "grad_norm": 37.8125,
+      "learning_rate": 9.995846910921044e-06,
+      "loss": 17.4437,
+      "step": 1470
+    },
+    {
+      "epoch": 0.026760630230923444,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.995818658614384e-06,
+      "loss": 17.8306,
+      "step": 1480
+    },
+    {
+      "epoch": 0.026941445300051305,
+      "grad_norm": 38.875,
+      "learning_rate": 9.995790406307723e-06,
+      "loss": 17.4603,
+      "step": 1490
+    },
+    {
+      "epoch": 0.02712226036917917,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.995762154001064e-06,
+      "loss": 17.4142,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02730307543830703,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.995733901694404e-06,
+      "loss": 17.9184,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02748389050743489,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.995705649387745e-06,
+      "loss": 17.6185,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02766470557656275,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.995677397081086e-06,
+      "loss": 17.5548,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02784552064569061,
+      "grad_norm": 40.5,
+      "learning_rate": 9.995649144774426e-06,
+      "loss": 17.4174,
+      "step": 1540
+    },
+    {
+      "epoch": 0.028026335714818475,
+      "grad_norm": 41.5,
+      "learning_rate": 9.995620892467767e-06,
+      "loss": 17.5004,
+      "step": 1550
+    },
+    {
+      "epoch": 0.028207150783946335,
+      "grad_norm": 40.625,
+      "learning_rate": 9.995592640161107e-06,
+      "loss": 17.4335,
+      "step": 1560
+    },
+    {
+      "epoch": 0.028387965853074196,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.995564387854448e-06,
+      "loss": 17.5609,
+      "step": 1570
+    },
+    {
+      "epoch": 0.028568780922202056,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.995536135547787e-06,
+      "loss": 17.2265,
+      "step": 1580
+    },
+    {
+      "epoch": 0.028749595991329917,
+      "grad_norm": 39.5,
+      "learning_rate": 9.995507883241128e-06,
+      "loss": 17.6429,
+      "step": 1590
+    },
+    {
+      "epoch": 0.028930411060457777,
+      "grad_norm": 37.625,
+      "learning_rate": 9.995479630934468e-06,
+      "loss": 17.6421,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02911122612958564,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.995451378627809e-06,
+      "loss": 17.8708,
+      "step": 1610
+    },
+    {
+      "epoch": 0.029292041198713502,
+      "grad_norm": 38.375,
+      "learning_rate": 9.99542312632115e-06,
+      "loss": 17.6356,
+      "step": 1620
+    },
+    {
+      "epoch": 0.029472856267841362,
+      "grad_norm": 37.03125,
+      "learning_rate": 9.99539487401449e-06,
+      "loss": 17.4266,
+      "step": 1630
+    },
+    {
+      "epoch": 0.029653671336969223,
+      "grad_norm": 38.125,
+      "learning_rate": 9.99536662170783e-06,
+      "loss": 17.4123,
+      "step": 1640
+    },
+    {
+      "epoch": 0.029834486406097083,
+      "grad_norm": 38.3125,
+      "learning_rate": 9.995338369401171e-06,
+      "loss": 17.4872,
+      "step": 1650
+    },
+    {
+      "epoch": 0.030015301475224944,
+      "grad_norm": 40.0,
+      "learning_rate": 9.99531011709451e-06,
+      "loss": 17.6213,
+      "step": 1660
+    },
+    {
+      "epoch": 0.030196116544352808,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.995281864787851e-06,
+      "loss": 17.2009,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03037693161348067,
+      "grad_norm": 42.125,
+      "learning_rate": 9.995253612481191e-06,
+      "loss": 17.5288,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03055774668260853,
+      "grad_norm": 37.625,
+      "learning_rate": 9.995225360174532e-06,
+      "loss": 17.842,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03073856175173639,
+      "grad_norm": 37.90625,
+      "learning_rate": 9.995197107867873e-06,
+      "loss": 17.5651,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03091937682086425,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.995168855561213e-06,
+      "loss": 17.3393,
+      "step": 1710
+    },
+    {
+      "epoch": 0.03110019188999211,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.995140603254554e-06,
+      "loss": 17.9935,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03128100695911997,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.995112350947895e-06,
+      "loss": 17.5959,
+      "step": 1730
+    },
+    {
+      "epoch": 0.031461822028247835,
+      "grad_norm": 40.25,
+      "learning_rate": 9.995084098641235e-06,
+      "loss": 17.9083,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03164263709737569,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.995055846334574e-06,
+      "loss": 17.3785,
+      "step": 1750
+    },
+    {
+      "epoch": 0.031823452166503556,
+      "grad_norm": 40.625,
+      "learning_rate": 9.995027594027915e-06,
+      "loss": 17.5619,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03200426723563142,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.994999341721255e-06,
+      "loss": 17.3343,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03218508230475928,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.994971089414596e-06,
+      "loss": 17.5575,
+      "step": 1780
+    },
+    {
+      "epoch": 0.03236589737388714,
+      "grad_norm": 38.125,
+      "learning_rate": 9.994942837107937e-06,
+      "loss": 17.4286,
+      "step": 1790
+    },
+    {
+      "epoch": 0.032546712443015,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.994914584801277e-06,
+      "loss": 17.5324,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03272752751214286,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.994886332494618e-06,
+      "loss": 17.3849,
+      "step": 1810
+    },
+    {
+      "epoch": 0.032908342581270726,
+      "grad_norm": 41.25,
+      "learning_rate": 9.994858080187959e-06,
+      "loss": 17.6226,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03308915765039858,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.994829827881299e-06,
+      "loss": 17.5047,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03326997271952645,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.994801575574638e-06,
+      "loss": 17.7064,
+      "step": 1840
+    },
+    {
+      "epoch": 0.033450787788654304,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.994773323267979e-06,
+      "loss": 17.5485,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03363160285778217,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.99474507096132e-06,
+      "loss": 17.5458,
+      "step": 1860
+    },
+    {
+      "epoch": 0.033812417926910025,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.99471681865466e-06,
+      "loss": 17.452,
+      "step": 1870
+    },
+    {
+      "epoch": 0.03399323299603789,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.994688566348e-06,
+      "loss": 18.0397,
+      "step": 1880
+    },
+    {
+      "epoch": 0.03417404806516575,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.994660314041341e-06,
+      "loss": 17.6931,
+      "step": 1890
+    },
+    {
+      "epoch": 0.03435486313429361,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.994632061734682e-06,
+      "loss": 17.7594,
+      "step": 1900
+    },
+    {
+      "epoch": 0.034535678203421474,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.994603809428022e-06,
+      "loss": 17.4341,
+      "step": 1910
+    },
+    {
+      "epoch": 0.03471649327254933,
+      "grad_norm": 40.5,
+      "learning_rate": 9.994575557121361e-06,
+      "loss": 17.5042,
+      "step": 1920
+    },
+    {
+      "epoch": 0.034897308341677195,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.994547304814702e-06,
+      "loss": 17.4923,
+      "step": 1930
+    },
+    {
+      "epoch": 0.03507812341080506,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.994519052508043e-06,
+      "loss": 17.8844,
+      "step": 1940
+    },
+    {
+      "epoch": 0.035258938479932916,
+      "grad_norm": 38.875,
+      "learning_rate": 9.994490800201383e-06,
+      "loss": 17.2922,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03543975354906078,
+      "grad_norm": 38.40625,
+      "learning_rate": 9.994462547894724e-06,
+      "loss": 17.5523,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03562056861818864,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.994434295588063e-06,
+      "loss": 17.504,
+      "step": 1970
+    },
+    {
+      "epoch": 0.0358013836873165,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.994406043281405e-06,
+      "loss": 17.8244,
+      "step": 1980
+    },
+    {
+      "epoch": 0.035982198756444365,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.994377790974746e-06,
+      "loss": 17.5912,
+      "step": 1990
+    },
+    {
+      "epoch": 0.03616301382557222,
+      "grad_norm": 37.8125,
+      "learning_rate": 9.994349538668086e-06,
+      "loss": 17.4715,
+      "step": 2000
+    },
+    {
+      "epoch": 0.036343828894700086,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.994321286361425e-06,
+      "loss": 17.6859,
+      "step": 2010
+    },
+    {
+      "epoch": 0.036524643963827944,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.994293034054766e-06,
+      "loss": 17.9146,
+      "step": 2020
+    },
+    {
+      "epoch": 0.03670545903295581,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.994264781748106e-06,
+      "loss": 17.4209,
+      "step": 2030
+    },
+    {
+      "epoch": 0.036886274102083665,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.994236529441447e-06,
+      "loss": 17.432,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03706708917121153,
+      "grad_norm": 40.875,
+      "learning_rate": 9.994208277134788e-06,
+      "loss": 17.6924,
+      "step": 2050
+    },
+    {
+      "epoch": 0.03724790424033939,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.994180024828128e-06,
+      "loss": 17.1496,
+      "step": 2060
+    },
+    {
+      "epoch": 0.03742871930946725,
+      "grad_norm": 38.3125,
+      "learning_rate": 9.994151772521469e-06,
+      "loss": 17.0407,
+      "step": 2070
+    },
+    {
+      "epoch": 0.037609534378595114,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.99412352021481e-06,
+      "loss": 18.0761,
+      "step": 2080
+    },
+    {
+      "epoch": 0.03779034944772297,
+      "grad_norm": 37.71875,
+      "learning_rate": 9.994095267908149e-06,
+      "loss": 17.3811,
+      "step": 2090
+    },
+    {
+      "epoch": 0.037971164516850835,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.99406701560149e-06,
+      "loss": 17.4023,
+      "step": 2100
+    },
+    {
+      "epoch": 0.0381519795859787,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.99403876329483e-06,
+      "loss": 17.4827,
+      "step": 2110
+    },
+    {
+      "epoch": 0.038332794655106556,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.99401051098817e-06,
+      "loss": 17.3661,
+      "step": 2120
+    },
+    {
+      "epoch": 0.03851360972423442,
+      "grad_norm": 39.375,
+      "learning_rate": 9.993982258681511e-06,
+      "loss": 17.6976,
+      "step": 2130
+    },
+    {
+      "epoch": 0.03869442479336228,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.993954006374852e-06,
+      "loss": 17.472,
+      "step": 2140
+    },
+    {
+      "epoch": 0.03887523986249014,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.993925754068192e-06,
+      "loss": 17.3894,
+      "step": 2150
+    },
+    {
+      "epoch": 0.039056054931618,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.993897501761533e-06,
+      "loss": 17.4943,
+      "step": 2160
+    },
+    {
+      "epoch": 0.03923687000074586,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.993869249454874e-06,
+      "loss": 17.8319,
+      "step": 2170
+    },
+    {
+      "epoch": 0.039417685069873726,
+      "grad_norm": 37.40625,
+      "learning_rate": 9.993840997148212e-06,
+      "loss": 17.5854,
+      "step": 2180
+    },
+    {
+      "epoch": 0.03959850013900158,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.993812744841553e-06,
+      "loss": 17.6529,
+      "step": 2190
+    },
+    {
+      "epoch": 0.03977931520812945,
+      "grad_norm": 38.375,
+      "learning_rate": 9.993784492534894e-06,
+      "loss": 17.0541,
+      "step": 2200
+    },
+    {
+      "epoch": 0.039960130277257304,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.993756240228234e-06,
+      "loss": 17.2947,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04014094534638517,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.993727987921575e-06,
+      "loss": 17.6078,
+      "step": 2220
+    },
+    {
+      "epoch": 0.04032176041551303,
+      "grad_norm": 38.375,
+      "learning_rate": 9.993699735614914e-06,
+      "loss": 17.3897,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04050257548464089,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.993671483308256e-06,
+      "loss": 17.3831,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04068339055376875,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.993643231001597e-06,
+      "loss": 17.4514,
+      "step": 2250
+    },
+    {
+      "epoch": 0.04086420562289661,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.993614978694937e-06,
+      "loss": 17.9483,
+      "step": 2260
+    },
+    {
+      "epoch": 0.041045020692024474,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.993586726388276e-06,
+      "loss": 17.9368,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04122583576115233,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.993558474081617e-06,
+      "loss": 17.7421,
+      "step": 2280
+    },
+    {
+      "epoch": 0.041406650830280195,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.993530221774958e-06,
+      "loss": 17.6865,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04158746589940806,
+      "grad_norm": 39.625,
+      "learning_rate": 9.993501969468298e-06,
+      "loss": 17.6578,
+      "step": 2300
+    },
+    {
+      "epoch": 0.041768280968535916,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.993473717161639e-06,
+      "loss": 17.5698,
+      "step": 2310
+    },
+    {
+      "epoch": 0.04194909603766378,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.993445464854978e-06,
+      "loss": 17.1494,
+      "step": 2320
+    },
+    {
+      "epoch": 0.04212991110679164,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.99341721254832e-06,
+      "loss": 17.4623,
+      "step": 2330
+    },
+    {
+      "epoch": 0.0423107261759195,
+      "grad_norm": 40.5,
+      "learning_rate": 9.99338896024166e-06,
+      "loss": 17.5555,
+      "step": 2340
+    },
+    {
+      "epoch": 0.042491541245047365,
+      "grad_norm": 42.125,
+      "learning_rate": 9.993360707935e-06,
+      "loss": 17.5949,
+      "step": 2350
+    },
+    {
+      "epoch": 0.04267235631417522,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.99333245562834e-06,
+      "loss": 17.2163,
+      "step": 2360
+    },
+    {
+      "epoch": 0.042853171383303086,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.993304203321681e-06,
+      "loss": 17.95,
+      "step": 2370
+    },
+    {
+      "epoch": 0.04303398645243094,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.993275951015021e-06,
+      "loss": 17.4045,
+      "step": 2380
+    },
+    {
+      "epoch": 0.04321480152155881,
+      "grad_norm": 37.78125,
+      "learning_rate": 9.993247698708362e-06,
+      "loss": 17.2573,
+      "step": 2390
+    },
+    {
+      "epoch": 0.04339561659068667,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.993219446401701e-06,
+      "loss": 17.1407,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04357643165981453,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.993191194095043e-06,
+      "loss": 17.4419,
+      "step": 2410
+    },
+    {
+      "epoch": 0.04375724672894239,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.993162941788384e-06,
+      "loss": 17.9597,
+      "step": 2420
+    },
+    {
+      "epoch": 0.04393806179807025,
+      "grad_norm": 40.5,
+      "learning_rate": 9.993134689481725e-06,
+      "loss": 17.3377,
+      "step": 2430
+    },
+    {
+      "epoch": 0.04411887686719811,
+      "grad_norm": 37.03125,
+      "learning_rate": 9.993106437175064e-06,
+      "loss": 17.517,
+      "step": 2440
+    },
+    {
+      "epoch": 0.04429969193632597,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.993078184868404e-06,
+      "loss": 17.1929,
+      "step": 2450
+    },
+    {
+      "epoch": 0.044480507005453834,
+      "grad_norm": 38.75,
+      "learning_rate": 9.993049932561745e-06,
+      "loss": 17.5776,
+      "step": 2460
+    },
+    {
+      "epoch": 0.0446613220745817,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.993021680255085e-06,
+      "loss": 17.2231,
+      "step": 2470
+    },
+    {
+      "epoch": 0.044842137143709555,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.992993427948426e-06,
+      "loss": 17.4934,
+      "step": 2480
+    },
+    {
+      "epoch": 0.04502295221283742,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.992965175641765e-06,
+      "loss": 17.5628,
+      "step": 2490
+    },
+    {
+      "epoch": 0.045203767281965276,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.992936923335107e-06,
+      "loss": 17.804,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04538458235109314,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.992908671028448e-06,
+      "loss": 17.9538,
+      "step": 2510
+    },
+    {
+      "epoch": 0.045565397420221004,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.992880418721787e-06,
+      "loss": 17.6877,
+      "step": 2520
+    },
+    {
+      "epoch": 0.04574621248934886,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.992852166415127e-06,
+      "loss": 18.0231,
+      "step": 2530
+    },
+    {
+      "epoch": 0.045927027558476725,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.992823914108468e-06,
+      "loss": 17.272,
+      "step": 2540
+    },
+    {
+      "epoch": 0.04610784262760458,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.992795661801809e-06,
+      "loss": 17.1814,
+      "step": 2550
+    },
+    {
+      "epoch": 0.046288657696732446,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.99276740949515e-06,
+      "loss": 17.5537,
+      "step": 2560
+    },
+    {
+      "epoch": 0.046469472765860304,
+      "grad_norm": 39.75,
+      "learning_rate": 9.992739157188488e-06,
+      "loss": 17.979,
+      "step": 2570
+    },
+    {
+      "epoch": 0.04665028783498817,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.992710904881829e-06,
+      "loss": 17.4319,
+      "step": 2580
+    },
+    {
+      "epoch": 0.04683110290411603,
+      "grad_norm": 38.28125,
+      "learning_rate": 9.992682652575171e-06,
+      "loss": 17.3438,
+      "step": 2590
+    },
+    {
+      "epoch": 0.04701191797324389,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.992654400268512e-06,
+      "loss": 17.3572,
+      "step": 2600
+    },
+    {
+      "epoch": 0.04719273304237175,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.99262614796185e-06,
+      "loss": 17.6413,
+      "step": 2610
+    },
+    {
+      "epoch": 0.04737354811149961,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.992597895655191e-06,
+      "loss": 17.2812,
+      "step": 2620
+    },
+    {
+      "epoch": 0.047554363180627474,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.992569643348532e-06,
+      "loss": 17.3716,
+      "step": 2630
+    },
+    {
+      "epoch": 0.04773517824975534,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.992541391041873e-06,
+      "loss": 17.6471,
+      "step": 2640
+    },
+    {
+      "epoch": 0.047915993318883195,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.992513138735213e-06,
+      "loss": 17.5608,
+      "step": 2650
+    },
+    {
+      "epoch": 0.04809680838801106,
+      "grad_norm": 38.15625,
+      "learning_rate": 9.992484886428552e-06,
+      "loss": 17.6671,
+      "step": 2660
+    },
+    {
+      "epoch": 0.048277623457138916,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.992456634121893e-06,
+      "loss": 17.71,
+      "step": 2670
+    },
+    {
+      "epoch": 0.04845843852626678,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.992428381815235e-06,
+      "loss": 17.377,
+      "step": 2680
+    },
+    {
+      "epoch": 0.04863925359539464,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.992400129508574e-06,
+      "loss": 17.4225,
+      "step": 2690
+    },
+    {
+      "epoch": 0.0488200686645225,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.992371877201915e-06,
+      "loss": 17.4072,
+      "step": 2700
+    },
+    {
+      "epoch": 0.049000883733650365,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.992343624895255e-06,
+      "loss": 17.1139,
+      "step": 2710
+    },
+    {
+      "epoch": 0.04918169880277822,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.992315372588596e-06,
+      "loss": 17.3573,
+      "step": 2720
+    },
+    {
+      "epoch": 0.049362513871906086,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.992287120281937e-06,
+      "loss": 17.2752,
+      "step": 2730
+    },
+    {
+      "epoch": 0.04954332894103394,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.992258867975277e-06,
+      "loss": 17.4226,
+      "step": 2740
+    },
+    {
+      "epoch": 0.04972414401016181,
+      "grad_norm": 40.25,
+      "learning_rate": 9.992230615668616e-06,
+      "loss": 17.6635,
+      "step": 2750
+    },
+    {
+      "epoch": 0.04990495907928967,
+      "grad_norm": 42.0,
+      "learning_rate": 9.992202363361958e-06,
+      "loss": 17.8238,
+      "step": 2760
+    },
+    {
+      "epoch": 0.05008577414841753,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.992174111055299e-06,
+      "loss": 17.3426,
+      "step": 2770
+    },
+    {
+      "epoch": 0.05026658921754539,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.992145858748638e-06,
+      "loss": 17.4383,
+      "step": 2780
+    },
+    {
+      "epoch": 0.05044740428667325,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.992117606441979e-06,
+      "loss": 17.505,
+      "step": 2790
+    },
+    {
+      "epoch": 0.05062821935580111,
+      "grad_norm": 42.5,
+      "learning_rate": 9.99208935413532e-06,
+      "loss": 17.3786,
+      "step": 2800
+    },
+    {
+      "epoch": 0.05080903442492898,
+      "grad_norm": 43.625,
+      "learning_rate": 9.99206110182866e-06,
+      "loss": 17.5904,
+      "step": 2810
+    },
+    {
+      "epoch": 0.050989849494056834,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.992032849522e-06,
+      "loss": 17.7643,
+      "step": 2820
+    },
+    {
+      "epoch": 0.0511706645631847,
+      "grad_norm": 39.875,
+      "learning_rate": 9.99200459721534e-06,
+      "loss": 17.3822,
+      "step": 2830
+    },
+    {
+      "epoch": 0.051351479632312555,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.99197634490868e-06,
+      "loss": 17.8304,
+      "step": 2840
+    },
+    {
+      "epoch": 0.05153229470144042,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.991948092602022e-06,
+      "loss": 17.4109,
+      "step": 2850
+    },
+    {
+      "epoch": 0.051713109770568276,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.991919840295363e-06,
+      "loss": 17.5073,
+      "step": 2860
+    },
+    {
+      "epoch": 0.05189392483969614,
+      "grad_norm": 36.53125,
+      "learning_rate": 9.991891587988702e-06,
+      "loss": 17.6826,
+      "step": 2870
+    },
+    {
+      "epoch": 0.052074739908824004,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.991863335682042e-06,
+      "loss": 17.5374,
+      "step": 2880
+    },
+    {
+      "epoch": 0.05225555497795186,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.991835083375383e-06,
+      "loss": 17.7052,
+      "step": 2890
+    },
+    {
+      "epoch": 0.052436370047079725,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.991806831068724e-06,
+      "loss": 17.8823,
+      "step": 2900
+    },
+    {
+      "epoch": 0.05261718511620758,
+      "grad_norm": 39.125,
+      "learning_rate": 9.991778578762064e-06,
+      "loss": 17.2184,
+      "step": 2910
+    },
+    {
+      "epoch": 0.052798000185335446,
+      "grad_norm": 41.375,
+      "learning_rate": 9.991750326455403e-06,
+      "loss": 17.5451,
+      "step": 2920
+    },
+    {
+      "epoch": 0.05297881525446331,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.991722074148744e-06,
+      "loss": 17.6575,
+      "step": 2930
+    },
+    {
+      "epoch": 0.05315963032359117,
+      "grad_norm": 41.625,
+      "learning_rate": 9.991693821842086e-06,
+      "loss": 17.2601,
+      "step": 2940
+    },
+    {
+      "epoch": 0.05334044539271903,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.991665569535425e-06,
+      "loss": 17.8104,
+      "step": 2950
+    },
+    {
+      "epoch": 0.05352126046184689,
+      "grad_norm": 36.96875,
+      "learning_rate": 9.991637317228766e-06,
+      "loss": 17.6773,
+      "step": 2960
+    },
+    {
+      "epoch": 0.05370207553097475,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.991609064922106e-06,
+      "loss": 17.3477,
+      "step": 2970
+    },
+    {
+      "epoch": 0.05388289060010261,
+      "grad_norm": 41.5,
+      "learning_rate": 9.991580812615447e-06,
+      "loss": 17.2803,
+      "step": 2980
+    },
+    {
+      "epoch": 0.05406370566923047,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.991552560308788e-06,
+      "loss": 17.2301,
+      "step": 2990
+    },
+    {
+      "epoch": 0.05424452073835834,
+      "grad_norm": 37.625,
+      "learning_rate": 9.991524308002127e-06,
+      "loss": 17.5987,
+      "step": 3000
+    },
+    {
+      "epoch": 0.054425335807486194,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.991496055695467e-06,
+      "loss": 17.4069,
+      "step": 3010
+    },
+    {
+      "epoch": 0.05460615087661406,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.991467803388808e-06,
+      "loss": 17.0933,
+      "step": 3020
+    },
+    {
+      "epoch": 0.054786965945741915,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.99143955108215e-06,
+      "loss": 17.2305,
+      "step": 3030
+    },
+    {
+      "epoch": 0.05496778101486978,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.991411298775489e-06,
+      "loss": 17.7806,
+      "step": 3040
+    },
+    {
+      "epoch": 0.05514859608399764,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.99138304646883e-06,
+      "loss": 17.7382,
+      "step": 3050
+    },
+    {
+      "epoch": 0.0553294111531255,
+      "grad_norm": 42.875,
+      "learning_rate": 9.99135479416217e-06,
+      "loss": 17.4769,
+      "step": 3060
+    },
+    {
+      "epoch": 0.055510226222253364,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.991326541855511e-06,
+      "loss": 17.1672,
+      "step": 3070
+    },
+    {
+      "epoch": 0.05569104129138122,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.991298289548852e-06,
+      "loss": 17.4499,
+      "step": 3080
+    },
+    {
+      "epoch": 0.055871856360509085,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.99127003724219e-06,
+      "loss": 17.5775,
+      "step": 3090
+    },
+    {
+      "epoch": 0.05605267142963695,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.991241784935531e-06,
+      "loss": 17.6459,
+      "step": 3100
+    },
+    {
+      "epoch": 0.056233486498764806,
+      "grad_norm": 40.75,
+      "learning_rate": 9.991213532628873e-06,
+      "loss": 17.3404,
+      "step": 3110
+    },
+    {
+      "epoch": 0.05641430156789267,
+      "grad_norm": 41.5,
+      "learning_rate": 9.991185280322212e-06,
+      "loss": 17.4062,
+      "step": 3120
+    },
+    {
+      "epoch": 0.05659511663702053,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.991157028015553e-06,
+      "loss": 17.7076,
+      "step": 3130
+    },
+    {
+      "epoch": 0.05677593170614839,
+      "grad_norm": 38.625,
+      "learning_rate": 9.991128775708894e-06,
+      "loss": 17.6773,
+      "step": 3140
+    },
+    {
+      "epoch": 0.05695674677527625,
+      "grad_norm": 42.0,
+      "learning_rate": 9.991100523402234e-06,
+      "loss": 17.3732,
+      "step": 3150
+    },
+    {
+      "epoch": 0.05713756184440411,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.991072271095575e-06,
+      "loss": 17.4142,
+      "step": 3160
+    },
+    {
+      "epoch": 0.057318376913531977,
+      "grad_norm": 41.375,
+      "learning_rate": 9.991044018788915e-06,
+      "loss": 17.4713,
+      "step": 3170
+    },
+    {
+      "epoch": 0.057499191982659834,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.991015766482254e-06,
+      "loss": 17.4469,
+      "step": 3180
+    },
+    {
+      "epoch": 0.0576800070517877,
+      "grad_norm": 37.5,
+      "learning_rate": 9.990987514175595e-06,
+      "loss": 17.2832,
+      "step": 3190
+    },
+    {
+      "epoch": 0.057860822120915555,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.990959261868937e-06,
+      "loss": 17.5141,
+      "step": 3200
+    },
+    {
+      "epoch": 0.05804163719004342,
+      "grad_norm": 38.5625,
+      "learning_rate": 9.990931009562276e-06,
+      "loss": 17.3238,
+      "step": 3210
+    },
+    {
+      "epoch": 0.05822245225917128,
+      "grad_norm": 40.375,
+      "learning_rate": 9.990902757255617e-06,
+      "loss": 17.4824,
+      "step": 3220
+    },
+    {
+      "epoch": 0.05840326732829914,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.990874504948957e-06,
+      "loss": 17.458,
+      "step": 3230
+    },
+    {
+      "epoch": 0.058584082397427004,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.990846252642298e-06,
+      "loss": 17.5533,
+      "step": 3240
+    },
+    {
+      "epoch": 0.05876489746655486,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.990818000335639e-06,
+      "loss": 17.7103,
+      "step": 3250
+    },
+    {
+      "epoch": 0.058945712535682725,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.990789748028978e-06,
+      "loss": 17.4866,
+      "step": 3260
+    },
+    {
+      "epoch": 0.05912652760481058,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.990761495722318e-06,
+      "loss": 17.6451,
+      "step": 3270
+    },
+    {
+      "epoch": 0.059307342673938446,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.990733243415659e-06,
+      "loss": 18.1264,
+      "step": 3280
+    },
+    {
+      "epoch": 0.05948815774306631,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.990704991109001e-06,
+      "loss": 17.2858,
+      "step": 3290
+    },
+    {
+      "epoch": 0.05966897281219417,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.99067673880234e-06,
+      "loss": 17.6059,
+      "step": 3300
+    },
+    {
+      "epoch": 0.05984978788132203,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.99064848649568e-06,
+      "loss": 17.8113,
+      "step": 3310
+    },
+    {
+      "epoch": 0.06003060295044989,
+      "grad_norm": 42.0,
+      "learning_rate": 9.990620234189021e-06,
+      "loss": 17.2323,
+      "step": 3320
+    },
+    {
+      "epoch": 0.06021141801957775,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.990591981882362e-06,
+      "loss": 17.2295,
+      "step": 3330
+    },
+    {
+      "epoch": 0.060392233088705616,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.990563729575703e-06,
+      "loss": 17.2785,
+      "step": 3340
+    },
+    {
+      "epoch": 0.06057304815783347,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.990535477269042e-06,
+      "loss": 17.3906,
+      "step": 3350
+    },
+    {
+      "epoch": 0.06075386322696134,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.990507224962382e-06,
+      "loss": 17.4731,
+      "step": 3360
+    },
+    {
+      "epoch": 0.060934678296089194,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.990478972655723e-06,
+      "loss": 17.5231,
+      "step": 3370
+    },
+    {
+      "epoch": 0.06111549336521706,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.990450720349063e-06,
+      "loss": 17.6285,
+      "step": 3380
+    },
+    {
+      "epoch": 0.061296308434344915,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.990422468042404e-06,
+      "loss": 17.5402,
+      "step": 3390
+    },
+    {
+      "epoch": 0.06147712350347278,
+      "grad_norm": 41.25,
+      "learning_rate": 9.990394215735745e-06,
+      "loss": 17.6009,
+      "step": 3400
+    },
+    {
+      "epoch": 0.06165793857260064,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.990365963429085e-06,
+      "loss": 17.5682,
+      "step": 3410
+    },
+    {
+      "epoch": 0.0618387536417285,
+      "grad_norm": 41.125,
+      "learning_rate": 9.990337711122426e-06,
+      "loss": 17.8649,
+      "step": 3420
+    },
+    {
+      "epoch": 0.062019568710856364,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.990309458815765e-06,
+      "loss": 17.6951,
+      "step": 3430
+    },
+    {
+      "epoch": 0.06220038377998422,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.990281206509105e-06,
+      "loss": 17.3738,
+      "step": 3440
+    },
+    {
+      "epoch": 0.062381198849112085,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.990252954202446e-06,
+      "loss": 17.4536,
+      "step": 3450
+    },
+    {
+      "epoch": 0.06256201391823994,
+      "grad_norm": 38.5625,
+      "learning_rate": 9.990224701895788e-06,
+      "loss": 17.4441,
+      "step": 3460
+    },
+    {
+      "epoch": 0.06274282898736781,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.990196449589127e-06,
+      "loss": 17.284,
+      "step": 3470
+    },
+    {
+      "epoch": 0.06292364405649567,
+      "grad_norm": 41.25,
+      "learning_rate": 9.990168197282468e-06,
+      "loss": 17.529,
+      "step": 3480
+    },
+    {
+      "epoch": 0.06310445912562353,
+      "grad_norm": 39.25,
+      "learning_rate": 9.990139944975809e-06,
+      "loss": 17.4169,
+      "step": 3490
+    },
+    {
+      "epoch": 0.06328527419475138,
+      "grad_norm": 37.78125,
+      "learning_rate": 9.99011169266915e-06,
+      "loss": 17.2194,
+      "step": 3500
+    },
+    {
+      "epoch": 0.06346608926387926,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.99008344036249e-06,
+      "loss": 17.5379,
+      "step": 3510
+    },
+    {
+      "epoch": 0.06364690433300711,
+      "grad_norm": 40.5,
+      "learning_rate": 9.990055188055829e-06,
+      "loss": 17.5937,
+      "step": 3520
+    },
+    {
+      "epoch": 0.06382771940213497,
+      "grad_norm": 41.125,
+      "learning_rate": 9.99002693574917e-06,
+      "loss": 17.8142,
+      "step": 3530
+    },
+    {
+      "epoch": 0.06400853447126284,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.98999868344251e-06,
+      "loss": 17.6586,
+      "step": 3540
+    },
+    {
+      "epoch": 0.0641893495403907,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.98997043113585e-06,
+      "loss": 17.0579,
+      "step": 3550
+    },
+    {
+      "epoch": 0.06437016460951855,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.989942178829191e-06,
+      "loss": 17.0986,
+      "step": 3560
+    },
+    {
+      "epoch": 0.06455097967864643,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.989913926522532e-06,
+      "loss": 17.384,
+      "step": 3570
+    },
+    {
+      "epoch": 0.06473179474777428,
+      "grad_norm": 41.625,
+      "learning_rate": 9.989885674215872e-06,
+      "loss": 17.5132,
+      "step": 3580
+    },
+    {
+      "epoch": 0.06491260981690214,
+      "grad_norm": 40.5,
+      "learning_rate": 9.989857421909213e-06,
+      "loss": 17.5032,
+      "step": 3590
+    },
+    {
+      "epoch": 0.06509342488603,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.989829169602554e-06,
+      "loss": 17.913,
+      "step": 3600
+    },
+    {
+      "epoch": 0.06527423995515787,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.989800917295893e-06,
+      "loss": 17.4051,
+      "step": 3610
+    },
+    {
+      "epoch": 0.06545505502428572,
+      "grad_norm": 38.3125,
+      "learning_rate": 9.989772664989233e-06,
+      "loss": 17.7376,
+      "step": 3620
+    },
+    {
+      "epoch": 0.06563587009341358,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.989744412682574e-06,
+      "loss": 17.9015,
+      "step": 3630
+    },
+    {
+      "epoch": 0.06581668516254145,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.989716160375915e-06,
+      "loss": 17.7884,
+      "step": 3640
+    },
+    {
+      "epoch": 0.06599750023166931,
+      "grad_norm": 36.15625,
+      "learning_rate": 9.989687908069255e-06,
+      "loss": 17.105,
+      "step": 3650
+    },
+    {
+      "epoch": 0.06617831530079717,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.989659655762596e-06,
+      "loss": 17.6172,
+      "step": 3660
+    },
+    {
+      "epoch": 0.06635913036992502,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.989631403455936e-06,
+      "loss": 17.6919,
+      "step": 3670
+    },
+    {
+      "epoch": 0.0665399454390529,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.989603151149277e-06,
+      "loss": 17.4987,
+      "step": 3680
+    },
+    {
+      "epoch": 0.06672076050818075,
+      "grad_norm": 40.875,
+      "learning_rate": 9.989574898842616e-06,
+      "loss": 17.4662,
+      "step": 3690
+    },
+    {
+      "epoch": 0.06690157557730861,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.989546646535957e-06,
+      "loss": 17.3278,
+      "step": 3700
+    },
+    {
+      "epoch": 0.06708239064643648,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.989518394229297e-06,
+      "loss": 17.5842,
+      "step": 3710
+    },
+    {
+      "epoch": 0.06726320571556434,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.989490141922638e-06,
+      "loss": 17.4774,
+      "step": 3720
+    },
+    {
+      "epoch": 0.0674440207846922,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.989461889615978e-06,
+      "loss": 17.6076,
+      "step": 3730
+    },
+    {
+      "epoch": 0.06762483585382005,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.989433637309319e-06,
+      "loss": 17.2923,
+      "step": 3740
+    },
+    {
+      "epoch": 0.06780565092294792,
+      "grad_norm": 40.625,
+      "learning_rate": 9.98940538500266e-06,
+      "loss": 17.6126,
+      "step": 3750
+    },
+    {
+      "epoch": 0.06798646599207578,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.989377132696e-06,
+      "loss": 17.6107,
+      "step": 3760
+    },
+    {
+      "epoch": 0.06816728106120364,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.989348880389341e-06,
+      "loss": 17.3268,
+      "step": 3770
+    },
+    {
+      "epoch": 0.0683480961303315,
+      "grad_norm": 39.25,
+      "learning_rate": 9.98932062808268e-06,
+      "loss": 17.3873,
+      "step": 3780
+    },
+    {
+      "epoch": 0.06852891119945936,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.98929237577602e-06,
+      "loss": 17.6123,
+      "step": 3790
+    },
+    {
+      "epoch": 0.06870972626858722,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.989264123469361e-06,
+      "loss": 17.2932,
+      "step": 3800
+    },
+    {
+      "epoch": 0.06889054133771509,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.989235871162702e-06,
+      "loss": 17.8571,
+      "step": 3810
+    },
+    {
+      "epoch": 0.06907135640684295,
+      "grad_norm": 38.625,
+      "learning_rate": 9.989207618856042e-06,
+      "loss": 17.2305,
+      "step": 3820
+    },
+    {
+      "epoch": 0.0692521714759708,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.989179366549383e-06,
+      "loss": 17.0817,
+      "step": 3830
+    },
+    {
+      "epoch": 0.06943298654509866,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.989151114242724e-06,
+      "loss": 17.5897,
+      "step": 3840
+    },
+    {
+      "epoch": 0.06961380161422653,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.989122861936064e-06,
+      "loss": 17.8866,
+      "step": 3850
+    },
+    {
+      "epoch": 0.06979461668335439,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.989094609629403e-06,
+      "loss": 17.5472,
+      "step": 3860
+    },
+    {
+      "epoch": 0.06997543175248225,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.989066357322744e-06,
+      "loss": 17.7536,
+      "step": 3870
+    },
+    {
+      "epoch": 0.07015624682161012,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.989038105016084e-06,
+      "loss": 17.7344,
+      "step": 3880
+    },
+    {
+      "epoch": 0.07033706189073798,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.989009852709425e-06,
+      "loss": 17.0403,
+      "step": 3890
+    },
+    {
+      "epoch": 0.07051787695986583,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.988981600402766e-06,
+      "loss": 17.368,
+      "step": 3900
+    },
+    {
+      "epoch": 0.07069869202899369,
+      "grad_norm": 37.21875,
+      "learning_rate": 9.988953348096106e-06,
+      "loss": 17.6145,
+      "step": 3910
+    },
+    {
+      "epoch": 0.07087950709812156,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.988925095789447e-06,
+      "loss": 17.3629,
+      "step": 3920
+    },
+    {
+      "epoch": 0.07106032216724942,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.988896843482787e-06,
+      "loss": 17.8597,
+      "step": 3930
+    },
+    {
+      "epoch": 0.07124113723637727,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.988868591176128e-06,
+      "loss": 17.4615,
+      "step": 3940
+    },
+    {
+      "epoch": 0.07142195230550515,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.988840338869467e-06,
+      "loss": 17.6934,
+      "step": 3950
+    },
+    {
+      "epoch": 0.071602767374633,
+      "grad_norm": 43.25,
+      "learning_rate": 9.988812086562808e-06,
+      "loss": 17.637,
+      "step": 3960
+    },
+    {
+      "epoch": 0.07178358244376086,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.988783834256148e-06,
+      "loss": 17.3732,
+      "step": 3970
+    },
+    {
+      "epoch": 0.07196439751288873,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.988755581949489e-06,
+      "loss": 17.4592,
+      "step": 3980
+    },
+    {
+      "epoch": 0.07214521258201659,
+      "grad_norm": 41.125,
+      "learning_rate": 9.98872732964283e-06,
+      "loss": 17.7686,
+      "step": 3990
+    },
+    {
+      "epoch": 0.07232602765114444,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.98869907733617e-06,
+      "loss": 17.8232,
+      "step": 4000
+    },
+    {
+      "epoch": 0.0725068427202723,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.98867082502951e-06,
+      "loss": 17.597,
+      "step": 4010
+    },
+    {
+      "epoch": 0.07268765778940017,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.988642572722851e-06,
+      "loss": 17.6848,
+      "step": 4020
+    },
+    {
+      "epoch": 0.07286847285852803,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.988614320416192e-06,
+      "loss": 17.1169,
+      "step": 4030
+    },
+    {
+      "epoch": 0.07304928792765589,
+      "grad_norm": 41.0,
+      "learning_rate": 9.988586068109531e-06,
+      "loss": 17.301,
+      "step": 4040
+    },
+    {
+      "epoch": 0.07323010299678376,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.988557815802872e-06,
+      "loss": 17.5104,
+      "step": 4050
+    },
+    {
+      "epoch": 0.07341091806591162,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.988529563496212e-06,
+      "loss": 17.302,
+      "step": 4060
+    },
+    {
+      "epoch": 0.07359173313503947,
+      "grad_norm": 39.625,
+      "learning_rate": 9.988501311189553e-06,
+      "loss": 17.4633,
+      "step": 4070
+    },
+    {
+      "epoch": 0.07377254820416733,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.988473058882893e-06,
+      "loss": 17.6813,
+      "step": 4080
+    },
+    {
+      "epoch": 0.0739533632732952,
+      "grad_norm": 37.4375,
+      "learning_rate": 9.988444806576234e-06,
+      "loss": 17.2189,
+      "step": 4090
+    },
+    {
+      "epoch": 0.07413417834242306,
+      "grad_norm": 39.5,
+      "learning_rate": 9.988416554269575e-06,
+      "loss": 17.2021,
+      "step": 4100
+    },
+    {
+      "epoch": 0.07431499341155091,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.988388301962915e-06,
+      "loss": 17.6964,
+      "step": 4110
+    },
+    {
+      "epoch": 0.07449580848067879,
+      "grad_norm": 39.375,
+      "learning_rate": 9.988360049656254e-06,
+      "loss": 17.2008,
+      "step": 4120
+    },
+    {
+      "epoch": 0.07467662354980664,
+      "grad_norm": 39.0,
+      "learning_rate": 9.988331797349595e-06,
+      "loss": 17.7864,
+      "step": 4130
+    },
+    {
+      "epoch": 0.0748574386189345,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.988303545042935e-06,
+      "loss": 17.8637,
+      "step": 4140
+    },
+    {
+      "epoch": 0.07503825368806236,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.988275292736276e-06,
+      "loss": 17.4759,
+      "step": 4150
+    },
+    {
+      "epoch": 0.07521906875719023,
+      "grad_norm": 41.5,
+      "learning_rate": 9.988247040429617e-06,
+      "loss": 17.9535,
+      "step": 4160
+    },
+    {
+      "epoch": 0.07539988382631808,
+      "grad_norm": 47.34375,
+      "learning_rate": 9.988218788122957e-06,
+      "loss": 17.3817,
+      "step": 4170
+    },
+    {
+      "epoch": 0.07558069889544594,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.988190535816298e-06,
+      "loss": 17.4104,
+      "step": 4180
+    },
+    {
+      "epoch": 0.07576151396457381,
+      "grad_norm": 40.25,
+      "learning_rate": 9.988162283509639e-06,
+      "loss": 17.018,
+      "step": 4190
+    },
+    {
+      "epoch": 0.07594232903370167,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.98813403120298e-06,
+      "loss": 17.7528,
+      "step": 4200
+    },
+    {
+      "epoch": 0.07612314410282953,
+      "grad_norm": 40.125,
+      "learning_rate": 9.988105778896318e-06,
+      "loss": 17.5826,
+      "step": 4210
+    },
+    {
+      "epoch": 0.0763039591719574,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.988077526589659e-06,
+      "loss": 17.3862,
+      "step": 4220
+    },
+    {
+      "epoch": 0.07648477424108525,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.988049274283e-06,
+      "loss": 17.879,
+      "step": 4230
+    },
+    {
+      "epoch": 0.07666558931021311,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.98802102197634e-06,
+      "loss": 17.4999,
+      "step": 4240
+    },
+    {
+      "epoch": 0.07684640437934097,
+      "grad_norm": 42.0,
+      "learning_rate": 9.98799276966968e-06,
+      "loss": 17.3519,
+      "step": 4250
+    },
+    {
+      "epoch": 0.07702721944846884,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.987964517363021e-06,
+      "loss": 17.3579,
+      "step": 4260
+    },
+    {
+      "epoch": 0.0772080345175967,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.987936265056362e-06,
+      "loss": 17.225,
+      "step": 4270
+    },
+    {
+      "epoch": 0.07738884958672455,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.987908012749702e-06,
+      "loss": 17.7747,
+      "step": 4280
+    },
+    {
+      "epoch": 0.07756966465585242,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.987879760443041e-06,
+      "loss": 17.5272,
+      "step": 4290
+    },
+    {
+      "epoch": 0.07775047972498028,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.987851508136382e-06,
+      "loss": 17.151,
+      "step": 4300
+    },
+    {
+      "epoch": 0.07793129479410814,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.987823255829723e-06,
+      "loss": 17.4248,
+      "step": 4310
+    },
+    {
+      "epoch": 0.078112109863236,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.987795003523063e-06,
+      "loss": 17.5838,
+      "step": 4320
+    },
+    {
+      "epoch": 0.07829292493236387,
+      "grad_norm": 40.125,
+      "learning_rate": 9.987766751216404e-06,
+      "loss": 17.3888,
+      "step": 4330
+    },
+    {
+      "epoch": 0.07847374000149172,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.987738498909745e-06,
+      "loss": 17.7687,
+      "step": 4340
+    },
+    {
+      "epoch": 0.07865455507061958,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.987710246603085e-06,
+      "loss": 17.5069,
+      "step": 4350
+    },
+    {
+      "epoch": 0.07883537013974745,
+      "grad_norm": 38.375,
+      "learning_rate": 9.987681994296426e-06,
+      "loss": 17.6491,
+      "step": 4360
+    },
+    {
+      "epoch": 0.07901618520887531,
+      "grad_norm": 39.75,
+      "learning_rate": 9.987653741989766e-06,
+      "loss": 17.67,
+      "step": 4370
+    },
+    {
+      "epoch": 0.07919700027800317,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.987625489683105e-06,
+      "loss": 17.4607,
+      "step": 4380
+    },
+    {
+      "epoch": 0.07937781534713104,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.987597237376446e-06,
+      "loss": 17.0653,
+      "step": 4390
+    },
+    {
+      "epoch": 0.0795586304162589,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.987568985069787e-06,
+      "loss": 17.7432,
+      "step": 4400
+    },
+    {
+      "epoch": 0.07973944548538675,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.987540732763127e-06,
+      "loss": 17.5271,
+      "step": 4410
+    },
+    {
+      "epoch": 0.07992026055451461,
+      "grad_norm": 43.625,
+      "learning_rate": 9.987512480456468e-06,
+      "loss": 17.3105,
+      "step": 4420
+    },
+    {
+      "epoch": 0.08010107562364248,
+      "grad_norm": 37.25,
+      "learning_rate": 9.987484228149808e-06,
+      "loss": 17.6018,
+      "step": 4430
+    },
+    {
+      "epoch": 0.08028189069277034,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.987455975843149e-06,
+      "loss": 17.2744,
+      "step": 4440
+    },
+    {
+      "epoch": 0.08046270576189819,
+      "grad_norm": 37.03125,
+      "learning_rate": 9.98742772353649e-06,
+      "loss": 17.3233,
+      "step": 4450
+    },
+    {
+      "epoch": 0.08064352083102606,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.98739947122983e-06,
+      "loss": 17.4399,
+      "step": 4460
+    },
+    {
+      "epoch": 0.08082433590015392,
+      "grad_norm": 40.5,
+      "learning_rate": 9.98737121892317e-06,
+      "loss": 16.9782,
+      "step": 4470
+    },
+    {
+      "epoch": 0.08100515096928178,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.98734296661651e-06,
+      "loss": 16.7517,
+      "step": 4480
+    },
+    {
+      "epoch": 0.08118596603840963,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.98731471430985e-06,
+      "loss": 17.4957,
+      "step": 4490
+    },
+    {
+      "epoch": 0.0813667811075375,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.987286462003191e-06,
+      "loss": 17.537,
+      "step": 4500
+    },
+    {
+      "epoch": 0.08154759617666536,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.987258209696532e-06,
+      "loss": 17.3668,
+      "step": 4510
+    },
+    {
+      "epoch": 0.08172841124579322,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.987229957389872e-06,
+      "loss": 17.4374,
+      "step": 4520
+    },
+    {
+      "epoch": 0.08190922631492109,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.987201705083213e-06,
+      "loss": 17.6475,
+      "step": 4530
+    },
+    {
+      "epoch": 0.08209004138404895,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.987173452776554e-06,
+      "loss": 17.6052,
+      "step": 4540
+    },
+    {
+      "epoch": 0.0822708564531768,
+      "grad_norm": 36.59375,
+      "learning_rate": 9.987145200469892e-06,
+      "loss": 17.2227,
+      "step": 4550
+    },
+    {
+      "epoch": 0.08245167152230466,
+      "grad_norm": 38.625,
+      "learning_rate": 9.987116948163233e-06,
+      "loss": 17.2557,
+      "step": 4560
+    },
+    {
+      "epoch": 0.08263248659143253,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.987088695856574e-06,
+      "loss": 17.3571,
+      "step": 4570
+    },
+    {
+      "epoch": 0.08281330166056039,
+      "grad_norm": 43.625,
+      "learning_rate": 9.987060443549914e-06,
+      "loss": 17.6732,
+      "step": 4580
+    },
+    {
+      "epoch": 0.08299411672968825,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.987032191243255e-06,
+      "loss": 17.7018,
+      "step": 4590
+    },
+    {
+      "epoch": 0.08317493179881612,
+      "grad_norm": 41.875,
+      "learning_rate": 9.987003938936596e-06,
+      "loss": 17.6475,
+      "step": 4600
+    },
+    {
+      "epoch": 0.08335574686794398,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.986975686629936e-06,
+      "loss": 17.6004,
+      "step": 4610
+    },
+    {
+      "epoch": 0.08353656193707183,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.986947434323277e-06,
+      "loss": 17.108,
+      "step": 4620
+    },
+    {
+      "epoch": 0.0837173770061997,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.986919182016617e-06,
+      "loss": 17.4432,
+      "step": 4630
+    },
+    {
+      "epoch": 0.08389819207532756,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.986890929709956e-06,
+      "loss": 17.2823,
+      "step": 4640
+    },
+    {
+      "epoch": 0.08407900714445542,
+      "grad_norm": 38.40625,
+      "learning_rate": 9.986862677403297e-06,
+      "loss": 17.3734,
+      "step": 4650
+    },
+    {
+      "epoch": 0.08425982221358327,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.986834425096638e-06,
+      "loss": 17.538,
+      "step": 4660
+    },
+    {
+      "epoch": 0.08444063728271115,
+      "grad_norm": 38.46875,
+      "learning_rate": 9.986806172789978e-06,
+      "loss": 16.9801,
+      "step": 4670
+    },
+    {
+      "epoch": 0.084621452351839,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.986777920483319e-06,
+      "loss": 17.5482,
+      "step": 4680
+    },
+    {
+      "epoch": 0.08480226742096686,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.98674966817666e-06,
+      "loss": 17.2156,
+      "step": 4690
+    },
+    {
+      "epoch": 0.08498308249009473,
+      "grad_norm": 37.34375,
+      "learning_rate": 9.98672141587e-06,
+      "loss": 17.5817,
+      "step": 4700
+    },
+    {
+      "epoch": 0.08516389755922259,
+      "grad_norm": 39.0,
+      "learning_rate": 9.98669316356334e-06,
+      "loss": 17.4391,
+      "step": 4710
+    },
+    {
+      "epoch": 0.08534471262835044,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.98666491125668e-06,
+      "loss": 17.4249,
+      "step": 4720
+    },
+    {
+      "epoch": 0.0855255276974783,
+      "grad_norm": 38.25,
+      "learning_rate": 9.98663665895002e-06,
+      "loss": 17.2424,
+      "step": 4730
+    },
+    {
+      "epoch": 0.08570634276660617,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.986608406643361e-06,
+      "loss": 17.6555,
+      "step": 4740
+    },
+    {
+      "epoch": 0.08588715783573403,
+      "grad_norm": 39.625,
+      "learning_rate": 9.986580154336702e-06,
+      "loss": 17.662,
+      "step": 4750
+    },
+    {
+      "epoch": 0.08606797290486189,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.986551902030042e-06,
+      "loss": 17.5323,
+      "step": 4760
+    },
+    {
+      "epoch": 0.08624878797398976,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.986523649723381e-06,
+      "loss": 17.3838,
+      "step": 4770
+    },
+    {
+      "epoch": 0.08642960304311761,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.986495397416723e-06,
+      "loss": 17.4733,
+      "step": 4780
+    },
+    {
+      "epoch": 0.08661041811224547,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.986467145110064e-06,
+      "loss": 17.9583,
+      "step": 4790
+    },
+    {
+      "epoch": 0.08679123318137334,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.986438892803405e-06,
+      "loss": 17.6036,
+      "step": 4800
+    },
+    {
+      "epoch": 0.0869720482505012,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.986410640496744e-06,
+      "loss": 17.6934,
+      "step": 4810
+    },
+    {
+      "epoch": 0.08715286331962906,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.986382388190084e-06,
+      "loss": 17.2059,
+      "step": 4820
+    },
+    {
+      "epoch": 0.08733367838875691,
+      "grad_norm": 37.75,
+      "learning_rate": 9.986354135883425e-06,
+      "loss": 17.1567,
+      "step": 4830
+    },
+    {
+      "epoch": 0.08751449345788478,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.986325883576765e-06,
+      "loss": 17.3249,
+      "step": 4840
+    },
+    {
+      "epoch": 0.08769530852701264,
+      "grad_norm": 38.25,
+      "learning_rate": 9.986297631270106e-06,
+      "loss": 17.2945,
+      "step": 4850
+    },
+    {
+      "epoch": 0.0878761235961405,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.986269378963447e-06,
+      "loss": 17.2515,
+      "step": 4860
+    },
+    {
+      "epoch": 0.08805693866526837,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.986241126656787e-06,
+      "loss": 17.6603,
+      "step": 4870
+    },
+    {
+      "epoch": 0.08823775373439623,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.986212874350128e-06,
+      "loss": 17.6273,
+      "step": 4880
+    },
+    {
+      "epoch": 0.08841856880352408,
+      "grad_norm": 37.9375,
+      "learning_rate": 9.986184622043467e-06,
+      "loss": 17.3321,
+      "step": 4890
+    },
+    {
+      "epoch": 0.08859938387265194,
+      "grad_norm": 38.15625,
+      "learning_rate": 9.986156369736808e-06,
+      "loss": 17.5066,
+      "step": 4900
+    },
+    {
+      "epoch": 0.08878019894177981,
+      "grad_norm": 40.125,
+      "learning_rate": 9.986128117430148e-06,
+      "loss": 17.2144,
+      "step": 4910
+    },
+    {
+      "epoch": 0.08896101401090767,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.986099865123489e-06,
+      "loss": 17.3687,
+      "step": 4920
+    },
+    {
+      "epoch": 0.08914182908003553,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.98607161281683e-06,
+      "loss": 17.7729,
+      "step": 4930
+    },
+    {
+      "epoch": 0.0893226441491634,
+      "grad_norm": 38.0,
+      "learning_rate": 9.98604336051017e-06,
+      "loss": 17.2783,
+      "step": 4940
+    },
+    {
+      "epoch": 0.08950345921829125,
+      "grad_norm": 37.96875,
+      "learning_rate": 9.98601510820351e-06,
+      "loss": 17.4247,
+      "step": 4950
+    },
+    {
+      "epoch": 0.08968427428741911,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.985986855896851e-06,
+      "loss": 17.406,
+      "step": 4960
+    },
+    {
+      "epoch": 0.08986508935654697,
+      "grad_norm": 40.875,
+      "learning_rate": 9.985958603590192e-06,
+      "loss": 17.4239,
+      "step": 4970
+    },
+    {
+      "epoch": 0.09004590442567484,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.98593035128353e-06,
+      "loss": 17.6854,
+      "step": 4980
+    },
+    {
+      "epoch": 0.0902267194948027,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.985902098976871e-06,
+      "loss": 17.2287,
+      "step": 4990
+    },
+    {
+      "epoch": 0.09040753456393055,
+      "grad_norm": 39.625,
+      "learning_rate": 9.985873846670212e-06,
+      "loss": 17.9483,
+      "step": 5000
+    },
+    {
+      "epoch": 0.09040753456393055,
+      "eval_loss": 2.186025381088257,
+      "eval_runtime": 229.8838,
+      "eval_samples_per_second": 3158.374,
+      "eval_steps_per_second": 49.351,
+      "step": 5000
+    },
+    {
+      "epoch": 0.09058834963305842,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.985845594363553e-06,
+      "loss": 17.5609,
+      "step": 5010
+    },
+    {
+      "epoch": 0.09076916470218628,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.985817342056893e-06,
+      "loss": 17.8373,
+      "step": 5020
+    },
+    {
+      "epoch": 0.09094997977131414,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.985789089750232e-06,
+      "loss": 17.0478,
+      "step": 5030
+    },
+    {
+      "epoch": 0.09113079484044201,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.985760837443575e-06,
+      "loss": 17.4107,
+      "step": 5040
+    },
+    {
+      "epoch": 0.09131160990956987,
+      "grad_norm": 37.3125,
+      "learning_rate": 9.985732585136915e-06,
+      "loss": 17.773,
+      "step": 5050
+    },
+    {
+      "epoch": 0.09149242497869772,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.985704332830256e-06,
+      "loss": 16.9747,
+      "step": 5060
+    },
+    {
+      "epoch": 0.09167324004782558,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.985676080523595e-06,
+      "loss": 16.9955,
+      "step": 5070
+    },
+    {
+      "epoch": 0.09185405511695345,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.985647828216935e-06,
+      "loss": 17.2063,
+      "step": 5080
+    },
+    {
+      "epoch": 0.09203487018608131,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.985619575910276e-06,
+      "loss": 17.2967,
+      "step": 5090
+    },
+    {
+      "epoch": 0.09221568525520916,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.985591323603617e-06,
+      "loss": 17.641,
+      "step": 5100
+    },
+    {
+      "epoch": 0.09239650032433704,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.985563071296957e-06,
+      "loss": 17.6372,
+      "step": 5110
+    },
+    {
+      "epoch": 0.09257731539346489,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.985534818990296e-06,
+      "loss": 17.3905,
+      "step": 5120
+    },
+    {
+      "epoch": 0.09275813046259275,
+      "grad_norm": 38.375,
+      "learning_rate": 9.985506566683638e-06,
+      "loss": 17.634,
+      "step": 5130
+    },
+    {
+      "epoch": 0.09293894553172061,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.985478314376979e-06,
+      "loss": 17.8938,
+      "step": 5140
+    },
+    {
+      "epoch": 0.09311976060084848,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.985450062070318e-06,
+      "loss": 17.6147,
+      "step": 5150
+    },
+    {
+      "epoch": 0.09330057566997634,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.985421809763659e-06,
+      "loss": 17.5877,
+      "step": 5160
+    },
+    {
+      "epoch": 0.09348139073910419,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.985393557457e-06,
+      "loss": 17.0614,
+      "step": 5170
+    },
+    {
+      "epoch": 0.09366220580823206,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.98536530515034e-06,
+      "loss": 17.3577,
+      "step": 5180
+    },
+    {
+      "epoch": 0.09384302087735992,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.98533705284368e-06,
+      "loss": 17.6818,
+      "step": 5190
+    },
+    {
+      "epoch": 0.09402383594648778,
+      "grad_norm": 41.125,
+      "learning_rate": 9.98530880053702e-06,
+      "loss": 17.375,
+      "step": 5200
+    },
+    {
+      "epoch": 0.09420465101561565,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.985280548230362e-06,
+      "loss": 17.1533,
+      "step": 5210
+    },
+    {
+      "epoch": 0.0943854660847435,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.985252295923702e-06,
+      "loss": 17.5257,
+      "step": 5220
+    },
+    {
+      "epoch": 0.09456628115387136,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.985224043617043e-06,
+      "loss": 17.6709,
+      "step": 5230
+    },
+    {
+      "epoch": 0.09474709622299922,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.985195791310382e-06,
+      "loss": 17.9804,
+      "step": 5240
+    },
+    {
+      "epoch": 0.09492791129212709,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.985167539003723e-06,
+      "loss": 17.0051,
+      "step": 5250
+    },
+    {
+      "epoch": 0.09510872636125495,
+      "grad_norm": 39.0,
+      "learning_rate": 9.985139286697063e-06,
+      "loss": 17.7626,
+      "step": 5260
+    },
+    {
+      "epoch": 0.0952895414303828,
+      "grad_norm": 39.0,
+      "learning_rate": 9.985111034390404e-06,
+      "loss": 17.0476,
+      "step": 5270
+    },
+    {
+      "epoch": 0.09547035649951068,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.985082782083744e-06,
+      "loss": 17.5495,
+      "step": 5280
+    },
+    {
+      "epoch": 0.09565117156863853,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.985054529777083e-06,
+      "loss": 17.558,
+      "step": 5290
+    },
+    {
+      "epoch": 0.09583198663776639,
+      "grad_norm": 36.9375,
+      "learning_rate": 9.985026277470426e-06,
+      "loss": 17.4099,
+      "step": 5300
+    },
+    {
+      "epoch": 0.09601280170689425,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.984998025163766e-06,
+      "loss": 17.2957,
+      "step": 5310
+    },
+    {
+      "epoch": 0.09619361677602212,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.984969772857105e-06,
+      "loss": 17.5661,
+      "step": 5320
+    },
+    {
+      "epoch": 0.09637443184514997,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.984941520550446e-06,
+      "loss": 17.8033,
+      "step": 5330
+    },
+    {
+      "epoch": 0.09655524691427783,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.984913268243786e-06,
+      "loss": 17.4909,
+      "step": 5340
+    },
+    {
+      "epoch": 0.0967360619834057,
+      "grad_norm": 38.46875,
+      "learning_rate": 9.984885015937127e-06,
+      "loss": 17.7694,
+      "step": 5350
+    },
+    {
+      "epoch": 0.09691687705253356,
+      "grad_norm": 39.5,
+      "learning_rate": 9.984856763630468e-06,
+      "loss": 17.4628,
+      "step": 5360
+    },
+    {
+      "epoch": 0.09709769212166142,
+      "grad_norm": 40.75,
+      "learning_rate": 9.984828511323808e-06,
+      "loss": 17.667,
+      "step": 5370
+    },
+    {
+      "epoch": 0.09727850719078927,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.984800259017147e-06,
+      "loss": 17.372,
+      "step": 5380
+    },
+    {
+      "epoch": 0.09745932225991714,
+      "grad_norm": 41.0,
+      "learning_rate": 9.98477200671049e-06,
+      "loss": 17.2629,
+      "step": 5390
+    },
+    {
+      "epoch": 0.097640137329045,
+      "grad_norm": 39.5,
+      "learning_rate": 9.98474375440383e-06,
+      "loss": 17.4202,
+      "step": 5400
+    },
+    {
+      "epoch": 0.09782095239817286,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.984715502097169e-06,
+      "loss": 17.287,
+      "step": 5410
+    },
+    {
+      "epoch": 0.09800176746730073,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.98468724979051e-06,
+      "loss": 17.6449,
+      "step": 5420
+    },
+    {
+      "epoch": 0.09818258253642859,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.98465899748385e-06,
+      "loss": 16.9249,
+      "step": 5430
+    },
+    {
+      "epoch": 0.09836339760555644,
+      "grad_norm": 39.375,
+      "learning_rate": 9.984630745177191e-06,
+      "loss": 17.5399,
+      "step": 5440
+    },
+    {
+      "epoch": 0.09854421267468431,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.984602492870532e-06,
+      "loss": 17.6102,
+      "step": 5450
+    },
+    {
+      "epoch": 0.09872502774381217,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.98457424056387e-06,
+      "loss": 17.4345,
+      "step": 5460
+    },
+    {
+      "epoch": 0.09890584281294003,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.984545988257211e-06,
+      "loss": 17.5696,
+      "step": 5470
+    },
+    {
+      "epoch": 0.09908665788206789,
+      "grad_norm": 41.75,
+      "learning_rate": 9.984517735950553e-06,
+      "loss": 17.2881,
+      "step": 5480
+    },
+    {
+      "epoch": 0.09926747295119576,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.984489483643894e-06,
+      "loss": 17.1251,
+      "step": 5490
+    },
+    {
+      "epoch": 0.09944828802032361,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.984461231337233e-06,
+      "loss": 17.6069,
+      "step": 5500
+    },
+    {
+      "epoch": 0.09962910308945147,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.984432979030574e-06,
+      "loss": 17.2116,
+      "step": 5510
+    },
+    {
+      "epoch": 0.09980991815857934,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.984404726723914e-06,
+      "loss": 17.4008,
+      "step": 5520
+    },
+    {
+      "epoch": 0.0999907332277072,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.984376474417255e-06,
+      "loss": 17.622,
+      "step": 5530
+    },
+    {
+      "epoch": 0.10017154829683506,
+      "grad_norm": 41.125,
+      "learning_rate": 9.984348222110595e-06,
+      "loss": 17.6973,
+      "step": 5540
+    },
+    {
+      "epoch": 0.10035236336596291,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.984319969803934e-06,
+      "loss": 17.1871,
+      "step": 5550
+    },
+    {
+      "epoch": 0.10053317843509078,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.984291717497277e-06,
+      "loss": 17.496,
+      "step": 5560
+    },
+    {
+      "epoch": 0.10071399350421864,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.984263465190617e-06,
+      "loss": 17.375,
+      "step": 5570
+    },
+    {
+      "epoch": 0.1008948085733465,
+      "grad_norm": 41.625,
+      "learning_rate": 9.984235212883956e-06,
+      "loss": 17.9584,
+      "step": 5580
+    },
+    {
+      "epoch": 0.10107562364247437,
+      "grad_norm": 39.875,
+      "learning_rate": 9.984206960577297e-06,
+      "loss": 17.8407,
+      "step": 5590
+    },
+    {
+      "epoch": 0.10125643871160223,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.984178708270638e-06,
+      "loss": 17.1891,
+      "step": 5600
+    },
+    {
+      "epoch": 0.10143725378073008,
+      "grad_norm": 40.5,
+      "learning_rate": 9.984150455963978e-06,
+      "loss": 17.3538,
+      "step": 5610
+    },
+    {
+      "epoch": 0.10161806884985795,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.984122203657319e-06,
+      "loss": 17.3117,
+      "step": 5620
+    },
+    {
+      "epoch": 0.10179888391898581,
+      "grad_norm": 37.90625,
+      "learning_rate": 9.984093951350658e-06,
+      "loss": 17.2763,
+      "step": 5630
+    },
+    {
+      "epoch": 0.10197969898811367,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.984065699043998e-06,
+      "loss": 17.175,
+      "step": 5640
+    },
+    {
+      "epoch": 0.10216051405724152,
+      "grad_norm": 37.46875,
+      "learning_rate": 9.98403744673734e-06,
+      "loss": 17.5182,
+      "step": 5650
+    },
+    {
+      "epoch": 0.1023413291263694,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.984009194430681e-06,
+      "loss": 17.2687,
+      "step": 5660
+    },
+    {
+      "epoch": 0.10252214419549725,
+      "grad_norm": 39.875,
+      "learning_rate": 9.98398094212402e-06,
+      "loss": 17.5617,
+      "step": 5670
+    },
+    {
+      "epoch": 0.10270295926462511,
+      "grad_norm": 40.375,
+      "learning_rate": 9.98395268981736e-06,
+      "loss": 17.4767,
+      "step": 5680
+    },
+    {
+      "epoch": 0.10288377433375298,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.983924437510701e-06,
+      "loss": 17.7439,
+      "step": 5690
+    },
+    {
+      "epoch": 0.10306458940288084,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.983896185204042e-06,
+      "loss": 17.0064,
+      "step": 5700
+    },
+    {
+      "epoch": 0.1032454044720087,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.983867932897383e-06,
+      "loss": 17.699,
+      "step": 5710
+    },
+    {
+      "epoch": 0.10342621954113655,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.983839680590722e-06,
+      "loss": 17.5337,
+      "step": 5720
+    },
+    {
+      "epoch": 0.10360703461026442,
+      "grad_norm": 38.375,
+      "learning_rate": 9.983811428284062e-06,
+      "loss": 17.6822,
+      "step": 5730
+    },
+    {
+      "epoch": 0.10378784967939228,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.983783175977405e-06,
+      "loss": 17.4954,
+      "step": 5740
+    },
+    {
+      "epoch": 0.10396866474852014,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.983754923670743e-06,
+      "loss": 17.6689,
+      "step": 5750
+    },
+    {
+      "epoch": 0.10414947981764801,
+      "grad_norm": 40.75,
+      "learning_rate": 9.983726671364084e-06,
+      "loss": 17.2288,
+      "step": 5760
+    },
+    {
+      "epoch": 0.10433029488677587,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.983698419057425e-06,
+      "loss": 17.3645,
+      "step": 5770
+    },
+    {
+      "epoch": 0.10451110995590372,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.983670166750765e-06,
+      "loss": 17.2011,
+      "step": 5780
+    },
+    {
+      "epoch": 0.10469192502503159,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.983641914444106e-06,
+      "loss": 17.4304,
+      "step": 5790
+    },
+    {
+      "epoch": 0.10487274009415945,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.983613662137447e-06,
+      "loss": 17.1882,
+      "step": 5800
+    },
+    {
+      "epoch": 0.10505355516328731,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.983585409830785e-06,
+      "loss": 17.7,
+      "step": 5810
+    },
+    {
+      "epoch": 0.10523437023241516,
+      "grad_norm": 41.25,
+      "learning_rate": 9.983557157524126e-06,
+      "loss": 17.4315,
+      "step": 5820
+    },
+    {
+      "epoch": 0.10541518530154304,
+      "grad_norm": 40.75,
+      "learning_rate": 9.983528905217468e-06,
+      "loss": 17.2083,
+      "step": 5830
+    },
+    {
+      "epoch": 0.10559600037067089,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.983500652910807e-06,
+      "loss": 17.4995,
+      "step": 5840
+    },
+    {
+      "epoch": 0.10577681543979875,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.983472400604148e-06,
+      "loss": 17.4644,
+      "step": 5850
+    },
+    {
+      "epoch": 0.10595763050892662,
+      "grad_norm": 39.875,
+      "learning_rate": 9.983444148297489e-06,
+      "loss": 16.9171,
+      "step": 5860
+    },
+    {
+      "epoch": 0.10613844557805448,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.98341589599083e-06,
+      "loss": 17.2418,
+      "step": 5870
+    },
+    {
+      "epoch": 0.10631926064718233,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.98338764368417e-06,
+      "loss": 17.2949,
+      "step": 5880
+    },
+    {
+      "epoch": 0.10650007571631019,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.983359391377509e-06,
+      "loss": 17.4008,
+      "step": 5890
+    },
+    {
+      "epoch": 0.10668089078543806,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.98333113907085e-06,
+      "loss": 17.65,
+      "step": 5900
+    },
+    {
+      "epoch": 0.10686170585456592,
+      "grad_norm": 38.59375,
+      "learning_rate": 9.98330288676419e-06,
+      "loss": 16.9491,
+      "step": 5910
+    },
+    {
+      "epoch": 0.10704252092369378,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.983274634457532e-06,
+      "loss": 17.8632,
+      "step": 5920
+    },
+    {
+      "epoch": 0.10722333599282165,
+      "grad_norm": 39.875,
+      "learning_rate": 9.983246382150871e-06,
+      "loss": 17.4305,
+      "step": 5930
+    },
+    {
+      "epoch": 0.1074041510619495,
+      "grad_norm": 40.5,
+      "learning_rate": 9.983218129844212e-06,
+      "loss": 17.4213,
+      "step": 5940
+    },
+    {
+      "epoch": 0.10758496613107736,
+      "grad_norm": 39.25,
+      "learning_rate": 9.983189877537553e-06,
+      "loss": 17.216,
+      "step": 5950
+    },
+    {
+      "epoch": 0.10776578120020522,
+      "grad_norm": 38.375,
+      "learning_rate": 9.983161625230893e-06,
+      "loss": 17.5143,
+      "step": 5960
+    },
+    {
+      "epoch": 0.10794659626933309,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.983133372924234e-06,
+      "loss": 17.9775,
+      "step": 5970
+    },
+    {
+      "epoch": 0.10812741133846095,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.983105120617573e-06,
+      "loss": 17.2851,
+      "step": 5980
+    },
+    {
+      "epoch": 0.1083082264075888,
+      "grad_norm": 37.90625,
+      "learning_rate": 9.983076868310913e-06,
+      "loss": 17.2725,
+      "step": 5990
+    },
+    {
+      "epoch": 0.10848904147671667,
+      "grad_norm": 36.4375,
+      "learning_rate": 9.983048616004256e-06,
+      "loss": 17.4561,
+      "step": 6000
+    },
+    {
+      "epoch": 0.10866985654584453,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.983020363697595e-06,
+      "loss": 17.251,
+      "step": 6010
+    },
+    {
+      "epoch": 0.10885067161497239,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.982992111390935e-06,
+      "loss": 17.4456,
+      "step": 6020
+    },
+    {
+      "epoch": 0.10903148668410026,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.982963859084276e-06,
+      "loss": 17.8138,
+      "step": 6030
+    },
+    {
+      "epoch": 0.10921230175322812,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.982935606777616e-06,
+      "loss": 17.5189,
+      "step": 6040
+    },
+    {
+      "epoch": 0.10939311682235597,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.982907354470957e-06,
+      "loss": 17.5637,
+      "step": 6050
+    },
+    {
+      "epoch": 0.10957393189148383,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.982879102164296e-06,
+      "loss": 17.297,
+      "step": 6060
+    },
+    {
+      "epoch": 0.1097547469606117,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.982850849857637e-06,
+      "loss": 17.6551,
+      "step": 6070
+    },
+    {
+      "epoch": 0.10993556202973956,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.982822597550977e-06,
+      "loss": 17.2923,
+      "step": 6080
+    },
+    {
+      "epoch": 0.11011637709886742,
+      "grad_norm": 40.25,
+      "learning_rate": 9.98279434524432e-06,
+      "loss": 17.8248,
+      "step": 6090
+    },
+    {
+      "epoch": 0.11029719216799529,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.982766092937658e-06,
+      "loss": 17.666,
+      "step": 6100
+    },
+    {
+      "epoch": 0.11047800723712314,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.982737840630999e-06,
+      "loss": 17.6284,
+      "step": 6110
+    },
+    {
+      "epoch": 0.110658822306251,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.98270958832434e-06,
+      "loss": 16.9809,
+      "step": 6120
+    },
+    {
+      "epoch": 0.11083963737537886,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.98268133601768e-06,
+      "loss": 17.6463,
+      "step": 6130
+    },
+    {
+      "epoch": 0.11102045244450673,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.982653083711021e-06,
+      "loss": 17.5154,
+      "step": 6140
+    },
+    {
+      "epoch": 0.11120126751363459,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.98262483140436e-06,
+      "loss": 17.27,
+      "step": 6150
+    },
+    {
+      "epoch": 0.11138208258276244,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.9825965790977e-06,
+      "loss": 17.2718,
+      "step": 6160
+    },
+    {
+      "epoch": 0.11156289765189031,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.982568326791041e-06,
+      "loss": 17.3054,
+      "step": 6170
+    },
+    {
+      "epoch": 0.11174371272101817,
+      "grad_norm": 37.53125,
+      "learning_rate": 9.982540074484382e-06,
+      "loss": 17.0941,
+      "step": 6180
+    },
+    {
+      "epoch": 0.11192452779014603,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.982511822177722e-06,
+      "loss": 16.9333,
+      "step": 6190
+    },
+    {
+      "epoch": 0.1121053428592739,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.982483569871063e-06,
+      "loss": 17.4357,
+      "step": 6200
+    },
+    {
+      "epoch": 0.11228615792840176,
+      "grad_norm": 37.4375,
+      "learning_rate": 9.982455317564404e-06,
+      "loss": 17.1446,
+      "step": 6210
+    },
+    {
+      "epoch": 0.11246697299752961,
+      "grad_norm": 40.0,
+      "learning_rate": 9.982427065257744e-06,
+      "loss": 17.581,
+      "step": 6220
+    },
+    {
+      "epoch": 0.11264778806665747,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.982398812951085e-06,
+      "loss": 17.6953,
+      "step": 6230
+    },
+    {
+      "epoch": 0.11282860313578534,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.982370560644424e-06,
+      "loss": 17.8289,
+      "step": 6240
+    },
+    {
+      "epoch": 0.1130094182049132,
+      "grad_norm": 41.375,
+      "learning_rate": 9.982342308337764e-06,
+      "loss": 17.8248,
+      "step": 6250
+    },
+    {
+      "epoch": 0.11319023327404105,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.982314056031105e-06,
+      "loss": 17.7385,
+      "step": 6260
+    },
+    {
+      "epoch": 0.11337104834316893,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.982285803724446e-06,
+      "loss": 17.1312,
+      "step": 6270
+    },
+    {
+      "epoch": 0.11355186341229678,
+      "grad_norm": 37.65625,
+      "learning_rate": 9.982257551417786e-06,
+      "loss": 17.5599,
+      "step": 6280
+    },
+    {
+      "epoch": 0.11373267848142464,
+      "grad_norm": 37.71875,
+      "learning_rate": 9.982229299111127e-06,
+      "loss": 17.3697,
+      "step": 6290
+    },
+    {
+      "epoch": 0.1139134935505525,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.982201046804468e-06,
+      "loss": 17.8049,
+      "step": 6300
+    },
+    {
+      "epoch": 0.11409430861968037,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.982172794497808e-06,
+      "loss": 17.5022,
+      "step": 6310
+    },
+    {
+      "epoch": 0.11427512368880823,
+      "grad_norm": 42.5,
+      "learning_rate": 9.982144542191147e-06,
+      "loss": 17.1551,
+      "step": 6320
+    },
+    {
+      "epoch": 0.11445593875793608,
+      "grad_norm": 39.125,
+      "learning_rate": 9.982116289884488e-06,
+      "loss": 17.5444,
+      "step": 6330
+    },
+    {
+      "epoch": 0.11463675382706395,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.982088037577828e-06,
+      "loss": 16.926,
+      "step": 6340
+    },
+    {
+      "epoch": 0.11481756889619181,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.98205978527117e-06,
+      "loss": 17.1707,
+      "step": 6350
+    },
+    {
+      "epoch": 0.11499838396531967,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.98203153296451e-06,
+      "loss": 17.3683,
+      "step": 6360
+    },
+    {
+      "epoch": 0.11517919903444752,
+      "grad_norm": 41.25,
+      "learning_rate": 9.98200328065785e-06,
+      "loss": 17.3285,
+      "step": 6370
+    },
+    {
+      "epoch": 0.1153600141035754,
+      "grad_norm": 41.75,
+      "learning_rate": 9.98197502835119e-06,
+      "loss": 17.3364,
+      "step": 6380
+    },
+    {
+      "epoch": 0.11554082917270325,
+      "grad_norm": 40.5,
+      "learning_rate": 9.981946776044531e-06,
+      "loss": 17.5414,
+      "step": 6390
+    },
+    {
+      "epoch": 0.11572164424183111,
+      "grad_norm": 40.625,
+      "learning_rate": 9.981918523737872e-06,
+      "loss": 17.6574,
+      "step": 6400
+    },
+    {
+      "epoch": 0.11590245931095898,
+      "grad_norm": 40.75,
+      "learning_rate": 9.981890271431211e-06,
+      "loss": 17.3751,
+      "step": 6410
+    },
+    {
+      "epoch": 0.11608327438008684,
+      "grad_norm": 38.53125,
+      "learning_rate": 9.981862019124552e-06,
+      "loss": 16.8837,
+      "step": 6420
+    },
+    {
+      "epoch": 0.1162640894492147,
+      "grad_norm": 41.625,
+      "learning_rate": 9.981833766817892e-06,
+      "loss": 17.3399,
+      "step": 6430
+    },
+    {
+      "epoch": 0.11644490451834257,
+      "grad_norm": 41.375,
+      "learning_rate": 9.981805514511233e-06,
+      "loss": 17.4723,
+      "step": 6440
+    },
+    {
+      "epoch": 0.11662571958747042,
+      "grad_norm": 38.75,
+      "learning_rate": 9.981777262204573e-06,
+      "loss": 17.4737,
+      "step": 6450
+    },
+    {
+      "epoch": 0.11680653465659828,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.981749009897914e-06,
+      "loss": 17.5064,
+      "step": 6460
+    },
+    {
+      "epoch": 0.11698734972572614,
+      "grad_norm": 37.75,
+      "learning_rate": 9.981720757591255e-06,
+      "loss": 17.5465,
+      "step": 6470
+    },
+    {
+      "epoch": 0.11716816479485401,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.981692505284595e-06,
+      "loss": 17.2815,
+      "step": 6480
+    },
+    {
+      "epoch": 0.11734897986398186,
+      "grad_norm": 41.25,
+      "learning_rate": 9.981664252977934e-06,
+      "loss": 17.0824,
+      "step": 6490
+    },
+    {
+      "epoch": 0.11752979493310972,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.981636000671275e-06,
+      "loss": 17.3115,
+      "step": 6500
+    },
+    {
+      "epoch": 0.11771061000223759,
+      "grad_norm": 39.625,
+      "learning_rate": 9.981607748364616e-06,
+      "loss": 17.4792,
+      "step": 6510
+    },
+    {
+      "epoch": 0.11789142507136545,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.981579496057956e-06,
+      "loss": 17.1819,
+      "step": 6520
+    },
+    {
+      "epoch": 0.1180722401404933,
+      "grad_norm": 37.71875,
+      "learning_rate": 9.981551243751297e-06,
+      "loss": 17.3972,
+      "step": 6530
+    },
+    {
+      "epoch": 0.11825305520962116,
+      "grad_norm": 38.34375,
+      "learning_rate": 9.981522991444637e-06,
+      "loss": 17.372,
+      "step": 6540
+    },
+    {
+      "epoch": 0.11843387027874903,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.981494739137978e-06,
+      "loss": 17.452,
+      "step": 6550
+    },
+    {
+      "epoch": 0.11861468534787689,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.981466486831319e-06,
+      "loss": 17.7057,
+      "step": 6560
+    },
+    {
+      "epoch": 0.11879550041700475,
+      "grad_norm": 40.375,
+      "learning_rate": 9.98143823452466e-06,
+      "loss": 17.2223,
+      "step": 6570
+    },
+    {
+      "epoch": 0.11897631548613262,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.981409982217998e-06,
+      "loss": 17.7114,
+      "step": 6580
+    },
+    {
+      "epoch": 0.11915713055526048,
+      "grad_norm": 38.75,
+      "learning_rate": 9.981381729911339e-06,
+      "loss": 17.115,
+      "step": 6590
+    },
+    {
+      "epoch": 0.11933794562438833,
+      "grad_norm": 38.3125,
+      "learning_rate": 9.98135347760468e-06,
+      "loss": 17.3579,
+      "step": 6600
+    },
+    {
+      "epoch": 0.1195187606935162,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.98132522529802e-06,
+      "loss": 17.7693,
+      "step": 6610
+    },
+    {
+      "epoch": 0.11969957576264406,
+      "grad_norm": 39.25,
+      "learning_rate": 9.98129697299136e-06,
+      "loss": 17.7085,
+      "step": 6620
+    },
+    {
+      "epoch": 0.11988039083177192,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.981268720684701e-06,
+      "loss": 17.2631,
+      "step": 6630
+    },
+    {
+      "epoch": 0.12006120590089978,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.981240468378042e-06,
+      "loss": 17.6135,
+      "step": 6640
+    },
+    {
+      "epoch": 0.12024202097002765,
+      "grad_norm": 38.34375,
+      "learning_rate": 9.981212216071383e-06,
+      "loss": 17.5299,
+      "step": 6650
+    },
+    {
+      "epoch": 0.1204228360391555,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.981183963764723e-06,
+      "loss": 17.2463,
+      "step": 6660
+    },
+    {
+      "epoch": 0.12060365110828336,
+      "grad_norm": 39.75,
+      "learning_rate": 9.981155711458062e-06,
+      "loss": 17.5251,
+      "step": 6670
+    },
+    {
+      "epoch": 0.12078446617741123,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.981127459151403e-06,
+      "loss": 17.2288,
+      "step": 6680
+    },
+    {
+      "epoch": 0.12096528124653909,
+      "grad_norm": 43.625,
+      "learning_rate": 9.981099206844743e-06,
+      "loss": 17.6455,
+      "step": 6690
+    },
+    {
+      "epoch": 0.12114609631566695,
+      "grad_norm": 38.875,
+      "learning_rate": 9.981070954538084e-06,
+      "loss": 17.074,
+      "step": 6700
+    },
+    {
+      "epoch": 0.1213269113847948,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.981042702231425e-06,
+      "loss": 17.5405,
+      "step": 6710
+    },
+    {
+      "epoch": 0.12150772645392267,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.981014449924765e-06,
+      "loss": 17.4071,
+      "step": 6720
+    },
+    {
+      "epoch": 0.12168854152305053,
+      "grad_norm": 39.75,
+      "learning_rate": 9.980986197618106e-06,
+      "loss": 17.665,
+      "step": 6730
+    },
+    {
+      "epoch": 0.12186935659217839,
+      "grad_norm": 39.625,
+      "learning_rate": 9.980957945311446e-06,
+      "loss": 17.5599,
+      "step": 6740
+    },
+    {
+      "epoch": 0.12205017166130626,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.980929693004785e-06,
+      "loss": 17.3555,
+      "step": 6750
+    },
+    {
+      "epoch": 0.12223098673043412,
+      "grad_norm": 38.625,
+      "learning_rate": 9.980901440698126e-06,
+      "loss": 17.6027,
+      "step": 6760
+    },
+    {
+      "epoch": 0.12241180179956197,
+      "grad_norm": 37.125,
+      "learning_rate": 9.980873188391467e-06,
+      "loss": 17.5312,
+      "step": 6770
+    },
+    {
+      "epoch": 0.12259261686868983,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.980844936084807e-06,
+      "loss": 17.3242,
+      "step": 6780
+    },
+    {
+      "epoch": 0.1227734319378177,
+      "grad_norm": 39.375,
+      "learning_rate": 9.980816683778148e-06,
+      "loss": 17.6528,
+      "step": 6790
+    },
+    {
+      "epoch": 0.12295424700694556,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.980788431471488e-06,
+      "loss": 17.2223,
+      "step": 6800
+    },
+    {
+      "epoch": 0.12313506207607341,
+      "grad_norm": 40.0,
+      "learning_rate": 9.980760179164829e-06,
+      "loss": 17.4004,
+      "step": 6810
+    },
+    {
+      "epoch": 0.12331587714520129,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.98073192685817e-06,
+      "loss": 17.1266,
+      "step": 6820
+    },
+    {
+      "epoch": 0.12349669221432914,
+      "grad_norm": 40.125,
+      "learning_rate": 9.98070367455151e-06,
+      "loss": 17.7438,
+      "step": 6830
+    },
+    {
+      "epoch": 0.123677507283457,
+      "grad_norm": 37.96875,
+      "learning_rate": 9.98067542224485e-06,
+      "loss": 17.5249,
+      "step": 6840
+    },
+    {
+      "epoch": 0.12385832235258487,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.98064716993819e-06,
+      "loss": 17.5363,
+      "step": 6850
+    },
+    {
+      "epoch": 0.12403913742171273,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.98061891763153e-06,
+      "loss": 17.8214,
+      "step": 6860
+    },
+    {
+      "epoch": 0.12421995249084058,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.980590665324871e-06,
+      "loss": 17.4741,
+      "step": 6870
+    },
+    {
+      "epoch": 0.12440076755996844,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.980562413018212e-06,
+      "loss": 17.3098,
+      "step": 6880
+    },
+    {
+      "epoch": 0.12458158262909631,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.980534160711552e-06,
+      "loss": 17.4718,
+      "step": 6890
+    },
+    {
+      "epoch": 0.12476239769822417,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.980505908404893e-06,
+      "loss": 17.3257,
+      "step": 6900
+    },
+    {
+      "epoch": 0.12494321276735203,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.980477656098234e-06,
+      "loss": 17.6889,
+      "step": 6910
+    },
+    {
+      "epoch": 0.12512402783647988,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.980449403791573e-06,
+      "loss": 17.4467,
+      "step": 6920
+    },
+    {
+      "epoch": 0.12530484290560776,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.980421151484913e-06,
+      "loss": 17.0792,
+      "step": 6930
+    },
+    {
+      "epoch": 0.12548565797473563,
+      "grad_norm": 40.75,
+      "learning_rate": 9.980392899178254e-06,
+      "loss": 17.2743,
+      "step": 6940
+    },
+    {
+      "epoch": 0.12566647304386347,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.980364646871594e-06,
+      "loss": 17.3742,
+      "step": 6950
+    },
+    {
+      "epoch": 0.12584728811299134,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.980336394564935e-06,
+      "loss": 17.4063,
+      "step": 6960
+    },
+    {
+      "epoch": 0.1260281031821192,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.980308142258276e-06,
+      "loss": 17.4938,
+      "step": 6970
+    },
+    {
+      "epoch": 0.12620891825124705,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.980279889951616e-06,
+      "loss": 17.3211,
+      "step": 6980
+    },
+    {
+      "epoch": 0.12638973332037493,
+      "grad_norm": 39.125,
+      "learning_rate": 9.980251637644957e-06,
+      "loss": 17.2774,
+      "step": 6990
+    },
+    {
+      "epoch": 0.12657054838950277,
+      "grad_norm": 43.125,
+      "learning_rate": 9.980223385338298e-06,
+      "loss": 17.5812,
+      "step": 7000
+    },
+    {
+      "epoch": 0.12675136345863064,
+      "grad_norm": 40.75,
+      "learning_rate": 9.980195133031636e-06,
+      "loss": 17.6814,
+      "step": 7010
+    },
+    {
+      "epoch": 0.1269321785277585,
+      "grad_norm": 36.75,
+      "learning_rate": 9.980166880724977e-06,
+      "loss": 17.2395,
+      "step": 7020
+    },
+    {
+      "epoch": 0.12711299359688635,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.980138628418318e-06,
+      "loss": 17.2053,
+      "step": 7030
+    },
+    {
+      "epoch": 0.12729380866601422,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.980110376111658e-06,
+      "loss": 17.4385,
+      "step": 7040
+    },
+    {
+      "epoch": 0.1274746237351421,
+      "grad_norm": 42.375,
+      "learning_rate": 9.980082123804999e-06,
+      "loss": 17.9981,
+      "step": 7050
+    },
+    {
+      "epoch": 0.12765543880426994,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.98005387149834e-06,
+      "loss": 17.6429,
+      "step": 7060
+    },
+    {
+      "epoch": 0.1278362538733978,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.98002561919168e-06,
+      "loss": 17.5678,
+      "step": 7070
+    },
+    {
+      "epoch": 0.12801706894252568,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.97999736688502e-06,
+      "loss": 17.5928,
+      "step": 7080
+    },
+    {
+      "epoch": 0.12819788401165352,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.979969114578361e-06,
+      "loss": 17.4203,
+      "step": 7090
+    },
+    {
+      "epoch": 0.1283786990807814,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.9799408622717e-06,
+      "loss": 17.4189,
+      "step": 7100
+    },
+    {
+      "epoch": 0.12855951414990927,
+      "grad_norm": 40.625,
+      "learning_rate": 9.979912609965041e-06,
+      "loss": 17.4796,
+      "step": 7110
+    },
+    {
+      "epoch": 0.1287403292190371,
+      "grad_norm": 38.875,
+      "learning_rate": 9.979884357658382e-06,
+      "loss": 17.6867,
+      "step": 7120
+    },
+    {
+      "epoch": 0.12892114428816498,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.979856105351722e-06,
+      "loss": 17.7465,
+      "step": 7130
+    },
+    {
+      "epoch": 0.12910195935729285,
+      "grad_norm": 39.125,
+      "learning_rate": 9.979827853045063e-06,
+      "loss": 17.5646,
+      "step": 7140
+    },
+    {
+      "epoch": 0.1292827744264207,
+      "grad_norm": 38.46875,
+      "learning_rate": 9.979799600738403e-06,
+      "loss": 17.3168,
+      "step": 7150
+    },
+    {
+      "epoch": 0.12946358949554856,
+      "grad_norm": 41.0,
+      "learning_rate": 9.979771348431744e-06,
+      "loss": 17.292,
+      "step": 7160
+    },
+    {
+      "epoch": 0.1296444045646764,
+      "grad_norm": 42.75,
+      "learning_rate": 9.979743096125085e-06,
+      "loss": 17.3749,
+      "step": 7170
+    },
+    {
+      "epoch": 0.12982521963380428,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.979714843818424e-06,
+      "loss": 17.1644,
+      "step": 7180
+    },
+    {
+      "epoch": 0.13000603470293215,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.979686591511764e-06,
+      "loss": 17.7358,
+      "step": 7190
+    },
+    {
+      "epoch": 0.13018684977206,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.979658339205105e-06,
+      "loss": 17.423,
+      "step": 7200
+    },
+    {
+      "epoch": 0.13036766484118786,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.979630086898446e-06,
+      "loss": 17.3277,
+      "step": 7210
+    },
+    {
+      "epoch": 0.13054847991031573,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.979601834591786e-06,
+      "loss": 17.3163,
+      "step": 7220
+    },
+    {
+      "epoch": 0.13072929497944358,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.979573582285127e-06,
+      "loss": 17.2494,
+      "step": 7230
+    },
+    {
+      "epoch": 0.13091011004857145,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.979545329978467e-06,
+      "loss": 17.5062,
+      "step": 7240
+    },
+    {
+      "epoch": 0.13109092511769932,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.979517077671808e-06,
+      "loss": 17.5579,
+      "step": 7250
+    },
+    {
+      "epoch": 0.13127174018682716,
+      "grad_norm": 40.25,
+      "learning_rate": 9.979488825365149e-06,
+      "loss": 17.6391,
+      "step": 7260
+    },
+    {
+      "epoch": 0.13145255525595503,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.979460573058488e-06,
+      "loss": 17.387,
+      "step": 7270
+    },
+    {
+      "epoch": 0.1316333703250829,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.979432320751828e-06,
+      "loss": 17.5774,
+      "step": 7280
+    },
+    {
+      "epoch": 0.13181418539421075,
+      "grad_norm": 41.125,
+      "learning_rate": 9.979404068445169e-06,
+      "loss": 17.1301,
+      "step": 7290
+    },
+    {
+      "epoch": 0.13199500046333862,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.97937581613851e-06,
+      "loss": 17.3479,
+      "step": 7300
+    },
+    {
+      "epoch": 0.1321758155324665,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.97934756383185e-06,
+      "loss": 17.3513,
+      "step": 7310
+    },
+    {
+      "epoch": 0.13235663060159433,
+      "grad_norm": 41.75,
+      "learning_rate": 9.97931931152519e-06,
+      "loss": 16.835,
+      "step": 7320
+    },
+    {
+      "epoch": 0.1325374456707222,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.979291059218531e-06,
+      "loss": 17.5654,
+      "step": 7330
+    },
+    {
+      "epoch": 0.13271826073985005,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.979262806911872e-06,
+      "loss": 17.58,
+      "step": 7340
+    },
+    {
+      "epoch": 0.13289907580897792,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.97923455460521e-06,
+      "loss": 17.6938,
+      "step": 7350
+    },
+    {
+      "epoch": 0.1330798908781058,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.979206302298551e-06,
+      "loss": 16.8821,
+      "step": 7360
+    },
+    {
+      "epoch": 0.13326070594723363,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.979178049991892e-06,
+      "loss": 17.4715,
+      "step": 7370
+    },
+    {
+      "epoch": 0.1334415210163615,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.979149797685233e-06,
+      "loss": 17.1391,
+      "step": 7380
+    },
+    {
+      "epoch": 0.13362233608548937,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.979121545378573e-06,
+      "loss": 17.4993,
+      "step": 7390
+    },
+    {
+      "epoch": 0.13380315115461722,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.979093293071914e-06,
+      "loss": 17.5661,
+      "step": 7400
+    },
+    {
+      "epoch": 0.1339839662237451,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.979065040765255e-06,
+      "loss": 17.7374,
+      "step": 7410
+    },
+    {
+      "epoch": 0.13416478129287296,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.979036788458595e-06,
+      "loss": 17.6661,
+      "step": 7420
+    },
+    {
+      "epoch": 0.1343455963620008,
+      "grad_norm": 38.03125,
+      "learning_rate": 9.979008536151936e-06,
+      "loss": 16.8962,
+      "step": 7430
+    },
+    {
+      "epoch": 0.13452641143112867,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.978980283845275e-06,
+      "loss": 17.0925,
+      "step": 7440
+    },
+    {
+      "epoch": 0.13470722650025654,
+      "grad_norm": 38.03125,
+      "learning_rate": 9.978952031538615e-06,
+      "loss": 17.3493,
+      "step": 7450
+    },
+    {
+      "epoch": 0.1348880415693844,
+      "grad_norm": 40.75,
+      "learning_rate": 9.978923779231956e-06,
+      "loss": 17.4892,
+      "step": 7460
+    },
+    {
+      "epoch": 0.13506885663851226,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.978895526925297e-06,
+      "loss": 17.3092,
+      "step": 7470
+    },
+    {
+      "epoch": 0.1352496717076401,
+      "grad_norm": 41.5,
+      "learning_rate": 9.978867274618637e-06,
+      "loss": 17.4271,
+      "step": 7480
+    },
+    {
+      "epoch": 0.13543048677676797,
+      "grad_norm": 38.15625,
+      "learning_rate": 9.978839022311978e-06,
+      "loss": 17.9422,
+      "step": 7490
+    },
+    {
+      "epoch": 0.13561130184589584,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.978810770005318e-06,
+      "loss": 17.0924,
+      "step": 7500
+    },
+    {
+      "epoch": 0.1357921169150237,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.978782517698659e-06,
+      "loss": 17.3601,
+      "step": 7510
+    },
+    {
+      "epoch": 0.13597293198415156,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.978754265391998e-06,
+      "loss": 17.6276,
+      "step": 7520
+    },
+    {
+      "epoch": 0.13615374705327943,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.978726013085339e-06,
+      "loss": 17.1647,
+      "step": 7530
+    },
+    {
+      "epoch": 0.13633456212240727,
+      "grad_norm": 40.625,
+      "learning_rate": 9.97869776077868e-06,
+      "loss": 17.5063,
+      "step": 7540
+    },
+    {
+      "epoch": 0.13651537719153514,
+      "grad_norm": 40.0,
+      "learning_rate": 9.97866950847202e-06,
+      "loss": 17.4359,
+      "step": 7550
+    },
+    {
+      "epoch": 0.136696192260663,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.97864125616536e-06,
+      "loss": 17.7257,
+      "step": 7560
+    },
+    {
+      "epoch": 0.13687700732979086,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.978613003858701e-06,
+      "loss": 17.6415,
+      "step": 7570
+    },
+    {
+      "epoch": 0.13705782239891873,
+      "grad_norm": 41.25,
+      "learning_rate": 9.978584751552042e-06,
+      "loss": 17.0992,
+      "step": 7580
+    },
+    {
+      "epoch": 0.1372386374680466,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.978556499245382e-06,
+      "loss": 17.3908,
+      "step": 7590
+    },
+    {
+      "epoch": 0.13741945253717444,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.978528246938723e-06,
+      "loss": 17.303,
+      "step": 7600
+    },
+    {
+      "epoch": 0.1376002676063023,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.978499994632062e-06,
+      "loss": 17.3098,
+      "step": 7610
+    },
+    {
+      "epoch": 0.13778108267543018,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.978471742325403e-06,
+      "loss": 17.2351,
+      "step": 7620
+    },
+    {
+      "epoch": 0.13796189774455803,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.978443490018743e-06,
+      "loss": 17.6227,
+      "step": 7630
+    },
+    {
+      "epoch": 0.1381427128136859,
+      "grad_norm": 41.125,
+      "learning_rate": 9.978415237712084e-06,
+      "loss": 17.4154,
+      "step": 7640
+    },
+    {
+      "epoch": 0.13832352788281374,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.978386985405424e-06,
+      "loss": 17.1675,
+      "step": 7650
+    },
+    {
+      "epoch": 0.1385043429519416,
+      "grad_norm": 39.875,
+      "learning_rate": 9.978358733098763e-06,
+      "loss": 17.3523,
+      "step": 7660
+    },
+    {
+      "epoch": 0.13868515802106948,
+      "grad_norm": 40.625,
+      "learning_rate": 9.978330480792106e-06,
+      "loss": 17.9028,
+      "step": 7670
+    },
+    {
+      "epoch": 0.13886597309019733,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.978302228485446e-06,
+      "loss": 17.5639,
+      "step": 7680
+    },
+    {
+      "epoch": 0.1390467881593252,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.978273976178787e-06,
+      "loss": 17.2738,
+      "step": 7690
+    },
+    {
+      "epoch": 0.13922760322845307,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.978245723872126e-06,
+      "loss": 17.1962,
+      "step": 7700
+    },
+    {
+      "epoch": 0.1394084182975809,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.978217471565466e-06,
+      "loss": 17.376,
+      "step": 7710
+    },
+    {
+      "epoch": 0.13958923336670878,
+      "grad_norm": 37.5,
+      "learning_rate": 9.978189219258807e-06,
+      "loss": 17.0739,
+      "step": 7720
+    },
+    {
+      "epoch": 0.13977004843583665,
+      "grad_norm": 38.5,
+      "learning_rate": 9.978160966952148e-06,
+      "loss": 16.9097,
+      "step": 7730
+    },
+    {
+      "epoch": 0.1399508635049645,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.978132714645488e-06,
+      "loss": 17.4405,
+      "step": 7740
+    },
+    {
+      "epoch": 0.14013167857409237,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.978104462338829e-06,
+      "loss": 17.3024,
+      "step": 7750
+    },
+    {
+      "epoch": 0.14031249364322024,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.97807621003217e-06,
+      "loss": 17.0304,
+      "step": 7760
+    },
+    {
+      "epoch": 0.14049330871234808,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.97804795772551e-06,
+      "loss": 17.3244,
+      "step": 7770
+    },
+    {
+      "epoch": 0.14067412378147595,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.978019705418849e-06,
+      "loss": 17.3528,
+      "step": 7780
+    },
+    {
+      "epoch": 0.14085493885060382,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.97799145311219e-06,
+      "loss": 17.2244,
+      "step": 7790
+    },
+    {
+      "epoch": 0.14103575391973167,
+      "grad_norm": 38.0625,
+      "learning_rate": 9.97796320080553e-06,
+      "loss": 16.99,
+      "step": 7800
+    },
+    {
+      "epoch": 0.14121656898885954,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.977934948498871e-06,
+      "loss": 17.0707,
+      "step": 7810
+    },
+    {
+      "epoch": 0.14139738405798738,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.977906696192212e-06,
+      "loss": 17.4142,
+      "step": 7820
+    },
+    {
+      "epoch": 0.14157819912711525,
+      "grad_norm": 41.125,
+      "learning_rate": 9.97787844388555e-06,
+      "loss": 17.4165,
+      "step": 7830
+    },
+    {
+      "epoch": 0.14175901419624312,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.977850191578893e-06,
+      "loss": 17.5218,
+      "step": 7840
+    },
+    {
+      "epoch": 0.14193982926537096,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.977821939272233e-06,
+      "loss": 17.1668,
+      "step": 7850
+    },
+    {
+      "epoch": 0.14212064433449884,
+      "grad_norm": 37.46875,
+      "learning_rate": 9.977793686965574e-06,
+      "loss": 16.9904,
+      "step": 7860
+    },
+    {
+      "epoch": 0.1423014594036267,
+      "grad_norm": 37.21875,
+      "learning_rate": 9.977765434658913e-06,
+      "loss": 17.0856,
+      "step": 7870
+    },
+    {
+      "epoch": 0.14248227447275455,
+      "grad_norm": 36.71875,
+      "learning_rate": 9.977737182352254e-06,
+      "loss": 17.1422,
+      "step": 7880
+    },
+    {
+      "epoch": 0.14266308954188242,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.977708930045594e-06,
+      "loss": 17.0785,
+      "step": 7890
+    },
+    {
+      "epoch": 0.1428439046110103,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.977680677738935e-06,
+      "loss": 17.4634,
+      "step": 7900
+    },
+    {
+      "epoch": 0.14302471968013813,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.977652425432276e-06,
+      "loss": 17.3942,
+      "step": 7910
+    },
+    {
+      "epoch": 0.143205534749266,
+      "grad_norm": 43.75,
+      "learning_rate": 9.977624173125614e-06,
+      "loss": 17.7189,
+      "step": 7920
+    },
+    {
+      "epoch": 0.14338634981839388,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.977595920818957e-06,
+      "loss": 17.412,
+      "step": 7930
+    },
+    {
+      "epoch": 0.14356716488752172,
+      "grad_norm": 42.875,
+      "learning_rate": 9.977567668512297e-06,
+      "loss": 17.591,
+      "step": 7940
+    },
+    {
+      "epoch": 0.1437479799566496,
+      "grad_norm": 39.125,
+      "learning_rate": 9.977539416205636e-06,
+      "loss": 17.5684,
+      "step": 7950
+    },
+    {
+      "epoch": 0.14392879502577746,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.977511163898977e-06,
+      "loss": 17.3622,
+      "step": 7960
+    },
+    {
+      "epoch": 0.1441096100949053,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.977482911592318e-06,
+      "loss": 17.1179,
+      "step": 7970
+    },
+    {
+      "epoch": 0.14429042516403318,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.977454659285658e-06,
+      "loss": 17.5406,
+      "step": 7980
+    },
+    {
+      "epoch": 0.14447124023316102,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.977426406978999e-06,
+      "loss": 17.0921,
+      "step": 7990
+    },
+    {
+      "epoch": 0.1446520553022889,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.97739815467234e-06,
+      "loss": 17.3135,
+      "step": 8000
+    },
+    {
+      "epoch": 0.14483287037141676,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.977369902365678e-06,
+      "loss": 17.4295,
+      "step": 8010
+    },
+    {
+      "epoch": 0.1450136854405446,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.97734165005902e-06,
+      "loss": 17.4965,
+      "step": 8020
+    },
+    {
+      "epoch": 0.14519450050967248,
+      "grad_norm": 39.5,
+      "learning_rate": 9.977313397752361e-06,
+      "loss": 17.6603,
+      "step": 8030
+    },
+    {
+      "epoch": 0.14537531557880035,
+      "grad_norm": 40.25,
+      "learning_rate": 9.9772851454457e-06,
+      "loss": 16.9875,
+      "step": 8040
+    },
+    {
+      "epoch": 0.1455561306479282,
+      "grad_norm": 42.625,
+      "learning_rate": 9.977256893139041e-06,
+      "loss": 17.1946,
+      "step": 8050
+    },
+    {
+      "epoch": 0.14573694571705606,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.977228640832381e-06,
+      "loss": 17.4784,
+      "step": 8060
+    },
+    {
+      "epoch": 0.14591776078618393,
+      "grad_norm": 38.75,
+      "learning_rate": 9.977200388525722e-06,
+      "loss": 17.3109,
+      "step": 8070
+    },
+    {
+      "epoch": 0.14609857585531177,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.977172136219063e-06,
+      "loss": 17.7436,
+      "step": 8080
+    },
+    {
+      "epoch": 0.14627939092443965,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.977143883912402e-06,
+      "loss": 17.1768,
+      "step": 8090
+    },
+    {
+      "epoch": 0.14646020599356752,
+      "grad_norm": 40.125,
+      "learning_rate": 9.977115631605744e-06,
+      "loss": 17.3665,
+      "step": 8100
+    },
+    {
+      "epoch": 0.14664102106269536,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.977087379299085e-06,
+      "loss": 17.5109,
+      "step": 8110
+    },
+    {
+      "epoch": 0.14682183613182323,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.977059126992425e-06,
+      "loss": 17.9137,
+      "step": 8120
+    },
+    {
+      "epoch": 0.1470026512009511,
+      "grad_norm": 43.375,
+      "learning_rate": 9.977030874685764e-06,
+      "loss": 17.3533,
+      "step": 8130
+    },
+    {
+      "epoch": 0.14718346627007894,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.977002622379105e-06,
+      "loss": 17.4239,
+      "step": 8140
+    },
+    {
+      "epoch": 0.14736428133920682,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.976974370072445e-06,
+      "loss": 17.8833,
+      "step": 8150
+    },
+    {
+      "epoch": 0.14754509640833466,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.976946117765786e-06,
+      "loss": 17.5879,
+      "step": 8160
+    },
+    {
+      "epoch": 0.14772591147746253,
+      "grad_norm": 40.875,
+      "learning_rate": 9.976917865459127e-06,
+      "loss": 17.5351,
+      "step": 8170
+    },
+    {
+      "epoch": 0.1479067265465904,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.976889613152466e-06,
+      "loss": 17.1707,
+      "step": 8180
+    },
+    {
+      "epoch": 0.14808754161571824,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.976861360845808e-06,
+      "loss": 17.5665,
+      "step": 8190
+    },
+    {
+      "epoch": 0.14826835668484611,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.976833108539148e-06,
+      "loss": 17.9471,
+      "step": 8200
+    },
+    {
+      "epoch": 0.14844917175397399,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.976804856232487e-06,
+      "loss": 17.341,
+      "step": 8210
+    },
+    {
+      "epoch": 0.14862998682310183,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.976776603925828e-06,
+      "loss": 17.2915,
+      "step": 8220
+    },
+    {
+      "epoch": 0.1488108018922297,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.976748351619169e-06,
+      "loss": 17.2771,
+      "step": 8230
+    },
+    {
+      "epoch": 0.14899161696135757,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.97672009931251e-06,
+      "loss": 17.4774,
+      "step": 8240
+    },
+    {
+      "epoch": 0.1491724320304854,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.97669184700585e-06,
+      "loss": 17.43,
+      "step": 8250
+    },
+    {
+      "epoch": 0.14935324709961328,
+      "grad_norm": 39.875,
+      "learning_rate": 9.976663594699189e-06,
+      "loss": 17.3579,
+      "step": 8260
+    },
+    {
+      "epoch": 0.14953406216874116,
+      "grad_norm": 39.125,
+      "learning_rate": 9.97663534239253e-06,
+      "loss": 17.5017,
+      "step": 8270
+    },
+    {
+      "epoch": 0.149714877237869,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.976607090085872e-06,
+      "loss": 17.2222,
+      "step": 8280
+    },
+    {
+      "epoch": 0.14989569230699687,
+      "grad_norm": 38.375,
+      "learning_rate": 9.976578837779212e-06,
+      "loss": 17.222,
+      "step": 8290
+    },
+    {
+      "epoch": 0.1500765073761247,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.976550585472551e-06,
+      "loss": 17.5821,
+      "step": 8300
+    },
+    {
+      "epoch": 0.15025732244525258,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.976522333165892e-06,
+      "loss": 17.1667,
+      "step": 8310
+    },
+    {
+      "epoch": 0.15043813751438045,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.976494080859233e-06,
+      "loss": 17.5639,
+      "step": 8320
+    },
+    {
+      "epoch": 0.1506189525835083,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.976465828552573e-06,
+      "loss": 17.3709,
+      "step": 8330
+    },
+    {
+      "epoch": 0.15079976765263617,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.976437576245914e-06,
+      "loss": 17.223,
+      "step": 8340
+    },
+    {
+      "epoch": 0.15098058272176404,
+      "grad_norm": 36.84375,
+      "learning_rate": 9.976409323939253e-06,
+      "loss": 17.4693,
+      "step": 8350
+    },
+    {
+      "epoch": 0.15116139779089188,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.976381071632593e-06,
+      "loss": 17.4322,
+      "step": 8360
+    },
+    {
+      "epoch": 0.15134221286001975,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.976352819325936e-06,
+      "loss": 17.8045,
+      "step": 8370
+    },
+    {
+      "epoch": 0.15152302792914762,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.976324567019275e-06,
+      "loss": 17.2138,
+      "step": 8380
+    },
+    {
+      "epoch": 0.15170384299827547,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.976296314712615e-06,
+      "loss": 17.234,
+      "step": 8390
+    },
+    {
+      "epoch": 0.15188465806740334,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.976268062405956e-06,
+      "loss": 17.3728,
+      "step": 8400
+    },
+    {
+      "epoch": 0.1520654731365312,
+      "grad_norm": 38.75,
+      "learning_rate": 9.976239810099296e-06,
+      "loss": 17.6384,
+      "step": 8410
+    },
+    {
+      "epoch": 0.15224628820565905,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.976211557792637e-06,
+      "loss": 17.4172,
+      "step": 8420
+    },
+    {
+      "epoch": 0.15242710327478692,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.976183305485978e-06,
+      "loss": 17.0899,
+      "step": 8430
+    },
+    {
+      "epoch": 0.1526079183439148,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.976155053179317e-06,
+      "loss": 17.2965,
+      "step": 8440
+    },
+    {
+      "epoch": 0.15278873341304264,
+      "grad_norm": 40.75,
+      "learning_rate": 9.976126800872659e-06,
+      "loss": 17.5926,
+      "step": 8450
+    },
+    {
+      "epoch": 0.1529695484821705,
+      "grad_norm": 39.375,
+      "learning_rate": 9.976098548566e-06,
+      "loss": 17.4443,
+      "step": 8460
+    },
+    {
+      "epoch": 0.15315036355129835,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.976070296259339e-06,
+      "loss": 17.2535,
+      "step": 8470
+    },
+    {
+      "epoch": 0.15333117862042622,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.976042043952679e-06,
+      "loss": 17.4198,
+      "step": 8480
+    },
+    {
+      "epoch": 0.1535119936895541,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.97601379164602e-06,
+      "loss": 17.372,
+      "step": 8490
+    },
+    {
+      "epoch": 0.15369280875868194,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.97598553933936e-06,
+      "loss": 17.0233,
+      "step": 8500
+    },
+    {
+      "epoch": 0.1538736238278098,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.975957287032701e-06,
+      "loss": 17.4556,
+      "step": 8510
+    },
+    {
+      "epoch": 0.15405443889693768,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.97592903472604e-06,
+      "loss": 17.3115,
+      "step": 8520
+    },
+    {
+      "epoch": 0.15423525396606552,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.97590078241938e-06,
+      "loss": 17.3447,
+      "step": 8530
+    },
+    {
+      "epoch": 0.1544160690351934,
+      "grad_norm": 37.8125,
+      "learning_rate": 9.975872530112723e-06,
+      "loss": 17.3876,
+      "step": 8540
+    },
+    {
+      "epoch": 0.15459688410432126,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.975844277806064e-06,
+      "loss": 17.5301,
+      "step": 8550
+    },
+    {
+      "epoch": 0.1547776991734491,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.975816025499402e-06,
+      "loss": 17.4396,
+      "step": 8560
+    },
+    {
+      "epoch": 0.15495851424257698,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.975787773192743e-06,
+      "loss": 17.4825,
+      "step": 8570
+    },
+    {
+      "epoch": 0.15513932931170485,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.975759520886084e-06,
+      "loss": 17.7308,
+      "step": 8580
+    },
+    {
+      "epoch": 0.1553201443808327,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.975731268579424e-06,
+      "loss": 17.5016,
+      "step": 8590
+    },
+    {
+      "epoch": 0.15550095944996056,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.975703016272765e-06,
+      "loss": 17.2755,
+      "step": 8600
+    },
+    {
+      "epoch": 0.15568177451908843,
+      "grad_norm": 37.09375,
+      "learning_rate": 9.975674763966104e-06,
+      "loss": 17.3056,
+      "step": 8610
+    },
+    {
+      "epoch": 0.15586258958821628,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.975646511659444e-06,
+      "loss": 17.0374,
+      "step": 8620
+    },
+    {
+      "epoch": 0.15604340465734415,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.975618259352787e-06,
+      "loss": 17.2957,
+      "step": 8630
+    },
+    {
+      "epoch": 0.156224219726472,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.975590007046126e-06,
+      "loss": 17.1892,
+      "step": 8640
+    },
+    {
+      "epoch": 0.15640503479559986,
+      "grad_norm": 37.3125,
+      "learning_rate": 9.975561754739466e-06,
+      "loss": 17.7739,
+      "step": 8650
+    },
+    {
+      "epoch": 0.15658584986472773,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.975533502432807e-06,
+      "loss": 17.861,
+      "step": 8660
+    },
+    {
+      "epoch": 0.15676666493385558,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.975505250126148e-06,
+      "loss": 17.779,
+      "step": 8670
+    },
+    {
+      "epoch": 0.15694748000298345,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.975476997819488e-06,
+      "loss": 17.6016,
+      "step": 8680
+    },
+    {
+      "epoch": 0.15712829507211132,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.975448745512827e-06,
+      "loss": 17.1837,
+      "step": 8690
+    },
+    {
+      "epoch": 0.15730911014123916,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.975420493206168e-06,
+      "loss": 17.3432,
+      "step": 8700
+    },
+    {
+      "epoch": 0.15748992521036703,
+      "grad_norm": 40.5,
+      "learning_rate": 9.975392240899508e-06,
+      "loss": 17.4633,
+      "step": 8710
+    },
+    {
+      "epoch": 0.1576707402794949,
+      "grad_norm": 40.875,
+      "learning_rate": 9.97536398859285e-06,
+      "loss": 17.3936,
+      "step": 8720
+    },
+    {
+      "epoch": 0.15785155534862275,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.97533573628619e-06,
+      "loss": 17.2713,
+      "step": 8730
+    },
+    {
+      "epoch": 0.15803237041775062,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.97530748397953e-06,
+      "loss": 17.163,
+      "step": 8740
+    },
+    {
+      "epoch": 0.1582131854868785,
+      "grad_norm": 41.75,
+      "learning_rate": 9.975279231672871e-06,
+      "loss": 17.3989,
+      "step": 8750
+    },
+    {
+      "epoch": 0.15839400055600633,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.975250979366211e-06,
+      "loss": 17.3244,
+      "step": 8760
+    },
+    {
+      "epoch": 0.1585748156251342,
+      "grad_norm": 41.0,
+      "learning_rate": 9.975222727059552e-06,
+      "loss": 17.2113,
+      "step": 8770
+    },
+    {
+      "epoch": 0.15875563069426207,
+      "grad_norm": 37.875,
+      "learning_rate": 9.975194474752891e-06,
+      "loss": 17.142,
+      "step": 8780
+    },
+    {
+      "epoch": 0.15893644576338992,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.975166222446232e-06,
+      "loss": 17.2934,
+      "step": 8790
+    },
+    {
+      "epoch": 0.1591172608325178,
+      "grad_norm": 41.5,
+      "learning_rate": 9.975137970139574e-06,
+      "loss": 17.4551,
+      "step": 8800
+    },
+    {
+      "epoch": 0.15929807590164563,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.975109717832913e-06,
+      "loss": 17.2555,
+      "step": 8810
+    },
+    {
+      "epoch": 0.1594788909707735,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.975081465526254e-06,
+      "loss": 17.4699,
+      "step": 8820
+    },
+    {
+      "epoch": 0.15965970603990137,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.975053213219594e-06,
+      "loss": 17.3129,
+      "step": 8830
+    },
+    {
+      "epoch": 0.15984052110902922,
+      "grad_norm": 38.34375,
+      "learning_rate": 9.975024960912935e-06,
+      "loss": 17.7102,
+      "step": 8840
+    },
+    {
+      "epoch": 0.1600213361781571,
+      "grad_norm": 37.28125,
+      "learning_rate": 9.974996708606275e-06,
+      "loss": 16.9641,
+      "step": 8850
+    },
+    {
+      "epoch": 0.16020215124728496,
+      "grad_norm": 41.25,
+      "learning_rate": 9.974968456299616e-06,
+      "loss": 17.6302,
+      "step": 8860
+    },
+    {
+      "epoch": 0.1603829663164128,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.974940203992955e-06,
+      "loss": 17.493,
+      "step": 8870
+    },
+    {
+      "epoch": 0.16056378138554067,
+      "grad_norm": 38.625,
+      "learning_rate": 9.974911951686296e-06,
+      "loss": 17.4427,
+      "step": 8880
+    },
+    {
+      "epoch": 0.16074459645466854,
+      "grad_norm": 41.375,
+      "learning_rate": 9.974883699379638e-06,
+      "loss": 17.3359,
+      "step": 8890
+    },
+    {
+      "epoch": 0.16092541152379639,
+      "grad_norm": 38.375,
+      "learning_rate": 9.974855447072977e-06,
+      "loss": 17.4055,
+      "step": 8900
+    },
+    {
+      "epoch": 0.16110622659292426,
+      "grad_norm": 38.34375,
+      "learning_rate": 9.974827194766317e-06,
+      "loss": 17.0807,
+      "step": 8910
+    },
+    {
+      "epoch": 0.16128704166205213,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.974798942459658e-06,
+      "loss": 17.5087,
+      "step": 8920
+    },
+    {
+      "epoch": 0.16146785673117997,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.974770690152999e-06,
+      "loss": 17.475,
+      "step": 8930
+    },
+    {
+      "epoch": 0.16164867180030784,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.97474243784634e-06,
+      "loss": 17.2472,
+      "step": 8940
+    },
+    {
+      "epoch": 0.1618294868694357,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.974714185539678e-06,
+      "loss": 17.5978,
+      "step": 8950
+    },
+    {
+      "epoch": 0.16201030193856356,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.974685933233019e-06,
+      "loss": 17.1928,
+      "step": 8960
+    },
+    {
+      "epoch": 0.16219111700769143,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.97465768092636e-06,
+      "loss": 17.3915,
+      "step": 8970
+    },
+    {
+      "epoch": 0.16237193207681927,
+      "grad_norm": 42.625,
+      "learning_rate": 9.974629428619702e-06,
+      "loss": 17.7899,
+      "step": 8980
+    },
+    {
+      "epoch": 0.16255274714594714,
+      "grad_norm": 39.375,
+      "learning_rate": 9.97460117631304e-06,
+      "loss": 17.9558,
+      "step": 8990
+    },
+    {
+      "epoch": 0.162733562215075,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.974572924006381e-06,
+      "loss": 17.5389,
+      "step": 9000
+    },
+    {
+      "epoch": 0.16291437728420285,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.974544671699722e-06,
+      "loss": 17.6044,
+      "step": 9010
+    },
+    {
+      "epoch": 0.16309519235333073,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.974516419393063e-06,
+      "loss": 17.5651,
+      "step": 9020
+    },
+    {
+      "epoch": 0.1632760074224586,
+      "grad_norm": 40.25,
+      "learning_rate": 9.974488167086403e-06,
+      "loss": 17.4637,
+      "step": 9030
+    },
+    {
+      "epoch": 0.16345682249158644,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.974459914779742e-06,
+      "loss": 17.348,
+      "step": 9040
+    },
+    {
+      "epoch": 0.1636376375607143,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.974431662473083e-06,
+      "loss": 17.2592,
+      "step": 9050
+    },
+    {
+      "epoch": 0.16381845262984218,
+      "grad_norm": 39.75,
+      "learning_rate": 9.974403410166423e-06,
+      "loss": 17.5499,
+      "step": 9060
+    },
+    {
+      "epoch": 0.16399926769897002,
+      "grad_norm": 37.625,
+      "learning_rate": 9.974375157859764e-06,
+      "loss": 17.1442,
+      "step": 9070
+    },
+    {
+      "epoch": 0.1641800827680979,
+      "grad_norm": 40.5,
+      "learning_rate": 9.974346905553105e-06,
+      "loss": 17.0087,
+      "step": 9080
+    },
+    {
+      "epoch": 0.16436089783722577,
+      "grad_norm": 38.125,
+      "learning_rate": 9.974318653246445e-06,
+      "loss": 17.2556,
+      "step": 9090
+    },
+    {
+      "epoch": 0.1645417129063536,
+      "grad_norm": 37.9375,
+      "learning_rate": 9.974290400939786e-06,
+      "loss": 17.1482,
+      "step": 9100
+    },
+    {
+      "epoch": 0.16472252797548148,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.974262148633126e-06,
+      "loss": 17.108,
+      "step": 9110
+    },
+    {
+      "epoch": 0.16490334304460932,
+      "grad_norm": 39.875,
+      "learning_rate": 9.974233896326465e-06,
+      "loss": 17.4758,
+      "step": 9120
+    },
+    {
+      "epoch": 0.1650841581137372,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.974205644019806e-06,
+      "loss": 17.3085,
+      "step": 9130
+    },
+    {
+      "epoch": 0.16526497318286507,
+      "grad_norm": 38.59375,
+      "learning_rate": 9.974177391713147e-06,
+      "loss": 17.3685,
+      "step": 9140
+    },
+    {
+      "epoch": 0.1654457882519929,
+      "grad_norm": 43.0,
+      "learning_rate": 9.974149139406489e-06,
+      "loss": 17.382,
+      "step": 9150
+    },
+    {
+      "epoch": 0.16562660332112078,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.974120887099828e-06,
+      "loss": 17.3543,
+      "step": 9160
+    },
+    {
+      "epoch": 0.16580741839024865,
+      "grad_norm": 41.0,
+      "learning_rate": 9.974092634793169e-06,
+      "loss": 17.5792,
+      "step": 9170
+    },
+    {
+      "epoch": 0.1659882334593765,
+      "grad_norm": 38.46875,
+      "learning_rate": 9.97406438248651e-06,
+      "loss": 17.8702,
+      "step": 9180
+    },
+    {
+      "epoch": 0.16616904852850437,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.97403613017985e-06,
+      "loss": 17.4478,
+      "step": 9190
+    },
+    {
+      "epoch": 0.16634986359763224,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.97400787787319e-06,
+      "loss": 17.2742,
+      "step": 9200
+    },
+    {
+      "epoch": 0.16653067866676008,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.97397962556653e-06,
+      "loss": 17.5207,
+      "step": 9210
+    },
+    {
+      "epoch": 0.16671149373588795,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.97395137325987e-06,
+      "loss": 17.5423,
+      "step": 9220
+    },
+    {
+      "epoch": 0.16689230880501582,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.97392312095321e-06,
+      "loss": 17.708,
+      "step": 9230
+    },
+    {
+      "epoch": 0.16707312387414366,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.973894868646551e-06,
+      "loss": 17.2694,
+      "step": 9240
+    },
+    {
+      "epoch": 0.16725393894327154,
+      "grad_norm": 39.75,
+      "learning_rate": 9.973866616339892e-06,
+      "loss": 17.4095,
+      "step": 9250
+    },
+    {
+      "epoch": 0.1674347540123994,
+      "grad_norm": 41.375,
+      "learning_rate": 9.973838364033232e-06,
+      "loss": 17.4842,
+      "step": 9260
+    },
+    {
+      "epoch": 0.16761556908152725,
+      "grad_norm": 38.125,
+      "learning_rate": 9.973810111726573e-06,
+      "loss": 17.4138,
+      "step": 9270
+    },
+    {
+      "epoch": 0.16779638415065512,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.973781859419914e-06,
+      "loss": 17.4215,
+      "step": 9280
+    },
+    {
+      "epoch": 0.16797719921978296,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.973753607113254e-06,
+      "loss": 17.5086,
+      "step": 9290
+    },
+    {
+      "epoch": 0.16815801428891083,
+      "grad_norm": 38.34375,
+      "learning_rate": 9.973725354806593e-06,
+      "loss": 17.5304,
+      "step": 9300
+    },
+    {
+      "epoch": 0.1683388293580387,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.973697102499934e-06,
+      "loss": 17.6362,
+      "step": 9310
+    },
+    {
+      "epoch": 0.16851964442716655,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.973668850193274e-06,
+      "loss": 17.7299,
+      "step": 9320
+    },
+    {
+      "epoch": 0.16870045949629442,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.973640597886615e-06,
+      "loss": 17.6307,
+      "step": 9330
+    },
+    {
+      "epoch": 0.1688812745654223,
+      "grad_norm": 39.125,
+      "learning_rate": 9.973612345579956e-06,
+      "loss": 17.3122,
+      "step": 9340
+    },
+    {
+      "epoch": 0.16906208963455013,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.973584093273296e-06,
+      "loss": 17.3277,
+      "step": 9350
+    },
+    {
+      "epoch": 0.169242904703678,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.973555840966637e-06,
+      "loss": 17.4785,
+      "step": 9360
+    },
+    {
+      "epoch": 0.16942371977280588,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.973527588659978e-06,
+      "loss": 17.1665,
+      "step": 9370
+    },
+    {
+      "epoch": 0.16960453484193372,
+      "grad_norm": 39.625,
+      "learning_rate": 9.973499336353317e-06,
+      "loss": 17.2563,
+      "step": 9380
+    },
+    {
+      "epoch": 0.1697853499110616,
+      "grad_norm": 37.71875,
+      "learning_rate": 9.973471084046657e-06,
+      "loss": 17.3261,
+      "step": 9390
+    },
+    {
+      "epoch": 0.16996616498018946,
+      "grad_norm": 38.5,
+      "learning_rate": 9.973442831739998e-06,
+      "loss": 17.3375,
+      "step": 9400
+    },
+    {
+      "epoch": 0.1701469800493173,
+      "grad_norm": 43.25,
+      "learning_rate": 9.973414579433338e-06,
+      "loss": 17.1612,
+      "step": 9410
+    },
+    {
+      "epoch": 0.17032779511844517,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.973386327126679e-06,
+      "loss": 17.3999,
+      "step": 9420
+    },
+    {
+      "epoch": 0.17050861018757305,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.97335807482002e-06,
+      "loss": 17.4043,
+      "step": 9430
+    },
+    {
+      "epoch": 0.1706894252567009,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.97332982251336e-06,
+      "loss": 17.2566,
+      "step": 9440
+    },
+    {
+      "epoch": 0.17087024032582876,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.973301570206701e-06,
+      "loss": 17.311,
+      "step": 9450
+    },
+    {
+      "epoch": 0.1710510553949566,
+      "grad_norm": 41.25,
+      "learning_rate": 9.973273317900041e-06,
+      "loss": 17.0837,
+      "step": 9460
+    },
+    {
+      "epoch": 0.17123187046408447,
+      "grad_norm": 37.28125,
+      "learning_rate": 9.97324506559338e-06,
+      "loss": 17.1674,
+      "step": 9470
+    },
+    {
+      "epoch": 0.17141268553321234,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.973216813286721e-06,
+      "loss": 17.3728,
+      "step": 9480
+    },
+    {
+      "epoch": 0.1715935006023402,
+      "grad_norm": 42.125,
+      "learning_rate": 9.973188560980062e-06,
+      "loss": 17.5858,
+      "step": 9490
+    },
+    {
+      "epoch": 0.17177431567146806,
+      "grad_norm": 37.5625,
+      "learning_rate": 9.973160308673402e-06,
+      "loss": 17.2968,
+      "step": 9500
+    },
+    {
+      "epoch": 0.17195513074059593,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.973132056366743e-06,
+      "loss": 17.3369,
+      "step": 9510
+    },
+    {
+      "epoch": 0.17213594580972377,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.973103804060084e-06,
+      "loss": 17.0527,
+      "step": 9520
+    },
+    {
+      "epoch": 0.17231676087885164,
+      "grad_norm": 40.75,
+      "learning_rate": 9.973075551753424e-06,
+      "loss": 17.159,
+      "step": 9530
+    },
+    {
+      "epoch": 0.17249757594797951,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.973047299446765e-06,
+      "loss": 17.5794,
+      "step": 9540
+    },
+    {
+      "epoch": 0.17267839101710736,
+      "grad_norm": 42.0,
+      "learning_rate": 9.973019047140104e-06,
+      "loss": 17.6003,
+      "step": 9550
+    },
+    {
+      "epoch": 0.17285920608623523,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.972990794833444e-06,
+      "loss": 17.372,
+      "step": 9560
+    },
+    {
+      "epoch": 0.1730400211553631,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.972962542526785e-06,
+      "loss": 16.9928,
+      "step": 9570
+    },
+    {
+      "epoch": 0.17322083622449094,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.972934290220126e-06,
+      "loss": 17.6939,
+      "step": 9580
+    },
+    {
+      "epoch": 0.1734016512936188,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.972906037913466e-06,
+      "loss": 17.4905,
+      "step": 9590
+    },
+    {
+      "epoch": 0.17358246636274668,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.972877785606807e-06,
+      "loss": 17.1858,
+      "step": 9600
+    },
+    {
+      "epoch": 0.17376328143187453,
+      "grad_norm": 39.125,
+      "learning_rate": 9.972849533300147e-06,
+      "loss": 17.3789,
+      "step": 9610
+    },
+    {
+      "epoch": 0.1739440965010024,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.972821280993488e-06,
+      "loss": 17.4596,
+      "step": 9620
+    },
+    {
+      "epoch": 0.17412491157013024,
+      "grad_norm": 41.25,
+      "learning_rate": 9.972793028686829e-06,
+      "loss": 17.162,
+      "step": 9630
+    },
+    {
+      "epoch": 0.1743057266392581,
+      "grad_norm": 42.25,
+      "learning_rate": 9.972764776380168e-06,
+      "loss": 17.4351,
+      "step": 9640
+    },
+    {
+      "epoch": 0.17448654170838598,
+      "grad_norm": 40.625,
+      "learning_rate": 9.972736524073508e-06,
+      "loss": 17.6255,
+      "step": 9650
+    },
+    {
+      "epoch": 0.17466735677751383,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.972708271766849e-06,
+      "loss": 17.2791,
+      "step": 9660
+    },
+    {
+      "epoch": 0.1748481718466417,
+      "grad_norm": 43.0,
+      "learning_rate": 9.97268001946019e-06,
+      "loss": 17.5448,
+      "step": 9670
+    },
+    {
+      "epoch": 0.17502898691576957,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.97265176715353e-06,
+      "loss": 17.4885,
+      "step": 9680
+    },
+    {
+      "epoch": 0.1752098019848974,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.97262351484687e-06,
+      "loss": 17.652,
+      "step": 9690
+    },
+    {
+      "epoch": 0.17539061705402528,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.972595262540211e-06,
+      "loss": 17.7164,
+      "step": 9700
+    },
+    {
+      "epoch": 0.17557143212315315,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.972567010233552e-06,
+      "loss": 17.4173,
+      "step": 9710
+    },
+    {
+      "epoch": 0.175752247192281,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.972538757926891e-06,
+      "loss": 17.8984,
+      "step": 9720
+    },
+    {
+      "epoch": 0.17593306226140887,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.972510505620232e-06,
+      "loss": 17.6743,
+      "step": 9730
+    },
+    {
+      "epoch": 0.17611387733053674,
+      "grad_norm": 41.875,
+      "learning_rate": 9.972482253313572e-06,
+      "loss": 17.1897,
+      "step": 9740
+    },
+    {
+      "epoch": 0.17629469239966458,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.972454001006913e-06,
+      "loss": 17.4059,
+      "step": 9750
+    },
+    {
+      "epoch": 0.17647550746879245,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.972425748700253e-06,
+      "loss": 17.2298,
+      "step": 9760
+    },
+    {
+      "epoch": 0.17665632253792032,
+      "grad_norm": 39.25,
+      "learning_rate": 9.972397496393594e-06,
+      "loss": 17.4787,
+      "step": 9770
+    },
+    {
+      "epoch": 0.17683713760704817,
+      "grad_norm": 39.0,
+      "learning_rate": 9.972369244086935e-06,
+      "loss": 16.8661,
+      "step": 9780
+    },
+    {
+      "epoch": 0.17701795267617604,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.972340991780275e-06,
+      "loss": 16.9897,
+      "step": 9790
+    },
+    {
+      "epoch": 0.17719876774530388,
+      "grad_norm": 40.625,
+      "learning_rate": 9.972312739473616e-06,
+      "loss": 17.3103,
+      "step": 9800
+    },
+    {
+      "epoch": 0.17737958281443175,
+      "grad_norm": 41.75,
+      "learning_rate": 9.972284487166955e-06,
+      "loss": 17.4173,
+      "step": 9810
+    },
+    {
+      "epoch": 0.17756039788355962,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.972256234860295e-06,
+      "loss": 17.5176,
+      "step": 9820
+    },
+    {
+      "epoch": 0.17774121295268747,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.972227982553636e-06,
+      "loss": 17.8892,
+      "step": 9830
+    },
+    {
+      "epoch": 0.17792202802181534,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.972199730246977e-06,
+      "loss": 17.3229,
+      "step": 9840
+    },
+    {
+      "epoch": 0.1781028430909432,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.972171477940317e-06,
+      "loss": 17.4932,
+      "step": 9850
+    },
+    {
+      "epoch": 0.17828365816007105,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.972143225633658e-06,
+      "loss": 17.1002,
+      "step": 9860
+    },
+    {
+      "epoch": 0.17846447322919892,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.972114973326999e-06,
+      "loss": 17.3818,
+      "step": 9870
+    },
+    {
+      "epoch": 0.1786452882983268,
+      "grad_norm": 40.25,
+      "learning_rate": 9.97208672102034e-06,
+      "loss": 17.3064,
+      "step": 9880
+    },
+    {
+      "epoch": 0.17882610336745464,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.97205846871368e-06,
+      "loss": 17.0175,
+      "step": 9890
+    },
+    {
+      "epoch": 0.1790069184365825,
+      "grad_norm": 40.25,
+      "learning_rate": 9.972030216407019e-06,
+      "loss": 17.5675,
+      "step": 9900
+    },
+    {
+      "epoch": 0.17918773350571038,
+      "grad_norm": 40.75,
+      "learning_rate": 9.97200196410036e-06,
+      "loss": 17.2624,
+      "step": 9910
+    },
+    {
+      "epoch": 0.17936854857483822,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.9719737117937e-06,
+      "loss": 16.7743,
+      "step": 9920
+    },
+    {
+      "epoch": 0.1795493636439661,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.97194545948704e-06,
+      "loss": 17.372,
+      "step": 9930
+    },
+    {
+      "epoch": 0.17973017871309394,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.971917207180381e-06,
+      "loss": 17.2744,
+      "step": 9940
+    },
+    {
+      "epoch": 0.1799109937822218,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.971888954873722e-06,
+      "loss": 17.3897,
+      "step": 9950
+    },
+    {
+      "epoch": 0.18009180885134968,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.971860702567062e-06,
+      "loss": 17.1437,
+      "step": 9960
+    },
+    {
+      "epoch": 0.18027262392047752,
+      "grad_norm": 41.0,
+      "learning_rate": 9.971832450260403e-06,
+      "loss": 17.4114,
+      "step": 9970
+    },
+    {
+      "epoch": 0.1804534389896054,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.971804197953742e-06,
+      "loss": 17.2309,
+      "step": 9980
+    },
+    {
+      "epoch": 0.18063425405873326,
+      "grad_norm": 38.46875,
+      "learning_rate": 9.971775945647083e-06,
+      "loss": 17.4542,
+      "step": 9990
+    },
+    {
+      "epoch": 0.1808150691278611,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.971747693340423e-06,
+      "loss": 17.4604,
+      "step": 10000
+    },
+    {
+      "epoch": 0.1808150691278611,
+      "eval_loss": 2.173656463623047,
+      "eval_runtime": 230.3398,
+      "eval_samples_per_second": 3152.122,
+      "eval_steps_per_second": 49.253,
+      "step": 10000
+    },
+    {
+      "epoch": 0.18099588419698898,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.971719441033764e-06,
+      "loss": 17.6182,
+      "step": 10010
+    },
+    {
+      "epoch": 0.18117669926611685,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.971691188727104e-06,
+      "loss": 17.3686,
+      "step": 10020
+    },
+    {
+      "epoch": 0.1813575143352447,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.971662936420445e-06,
+      "loss": 16.8379,
+      "step": 10030
+    },
+    {
+      "epoch": 0.18153832940437256,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.971634684113786e-06,
+      "loss": 17.3498,
+      "step": 10040
+    },
+    {
+      "epoch": 0.18171914447350043,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.971606431807126e-06,
+      "loss": 17.3983,
+      "step": 10050
+    },
+    {
+      "epoch": 0.18189995954262828,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.971578179500467e-06,
+      "loss": 17.4824,
+      "step": 10060
+    },
+    {
+      "epoch": 0.18208077461175615,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.971549927193806e-06,
+      "loss": 17.3409,
+      "step": 10070
+    },
+    {
+      "epoch": 0.18226158968088402,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.971521674887147e-06,
+      "loss": 17.2479,
+      "step": 10080
+    },
+    {
+      "epoch": 0.18244240475001186,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.971493422580487e-06,
+      "loss": 17.4112,
+      "step": 10090
+    },
+    {
+      "epoch": 0.18262321981913973,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.971465170273828e-06,
+      "loss": 17.4876,
+      "step": 10100
+    },
+    {
+      "epoch": 0.18280403488826757,
+      "grad_norm": 39.125,
+      "learning_rate": 9.971436917967168e-06,
+      "loss": 17.2352,
+      "step": 10110
+    },
+    {
+      "epoch": 0.18298484995739545,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.971408665660509e-06,
+      "loss": 17.4973,
+      "step": 10120
+    },
+    {
+      "epoch": 0.18316566502652332,
+      "grad_norm": 38.875,
+      "learning_rate": 9.97138041335385e-06,
+      "loss": 17.0968,
+      "step": 10130
+    },
+    {
+      "epoch": 0.18334648009565116,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.97135216104719e-06,
+      "loss": 17.0191,
+      "step": 10140
+    },
+    {
+      "epoch": 0.18352729516477903,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.97132390874053e-06,
+      "loss": 17.3471,
+      "step": 10150
+    },
+    {
+      "epoch": 0.1837081102339069,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.97129565643387e-06,
+      "loss": 17.1329,
+      "step": 10160
+    },
+    {
+      "epoch": 0.18388892530303474,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.97126740412721e-06,
+      "loss": 17.4335,
+      "step": 10170
+    },
+    {
+      "epoch": 0.18406974037216262,
+      "grad_norm": 38.75,
+      "learning_rate": 9.971239151820551e-06,
+      "loss": 17.201,
+      "step": 10180
+    },
+    {
+      "epoch": 0.1842505554412905,
+      "grad_norm": 41.125,
+      "learning_rate": 9.971210899513892e-06,
+      "loss": 17.3499,
+      "step": 10190
+    },
+    {
+      "epoch": 0.18443137051041833,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.971182647207232e-06,
+      "loss": 17.1541,
+      "step": 10200
+    },
+    {
+      "epoch": 0.1846121855795462,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.971154394900573e-06,
+      "loss": 17.4216,
+      "step": 10210
+    },
+    {
+      "epoch": 0.18479300064867407,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.971126142593914e-06,
+      "loss": 17.228,
+      "step": 10220
+    },
+    {
+      "epoch": 0.18497381571780191,
+      "grad_norm": 42.0,
+      "learning_rate": 9.971097890287254e-06,
+      "loss": 17.3865,
+      "step": 10230
+    },
+    {
+      "epoch": 0.18515463078692979,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.971069637980593e-06,
+      "loss": 17.1773,
+      "step": 10240
+    },
+    {
+      "epoch": 0.18533544585605766,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.971041385673934e-06,
+      "loss": 17.2779,
+      "step": 10250
+    },
+    {
+      "epoch": 0.1855162609251855,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.971013133367274e-06,
+      "loss": 17.4225,
+      "step": 10260
+    },
+    {
+      "epoch": 0.18569707599431337,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.970984881060615e-06,
+      "loss": 17.1712,
+      "step": 10270
+    },
+    {
+      "epoch": 0.18587789106344121,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.970956628753956e-06,
+      "loss": 17.1064,
+      "step": 10280
+    },
+    {
+      "epoch": 0.18605870613256908,
+      "grad_norm": 38.875,
+      "learning_rate": 9.970928376447296e-06,
+      "loss": 17.4685,
+      "step": 10290
+    },
+    {
+      "epoch": 0.18623952120169696,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.970900124140637e-06,
+      "loss": 17.152,
+      "step": 10300
+    },
+    {
+      "epoch": 0.1864203362708248,
+      "grad_norm": 40.875,
+      "learning_rate": 9.970871871833977e-06,
+      "loss": 17.4354,
+      "step": 10310
+    },
+    {
+      "epoch": 0.18660115133995267,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.970843619527318e-06,
+      "loss": 17.0856,
+      "step": 10320
+    },
+    {
+      "epoch": 0.18678196640908054,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.970815367220657e-06,
+      "loss": 17.2967,
+      "step": 10330
+    },
+    {
+      "epoch": 0.18696278147820838,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.970787114913998e-06,
+      "loss": 17.6692,
+      "step": 10340
+    },
+    {
+      "epoch": 0.18714359654733626,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.970758862607338e-06,
+      "loss": 17.5644,
+      "step": 10350
+    },
+    {
+      "epoch": 0.18732441161646413,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.970730610300679e-06,
+      "loss": 17.5649,
+      "step": 10360
+    },
+    {
+      "epoch": 0.18750522668559197,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.97070235799402e-06,
+      "loss": 17.1662,
+      "step": 10370
+    },
+    {
+      "epoch": 0.18768604175471984,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.97067410568736e-06,
+      "loss": 17.7755,
+      "step": 10380
+    },
+    {
+      "epoch": 0.1878668568238477,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.9706458533807e-06,
+      "loss": 17.0753,
+      "step": 10390
+    },
+    {
+      "epoch": 0.18804767189297555,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.970617601074041e-06,
+      "loss": 17.3117,
+      "step": 10400
+    },
+    {
+      "epoch": 0.18822848696210343,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.97058934876738e-06,
+      "loss": 17.3471,
+      "step": 10410
+    },
+    {
+      "epoch": 0.1884093020312313,
+      "grad_norm": 37.625,
+      "learning_rate": 9.970561096460721e-06,
+      "loss": 17.3984,
+      "step": 10420
+    },
+    {
+      "epoch": 0.18859011710035914,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.970532844154062e-06,
+      "loss": 17.0128,
+      "step": 10430
+    },
+    {
+      "epoch": 0.188770932169487,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.970504591847402e-06,
+      "loss": 17.2128,
+      "step": 10440
+    },
+    {
+      "epoch": 0.18895174723861485,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.970476339540743e-06,
+      "loss": 17.5359,
+      "step": 10450
+    },
+    {
+      "epoch": 0.18913256230774272,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.970448087234082e-06,
+      "loss": 16.8257,
+      "step": 10460
+    },
+    {
+      "epoch": 0.1893133773768706,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.970419834927424e-06,
+      "loss": 17.617,
+      "step": 10470
+    },
+    {
+      "epoch": 0.18949419244599844,
+      "grad_norm": 40.0,
+      "learning_rate": 9.970391582620765e-06,
+      "loss": 18.0481,
+      "step": 10480
+    },
+    {
+      "epoch": 0.1896750075151263,
+      "grad_norm": 42.375,
+      "learning_rate": 9.970363330314105e-06,
+      "loss": 17.4429,
+      "step": 10490
+    },
+    {
+      "epoch": 0.18985582258425418,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.970335078007444e-06,
+      "loss": 17.5725,
+      "step": 10500
+    },
+    {
+      "epoch": 0.19003663765338202,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.970306825700785e-06,
+      "loss": 17.3861,
+      "step": 10510
+    },
+    {
+      "epoch": 0.1902174527225099,
+      "grad_norm": 38.875,
+      "learning_rate": 9.970278573394125e-06,
+      "loss": 17.8106,
+      "step": 10520
+    },
+    {
+      "epoch": 0.19039826779163777,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.970250321087466e-06,
+      "loss": 17.1235,
+      "step": 10530
+    },
+    {
+      "epoch": 0.1905790828607656,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.970222068780807e-06,
+      "loss": 17.1585,
+      "step": 10540
+    },
+    {
+      "epoch": 0.19075989792989348,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.970193816474146e-06,
+      "loss": 17.3798,
+      "step": 10550
+    },
+    {
+      "epoch": 0.19094071299902135,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.970165564167488e-06,
+      "loss": 16.9925,
+      "step": 10560
+    },
+    {
+      "epoch": 0.1911215280681492,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.970137311860829e-06,
+      "loss": 17.6112,
+      "step": 10570
+    },
+    {
+      "epoch": 0.19130234313727706,
+      "grad_norm": 41.5,
+      "learning_rate": 9.970109059554167e-06,
+      "loss": 17.0915,
+      "step": 10580
+    },
+    {
+      "epoch": 0.19148315820640494,
+      "grad_norm": 37.09375,
+      "learning_rate": 9.970080807247508e-06,
+      "loss": 17.5033,
+      "step": 10590
+    },
+    {
+      "epoch": 0.19166397327553278,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.970052554940849e-06,
+      "loss": 17.2897,
+      "step": 10600
+    },
+    {
+      "epoch": 0.19184478834466065,
+      "grad_norm": 41.375,
+      "learning_rate": 9.97002430263419e-06,
+      "loss": 17.5192,
+      "step": 10610
+    },
+    {
+      "epoch": 0.1920256034137885,
+      "grad_norm": 39.125,
+      "learning_rate": 9.96999605032753e-06,
+      "loss": 17.6841,
+      "step": 10620
+    },
+    {
+      "epoch": 0.19220641848291636,
+      "grad_norm": 37.90625,
+      "learning_rate": 9.96996779802087e-06,
+      "loss": 17.2809,
+      "step": 10630
+    },
+    {
+      "epoch": 0.19238723355204423,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.969939545714211e-06,
+      "loss": 17.3026,
+      "step": 10640
+    },
+    {
+      "epoch": 0.19256804862117208,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.969911293407552e-06,
+      "loss": 17.5597,
+      "step": 10650
+    },
+    {
+      "epoch": 0.19274886369029995,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.969883041100892e-06,
+      "loss": 17.0397,
+      "step": 10660
+    },
+    {
+      "epoch": 0.19292967875942782,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.969854788794231e-06,
+      "loss": 17.7761,
+      "step": 10670
+    },
+    {
+      "epoch": 0.19311049382855566,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.969826536487572e-06,
+      "loss": 17.5117,
+      "step": 10680
+    },
+    {
+      "epoch": 0.19329130889768353,
+      "grad_norm": 38.0,
+      "learning_rate": 9.969798284180913e-06,
+      "loss": 17.26,
+      "step": 10690
+    },
+    {
+      "epoch": 0.1934721239668114,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.969770031874253e-06,
+      "loss": 17.8356,
+      "step": 10700
+    },
+    {
+      "epoch": 0.19365293903593925,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.969741779567594e-06,
+      "loss": 17.7687,
+      "step": 10710
+    },
+    {
+      "epoch": 0.19383375410506712,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.969713527260933e-06,
+      "loss": 17.4783,
+      "step": 10720
+    },
+    {
+      "epoch": 0.194014569174195,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.969685274954275e-06,
+      "loss": 17.7937,
+      "step": 10730
+    },
+    {
+      "epoch": 0.19419538424332283,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.969657022647616e-06,
+      "loss": 17.655,
+      "step": 10740
+    },
+    {
+      "epoch": 0.1943761993124507,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.969628770340956e-06,
+      "loss": 17.4106,
+      "step": 10750
+    },
+    {
+      "epoch": 0.19455701438157855,
+      "grad_norm": 40.75,
+      "learning_rate": 9.969600518034295e-06,
+      "loss": 17.2935,
+      "step": 10760
+    },
+    {
+      "epoch": 0.19473782945070642,
+      "grad_norm": 38.75,
+      "learning_rate": 9.969572265727636e-06,
+      "loss": 17.3053,
+      "step": 10770
+    },
+    {
+      "epoch": 0.1949186445198343,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.969544013420977e-06,
+      "loss": 17.6152,
+      "step": 10780
+    },
+    {
+      "epoch": 0.19509945958896213,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.969515761114317e-06,
+      "loss": 17.6451,
+      "step": 10790
+    },
+    {
+      "epoch": 0.19528027465809,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.969487508807658e-06,
+      "loss": 17.323,
+      "step": 10800
+    },
+    {
+      "epoch": 0.19546108972721787,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.969459256500997e-06,
+      "loss": 17.0142,
+      "step": 10810
+    },
+    {
+      "epoch": 0.19564190479634572,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.969431004194339e-06,
+      "loss": 17.7659,
+      "step": 10820
+    },
+    {
+      "epoch": 0.1958227198654736,
+      "grad_norm": 43.375,
+      "learning_rate": 9.96940275188768e-06,
+      "loss": 17.0923,
+      "step": 10830
+    },
+    {
+      "epoch": 0.19600353493460146,
+      "grad_norm": 37.625,
+      "learning_rate": 9.969374499581019e-06,
+      "loss": 17.2729,
+      "step": 10840
+    },
+    {
+      "epoch": 0.1961843500037293,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.96934624727436e-06,
+      "loss": 17.4684,
+      "step": 10850
+    },
+    {
+      "epoch": 0.19636516507285717,
+      "grad_norm": 38.5,
+      "learning_rate": 9.9693179949677e-06,
+      "loss": 17.3878,
+      "step": 10860
+    },
+    {
+      "epoch": 0.19654598014198504,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.96928974266104e-06,
+      "loss": 17.1617,
+      "step": 10870
+    },
+    {
+      "epoch": 0.1967267952111129,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.969261490354381e-06,
+      "loss": 17.1731,
+      "step": 10880
+    },
+    {
+      "epoch": 0.19690761028024076,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.96923323804772e-06,
+      "loss": 17.5819,
+      "step": 10890
+    },
+    {
+      "epoch": 0.19708842534936863,
+      "grad_norm": 39.5,
+      "learning_rate": 9.96920498574106e-06,
+      "loss": 17.3935,
+      "step": 10900
+    },
+    {
+      "epoch": 0.19726924041849647,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.969176733434403e-06,
+      "loss": 17.4576,
+      "step": 10910
+    },
+    {
+      "epoch": 0.19745005548762434,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.969148481127744e-06,
+      "loss": 17.3355,
+      "step": 10920
+    },
+    {
+      "epoch": 0.1976308705567522,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.969120228821082e-06,
+      "loss": 17.1319,
+      "step": 10930
+    },
+    {
+      "epoch": 0.19781168562588006,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.969091976514423e-06,
+      "loss": 17.4143,
+      "step": 10940
+    },
+    {
+      "epoch": 0.19799250069500793,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.969063724207764e-06,
+      "loss": 17.4249,
+      "step": 10950
+    },
+    {
+      "epoch": 0.19817331576413577,
+      "grad_norm": 41.5,
+      "learning_rate": 9.969035471901104e-06,
+      "loss": 17.1293,
+      "step": 10960
+    },
+    {
+      "epoch": 0.19835413083326364,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.969007219594445e-06,
+      "loss": 17.4503,
+      "step": 10970
+    },
+    {
+      "epoch": 0.1985349459023915,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.968978967287784e-06,
+      "loss": 17.0614,
+      "step": 10980
+    },
+    {
+      "epoch": 0.19871576097151936,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.968950714981126e-06,
+      "loss": 17.3571,
+      "step": 10990
+    },
+    {
+      "epoch": 0.19889657604064723,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.968922462674467e-06,
+      "loss": 17.5243,
+      "step": 11000
+    },
+    {
+      "epoch": 0.1990773911097751,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.968894210367806e-06,
+      "loss": 17.303,
+      "step": 11010
+    },
+    {
+      "epoch": 0.19925820617890294,
+      "grad_norm": 40.375,
+      "learning_rate": 9.968865958061146e-06,
+      "loss": 17.3339,
+      "step": 11020
+    },
+    {
+      "epoch": 0.1994390212480308,
+      "grad_norm": 40.75,
+      "learning_rate": 9.968837705754487e-06,
+      "loss": 17.4386,
+      "step": 11030
+    },
+    {
+      "epoch": 0.19961983631715868,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.968809453447828e-06,
+      "loss": 17.4172,
+      "step": 11040
+    },
+    {
+      "epoch": 0.19980065138628653,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.968781201141168e-06,
+      "loss": 17.1305,
+      "step": 11050
+    },
+    {
+      "epoch": 0.1999814664554144,
+      "grad_norm": 39.625,
+      "learning_rate": 9.968752948834509e-06,
+      "loss": 17.4369,
+      "step": 11060
+    },
+    {
+      "epoch": 0.20016228152454227,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.968724696527848e-06,
+      "loss": 16.9403,
+      "step": 11070
+    },
+    {
+      "epoch": 0.2003430965936701,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.96869644422119e-06,
+      "loss": 16.5315,
+      "step": 11080
+    },
+    {
+      "epoch": 0.20052391166279798,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.96866819191453e-06,
+      "loss": 17.3027,
+      "step": 11090
+    },
+    {
+      "epoch": 0.20070472673192583,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.96863993960787e-06,
+      "loss": 17.1912,
+      "step": 11100
+    },
+    {
+      "epoch": 0.2008855418010537,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.96861168730121e-06,
+      "loss": 17.6024,
+      "step": 11110
+    },
+    {
+      "epoch": 0.20106635687018157,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.968583434994551e-06,
+      "loss": 17.4576,
+      "step": 11120
+    },
+    {
+      "epoch": 0.2012471719393094,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.968555182687892e-06,
+      "loss": 17.3747,
+      "step": 11130
+    },
+    {
+      "epoch": 0.20142798700843728,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.968526930381232e-06,
+      "loss": 17.0922,
+      "step": 11140
+    },
+    {
+      "epoch": 0.20160880207756515,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.968498678074571e-06,
+      "loss": 17.3724,
+      "step": 11150
+    },
+    {
+      "epoch": 0.201789617146693,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.968470425767912e-06,
+      "loss": 17.3961,
+      "step": 11160
+    },
+    {
+      "epoch": 0.20197043221582087,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.968442173461254e-06,
+      "loss": 17.4708,
+      "step": 11170
+    },
+    {
+      "epoch": 0.20215124728494874,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.968413921154595e-06,
+      "loss": 17.5796,
+      "step": 11180
+    },
+    {
+      "epoch": 0.20233206235407658,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.968385668847934e-06,
+      "loss": 17.803,
+      "step": 11190
+    },
+    {
+      "epoch": 0.20251287742320445,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.968357416541274e-06,
+      "loss": 17.0907,
+      "step": 11200
+    },
+    {
+      "epoch": 0.20269369249233232,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.968329164234615e-06,
+      "loss": 17.4098,
+      "step": 11210
+    },
+    {
+      "epoch": 0.20287450756146017,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.968300911927955e-06,
+      "loss": 17.3981,
+      "step": 11220
+    },
+    {
+      "epoch": 0.20305532263058804,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.968272659621296e-06,
+      "loss": 17.505,
+      "step": 11230
+    },
+    {
+      "epoch": 0.2032361376997159,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.968244407314635e-06,
+      "loss": 17.1375,
+      "step": 11240
+    },
+    {
+      "epoch": 0.20341695276884375,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.968216155007976e-06,
+      "loss": 17.1601,
+      "step": 11250
+    },
+    {
+      "epoch": 0.20359776783797162,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.968187902701318e-06,
+      "loss": 17.2334,
+      "step": 11260
+    },
+    {
+      "epoch": 0.20377858290709946,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.968159650394657e-06,
+      "loss": 16.8655,
+      "step": 11270
+    },
+    {
+      "epoch": 0.20395939797622734,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.968131398087997e-06,
+      "loss": 17.5414,
+      "step": 11280
+    },
+    {
+      "epoch": 0.2041402130453552,
+      "grad_norm": 38.0,
+      "learning_rate": 9.968103145781338e-06,
+      "loss": 17.5637,
+      "step": 11290
+    },
+    {
+      "epoch": 0.20432102811448305,
+      "grad_norm": 40.625,
+      "learning_rate": 9.968074893474679e-06,
+      "loss": 17.0316,
+      "step": 11300
+    },
+    {
+      "epoch": 0.20450184318361092,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.96804664116802e-06,
+      "loss": 17.5458,
+      "step": 11310
+    },
+    {
+      "epoch": 0.2046826582527388,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.968018388861358e-06,
+      "loss": 17.4147,
+      "step": 11320
+    },
+    {
+      "epoch": 0.20486347332186663,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.967990136554699e-06,
+      "loss": 17.158,
+      "step": 11330
+    },
+    {
+      "epoch": 0.2050442883909945,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.967961884248041e-06,
+      "loss": 17.5764,
+      "step": 11340
+    },
+    {
+      "epoch": 0.20522510346012238,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.967933631941382e-06,
+      "loss": 17.6255,
+      "step": 11350
+    },
+    {
+      "epoch": 0.20540591852925022,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.96790537963472e-06,
+      "loss": 17.2257,
+      "step": 11360
+    },
+    {
+      "epoch": 0.2055867335983781,
+      "grad_norm": 40.625,
+      "learning_rate": 9.967877127328061e-06,
+      "loss": 17.3469,
+      "step": 11370
+    },
+    {
+      "epoch": 0.20576754866750596,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.967848875021402e-06,
+      "loss": 17.1934,
+      "step": 11380
+    },
+    {
+      "epoch": 0.2059483637366338,
+      "grad_norm": 39.625,
+      "learning_rate": 9.967820622714743e-06,
+      "loss": 17.1529,
+      "step": 11390
+    },
+    {
+      "epoch": 0.20612917880576168,
+      "grad_norm": 42.0,
+      "learning_rate": 9.967792370408083e-06,
+      "loss": 17.0926,
+      "step": 11400
+    },
+    {
+      "epoch": 0.20630999387488955,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.967764118101422e-06,
+      "loss": 17.0699,
+      "step": 11410
+    },
+    {
+      "epoch": 0.2064908089440174,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.967735865794763e-06,
+      "loss": 17.3271,
+      "step": 11420
+    },
+    {
+      "epoch": 0.20667162401314526,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.967707613488105e-06,
+      "loss": 17.2057,
+      "step": 11430
+    },
+    {
+      "epoch": 0.2068524390822731,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.967679361181444e-06,
+      "loss": 17.5981,
+      "step": 11440
+    },
+    {
+      "epoch": 0.20703325415140097,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.967651108874785e-06,
+      "loss": 16.9635,
+      "step": 11450
+    },
+    {
+      "epoch": 0.20721406922052885,
+      "grad_norm": 40.375,
+      "learning_rate": 9.967622856568125e-06,
+      "loss": 17.6198,
+      "step": 11460
+    },
+    {
+      "epoch": 0.2073948842896567,
+      "grad_norm": 38.75,
+      "learning_rate": 9.967594604261466e-06,
+      "loss": 17.2423,
+      "step": 11470
+    },
+    {
+      "epoch": 0.20757569935878456,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.967566351954807e-06,
+      "loss": 17.2828,
+      "step": 11480
+    },
+    {
+      "epoch": 0.20775651442791243,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.967538099648147e-06,
+      "loss": 17.5125,
+      "step": 11490
+    },
+    {
+      "epoch": 0.20793732949704027,
+      "grad_norm": 42.625,
+      "learning_rate": 9.967509847341486e-06,
+      "loss": 17.0669,
+      "step": 11500
+    },
+    {
+      "epoch": 0.20811814456616815,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.967481595034827e-06,
+      "loss": 17.4618,
+      "step": 11510
+    },
+    {
+      "epoch": 0.20829895963529602,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.967453342728169e-06,
+      "loss": 17.1914,
+      "step": 11520
+    },
+    {
+      "epoch": 0.20847977470442386,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.967425090421508e-06,
+      "loss": 17.3149,
+      "step": 11530
+    },
+    {
+      "epoch": 0.20866058977355173,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.967396838114849e-06,
+      "loss": 17.5775,
+      "step": 11540
+    },
+    {
+      "epoch": 0.2088414048426796,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.96736858580819e-06,
+      "loss": 17.443,
+      "step": 11550
+    },
+    {
+      "epoch": 0.20902221991180744,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.96734033350153e-06,
+      "loss": 17.4273,
+      "step": 11560
+    },
+    {
+      "epoch": 0.20920303498093532,
+      "grad_norm": 41.75,
+      "learning_rate": 9.96731208119487e-06,
+      "loss": 17.0814,
+      "step": 11570
+    },
+    {
+      "epoch": 0.20938385005006319,
+      "grad_norm": 41.625,
+      "learning_rate": 9.96728382888821e-06,
+      "loss": 17.515,
+      "step": 11580
+    },
+    {
+      "epoch": 0.20956466511919103,
+      "grad_norm": 38.03125,
+      "learning_rate": 9.96725557658155e-06,
+      "loss": 17.6116,
+      "step": 11590
+    },
+    {
+      "epoch": 0.2097454801883189,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.96722732427489e-06,
+      "loss": 17.3167,
+      "step": 11600
+    },
+    {
+      "epoch": 0.20992629525744674,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.967199071968233e-06,
+      "loss": 17.2356,
+      "step": 11610
+    },
+    {
+      "epoch": 0.21010711032657461,
+      "grad_norm": 41.375,
+      "learning_rate": 9.967170819661572e-06,
+      "loss": 17.4695,
+      "step": 11620
+    },
+    {
+      "epoch": 0.21028792539570249,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.967142567354912e-06,
+      "loss": 17.5472,
+      "step": 11630
+    },
+    {
+      "epoch": 0.21046874046483033,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.967114315048253e-06,
+      "loss": 16.7093,
+      "step": 11640
+    },
+    {
+      "epoch": 0.2106495555339582,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.967086062741594e-06,
+      "loss": 17.725,
+      "step": 11650
+    },
+    {
+      "epoch": 0.21083037060308607,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.967057810434934e-06,
+      "loss": 17.4257,
+      "step": 11660
+    },
+    {
+      "epoch": 0.2110111856722139,
+      "grad_norm": 38.25,
+      "learning_rate": 9.967029558128273e-06,
+      "loss": 17.2519,
+      "step": 11670
+    },
+    {
+      "epoch": 0.21119200074134178,
+      "grad_norm": 38.46875,
+      "learning_rate": 9.967001305821614e-06,
+      "loss": 17.6406,
+      "step": 11680
+    },
+    {
+      "epoch": 0.21137281581046966,
+      "grad_norm": 38.28125,
+      "learning_rate": 9.966973053514956e-06,
+      "loss": 17.1831,
+      "step": 11690
+    },
+    {
+      "epoch": 0.2115536308795975,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.966944801208295e-06,
+      "loss": 17.2882,
+      "step": 11700
+    },
+    {
+      "epoch": 0.21173444594872537,
+      "grad_norm": 38.40625,
+      "learning_rate": 9.966916548901636e-06,
+      "loss": 17.6014,
+      "step": 11710
+    },
+    {
+      "epoch": 0.21191526101785324,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.966888296594976e-06,
+      "loss": 17.5026,
+      "step": 11720
+    },
+    {
+      "epoch": 0.21209607608698108,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.966860044288317e-06,
+      "loss": 17.4043,
+      "step": 11730
+    },
+    {
+      "epoch": 0.21227689115610895,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.966831791981658e-06,
+      "loss": 16.6975,
+      "step": 11740
+    },
+    {
+      "epoch": 0.2124577062252368,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.966803539674997e-06,
+      "loss": 17.4932,
+      "step": 11750
+    },
+    {
+      "epoch": 0.21263852129436467,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.966775287368337e-06,
+      "loss": 17.5085,
+      "step": 11760
+    },
+    {
+      "epoch": 0.21281933636349254,
+      "grad_norm": 41.0,
+      "learning_rate": 9.966747035061678e-06,
+      "loss": 17.6466,
+      "step": 11770
+    },
+    {
+      "epoch": 0.21300015143262038,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.96671878275502e-06,
+      "loss": 17.0325,
+      "step": 11780
+    },
+    {
+      "epoch": 0.21318096650174825,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.966690530448359e-06,
+      "loss": 17.2715,
+      "step": 11790
+    },
+    {
+      "epoch": 0.21336178157087612,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.9666622781417e-06,
+      "loss": 17.3877,
+      "step": 11800
+    },
+    {
+      "epoch": 0.21354259664000397,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.96663402583504e-06,
+      "loss": 17.3444,
+      "step": 11810
+    },
+    {
+      "epoch": 0.21372341170913184,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.966605773528381e-06,
+      "loss": 17.11,
+      "step": 11820
+    },
+    {
+      "epoch": 0.2139042267782597,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.966577521221722e-06,
+      "loss": 17.4604,
+      "step": 11830
+    },
+    {
+      "epoch": 0.21408504184738755,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.96654926891506e-06,
+      "loss": 17.2247,
+      "step": 11840
+    },
+    {
+      "epoch": 0.21426585691651542,
+      "grad_norm": 38.875,
+      "learning_rate": 9.966521016608401e-06,
+      "loss": 17.1911,
+      "step": 11850
+    },
+    {
+      "epoch": 0.2144466719856433,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.966492764301742e-06,
+      "loss": 17.3012,
+      "step": 11860
+    },
+    {
+      "epoch": 0.21462748705477114,
+      "grad_norm": 42.0,
+      "learning_rate": 9.966464511995082e-06,
+      "loss": 16.8401,
+      "step": 11870
+    },
+    {
+      "epoch": 0.214808302123899,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.966436259688423e-06,
+      "loss": 17.2553,
+      "step": 11880
+    },
+    {
+      "epoch": 0.21498911719302688,
+      "grad_norm": 40.625,
+      "learning_rate": 9.966408007381764e-06,
+      "loss": 17.3049,
+      "step": 11890
+    },
+    {
+      "epoch": 0.21516993226215472,
+      "grad_norm": 40.875,
+      "learning_rate": 9.966379755075104e-06,
+      "loss": 17.5202,
+      "step": 11900
+    },
+    {
+      "epoch": 0.2153507473312826,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.966351502768445e-06,
+      "loss": 17.5647,
+      "step": 11910
+    },
+    {
+      "epoch": 0.21553156240041044,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.966323250461785e-06,
+      "loss": 17.4072,
+      "step": 11920
+    },
+    {
+      "epoch": 0.2157123774695383,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.966294998155124e-06,
+      "loss": 17.3908,
+      "step": 11930
+    },
+    {
+      "epoch": 0.21589319253866618,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.966266745848465e-06,
+      "loss": 17.5294,
+      "step": 11940
+    },
+    {
+      "epoch": 0.21607400760779402,
+      "grad_norm": 37.75,
+      "learning_rate": 9.966238493541806e-06,
+      "loss": 17.1815,
+      "step": 11950
+    },
+    {
+      "epoch": 0.2162548226769219,
+      "grad_norm": 37.875,
+      "learning_rate": 9.966210241235146e-06,
+      "loss": 17.3322,
+      "step": 11960
+    },
+    {
+      "epoch": 0.21643563774604976,
+      "grad_norm": 36.75,
+      "learning_rate": 9.966181988928487e-06,
+      "loss": 16.9328,
+      "step": 11970
+    },
+    {
+      "epoch": 0.2166164528151776,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.966153736621828e-06,
+      "loss": 16.6897,
+      "step": 11980
+    },
+    {
+      "epoch": 0.21679726788430548,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.966125484315168e-06,
+      "loss": 17.0212,
+      "step": 11990
+    },
+    {
+      "epoch": 0.21697808295343335,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.966097232008509e-06,
+      "loss": 17.2172,
+      "step": 12000
+    },
+    {
+      "epoch": 0.2171588980225612,
+      "grad_norm": 42.125,
+      "learning_rate": 9.966068979701848e-06,
+      "loss": 16.9982,
+      "step": 12010
+    },
+    {
+      "epoch": 0.21733971309168906,
+      "grad_norm": 42.125,
+      "learning_rate": 9.966040727395188e-06,
+      "loss": 17.1451,
+      "step": 12020
+    },
+    {
+      "epoch": 0.21752052816081693,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.966012475088529e-06,
+      "loss": 17.435,
+      "step": 12030
+    },
+    {
+      "epoch": 0.21770134322994478,
+      "grad_norm": 38.625,
+      "learning_rate": 9.96598422278187e-06,
+      "loss": 17.0784,
+      "step": 12040
+    },
+    {
+      "epoch": 0.21788215829907265,
+      "grad_norm": 39.0,
+      "learning_rate": 9.96595597047521e-06,
+      "loss": 17.4566,
+      "step": 12050
+    },
+    {
+      "epoch": 0.21806297336820052,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.96592771816855e-06,
+      "loss": 17.1917,
+      "step": 12060
+    },
+    {
+      "epoch": 0.21824378843732836,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.965899465861891e-06,
+      "loss": 17.5761,
+      "step": 12070
+    },
+    {
+      "epoch": 0.21842460350645623,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.965871213555232e-06,
+      "loss": 17.2526,
+      "step": 12080
+    },
+    {
+      "epoch": 0.21860541857558408,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.965842961248573e-06,
+      "loss": 17.4308,
+      "step": 12090
+    },
+    {
+      "epoch": 0.21878623364471195,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.965814708941912e-06,
+      "loss": 17.687,
+      "step": 12100
+    },
+    {
+      "epoch": 0.21896704871383982,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.965786456635252e-06,
+      "loss": 17.2235,
+      "step": 12110
+    },
+    {
+      "epoch": 0.21914786378296766,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.965758204328593e-06,
+      "loss": 17.2053,
+      "step": 12120
+    },
+    {
+      "epoch": 0.21932867885209553,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.965729952021933e-06,
+      "loss": 16.9105,
+      "step": 12130
+    },
+    {
+      "epoch": 0.2195094939212234,
+      "grad_norm": 39.875,
+      "learning_rate": 9.965701699715274e-06,
+      "loss": 17.7573,
+      "step": 12140
+    },
+    {
+      "epoch": 0.21969030899035125,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.965673447408615e-06,
+      "loss": 18.0731,
+      "step": 12150
+    },
+    {
+      "epoch": 0.21987112405947912,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.965645195101955e-06,
+      "loss": 16.8835,
+      "step": 12160
+    },
+    {
+      "epoch": 0.220051939128607,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.965616942795296e-06,
+      "loss": 17.0859,
+      "step": 12170
+    },
+    {
+      "epoch": 0.22023275419773483,
+      "grad_norm": 40.25,
+      "learning_rate": 9.965588690488635e-06,
+      "loss": 17.2099,
+      "step": 12180
+    },
+    {
+      "epoch": 0.2204135692668627,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.965560438181975e-06,
+      "loss": 17.5276,
+      "step": 12190
+    },
+    {
+      "epoch": 0.22059438433599057,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.965532185875316e-06,
+      "loss": 17.3187,
+      "step": 12200
+    },
+    {
+      "epoch": 0.22077519940511842,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.965503933568657e-06,
+      "loss": 17.2944,
+      "step": 12210
+    },
+    {
+      "epoch": 0.2209560144742463,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.965475681261997e-06,
+      "loss": 17.1889,
+      "step": 12220
+    },
+    {
+      "epoch": 0.22113682954337416,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.965447428955338e-06,
+      "loss": 16.8959,
+      "step": 12230
+    },
+    {
+      "epoch": 0.221317644612502,
+      "grad_norm": 45.375,
+      "learning_rate": 9.965419176648679e-06,
+      "loss": 17.3403,
+      "step": 12240
+    },
+    {
+      "epoch": 0.22149845968162987,
+      "grad_norm": 37.5625,
+      "learning_rate": 9.96539092434202e-06,
+      "loss": 17.3305,
+      "step": 12250
+    },
+    {
+      "epoch": 0.22167927475075772,
+      "grad_norm": 45.03125,
+      "learning_rate": 9.96536267203536e-06,
+      "loss": 17.6105,
+      "step": 12260
+    },
+    {
+      "epoch": 0.2218600898198856,
+      "grad_norm": 41.875,
+      "learning_rate": 9.965334419728699e-06,
+      "loss": 17.2321,
+      "step": 12270
+    },
+    {
+      "epoch": 0.22204090488901346,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.96530616742204e-06,
+      "loss": 17.4887,
+      "step": 12280
+    },
+    {
+      "epoch": 0.2222217199581413,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.96527791511538e-06,
+      "loss": 17.6234,
+      "step": 12290
+    },
+    {
+      "epoch": 0.22240253502726917,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.96524966280872e-06,
+      "loss": 17.3018,
+      "step": 12300
+    },
+    {
+      "epoch": 0.22258335009639704,
+      "grad_norm": 38.28125,
+      "learning_rate": 9.965221410502061e-06,
+      "loss": 17.2957,
+      "step": 12310
+    },
+    {
+      "epoch": 0.22276416516552489,
+      "grad_norm": 41.625,
+      "learning_rate": 9.965193158195402e-06,
+      "loss": 17.1843,
+      "step": 12320
+    },
+    {
+      "epoch": 0.22294498023465276,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.965164905888743e-06,
+      "loss": 17.5083,
+      "step": 12330
+    },
+    {
+      "epoch": 0.22312579530378063,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.965136653582083e-06,
+      "loss": 17.018,
+      "step": 12340
+    },
+    {
+      "epoch": 0.22330661037290847,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.965108401275422e-06,
+      "loss": 17.7748,
+      "step": 12350
+    },
+    {
+      "epoch": 0.22348742544203634,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.965080148968763e-06,
+      "loss": 17.2274,
+      "step": 12360
+    },
+    {
+      "epoch": 0.2236682405111642,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.965051896662103e-06,
+      "loss": 17.2712,
+      "step": 12370
+    },
+    {
+      "epoch": 0.22384905558029206,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.965023644355444e-06,
+      "loss": 17.5226,
+      "step": 12380
+    },
+    {
+      "epoch": 0.22402987064941993,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.964995392048785e-06,
+      "loss": 17.2146,
+      "step": 12390
+    },
+    {
+      "epoch": 0.2242106857185478,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.964967139742125e-06,
+      "loss": 17.4363,
+      "step": 12400
+    },
+    {
+      "epoch": 0.22439150078767564,
+      "grad_norm": 38.40625,
+      "learning_rate": 9.964938887435466e-06,
+      "loss": 17.2412,
+      "step": 12410
+    },
+    {
+      "epoch": 0.2245723158568035,
+      "grad_norm": 37.0,
+      "learning_rate": 9.964910635128806e-06,
+      "loss": 17.4035,
+      "step": 12420
+    },
+    {
+      "epoch": 0.22475313092593135,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.964882382822147e-06,
+      "loss": 17.4648,
+      "step": 12430
+    },
+    {
+      "epoch": 0.22493394599505923,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.964854130515486e-06,
+      "loss": 17.2871,
+      "step": 12440
+    },
+    {
+      "epoch": 0.2251147610641871,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.964825878208827e-06,
+      "loss": 17.3342,
+      "step": 12450
+    },
+    {
+      "epoch": 0.22529557613331494,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.964797625902167e-06,
+      "loss": 17.5347,
+      "step": 12460
+    },
+    {
+      "epoch": 0.2254763912024428,
+      "grad_norm": 37.28125,
+      "learning_rate": 9.964769373595508e-06,
+      "loss": 17.2245,
+      "step": 12470
+    },
+    {
+      "epoch": 0.22565720627157068,
+      "grad_norm": 36.9375,
+      "learning_rate": 9.964741121288848e-06,
+      "loss": 16.9847,
+      "step": 12480
+    },
+    {
+      "epoch": 0.22583802134069852,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.964712868982189e-06,
+      "loss": 16.98,
+      "step": 12490
+    },
+    {
+      "epoch": 0.2260188364098264,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.96468461667553e-06,
+      "loss": 17.0928,
+      "step": 12500
+    },
+    {
+      "epoch": 0.22619965147895427,
+      "grad_norm": 37.59375,
+      "learning_rate": 9.96465636436887e-06,
+      "loss": 17.3024,
+      "step": 12510
+    },
+    {
+      "epoch": 0.2263804665480821,
+      "grad_norm": 41.75,
+      "learning_rate": 9.964628112062211e-06,
+      "loss": 17.6225,
+      "step": 12520
+    },
+    {
+      "epoch": 0.22656128161720998,
+      "grad_norm": 41.625,
+      "learning_rate": 9.96459985975555e-06,
+      "loss": 17.4294,
+      "step": 12530
+    },
+    {
+      "epoch": 0.22674209668633785,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.96457160744889e-06,
+      "loss": 17.0309,
+      "step": 12540
+    },
+    {
+      "epoch": 0.2269229117554657,
+      "grad_norm": 40.25,
+      "learning_rate": 9.964543355142231e-06,
+      "loss": 16.9067,
+      "step": 12550
+    },
+    {
+      "epoch": 0.22710372682459357,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.964515102835572e-06,
+      "loss": 17.5133,
+      "step": 12560
+    },
+    {
+      "epoch": 0.2272845418937214,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.964486850528912e-06,
+      "loss": 17.1679,
+      "step": 12570
+    },
+    {
+      "epoch": 0.22746535696284928,
+      "grad_norm": 40.5,
+      "learning_rate": 9.964458598222253e-06,
+      "loss": 17.345,
+      "step": 12580
+    },
+    {
+      "epoch": 0.22764617203197715,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.964430345915594e-06,
+      "loss": 17.2944,
+      "step": 12590
+    },
+    {
+      "epoch": 0.227826987101105,
+      "grad_norm": 42.375,
+      "learning_rate": 9.964402093608934e-06,
+      "loss": 16.8557,
+      "step": 12600
+    },
+    {
+      "epoch": 0.22800780217023287,
+      "grad_norm": 40.25,
+      "learning_rate": 9.964373841302273e-06,
+      "loss": 17.3471,
+      "step": 12610
+    },
+    {
+      "epoch": 0.22818861723936074,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.964345588995614e-06,
+      "loss": 16.9808,
+      "step": 12620
+    },
+    {
+      "epoch": 0.22836943230848858,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.964317336688954e-06,
+      "loss": 17.1642,
+      "step": 12630
+    },
+    {
+      "epoch": 0.22855024737761645,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.964289084382295e-06,
+      "loss": 17.1546,
+      "step": 12640
+    },
+    {
+      "epoch": 0.22873106244674432,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.964260832075636e-06,
+      "loss": 17.4697,
+      "step": 12650
+    },
+    {
+      "epoch": 0.22891187751587216,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.964232579768976e-06,
+      "loss": 17.0869,
+      "step": 12660
+    },
+    {
+      "epoch": 0.22909269258500004,
+      "grad_norm": 38.15625,
+      "learning_rate": 9.964204327462317e-06,
+      "loss": 17.1933,
+      "step": 12670
+    },
+    {
+      "epoch": 0.2292735076541279,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.964176075155658e-06,
+      "loss": 17.1562,
+      "step": 12680
+    },
+    {
+      "epoch": 0.22945432272325575,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.964147822848998e-06,
+      "loss": 17.2442,
+      "step": 12690
+    },
+    {
+      "epoch": 0.22963513779238362,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.964119570542337e-06,
+      "loss": 17.0829,
+      "step": 12700
+    },
+    {
+      "epoch": 0.2298159528615115,
+      "grad_norm": 39.75,
+      "learning_rate": 9.964091318235678e-06,
+      "loss": 17.4495,
+      "step": 12710
+    },
+    {
+      "epoch": 0.22999676793063933,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.964063065929018e-06,
+      "loss": 17.1042,
+      "step": 12720
+    },
+    {
+      "epoch": 0.2301775829997672,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.964034813622359e-06,
+      "loss": 17.2907,
+      "step": 12730
+    },
+    {
+      "epoch": 0.23035839806889505,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.9640065613157e-06,
+      "loss": 17.3119,
+      "step": 12740
+    },
+    {
+      "epoch": 0.23053921313802292,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.96397830900904e-06,
+      "loss": 17.4088,
+      "step": 12750
+    },
+    {
+      "epoch": 0.2307200282071508,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.96395005670238e-06,
+      "loss": 17.0552,
+      "step": 12760
+    },
+    {
+      "epoch": 0.23090084327627863,
+      "grad_norm": 37.9375,
+      "learning_rate": 9.963921804395721e-06,
+      "loss": 17.3551,
+      "step": 12770
+    },
+    {
+      "epoch": 0.2310816583454065,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.96389355208906e-06,
+      "loss": 17.0639,
+      "step": 12780
+    },
+    {
+      "epoch": 0.23126247341453438,
+      "grad_norm": 37.71875,
+      "learning_rate": 9.963865299782401e-06,
+      "loss": 17.2184,
+      "step": 12790
+    },
+    {
+      "epoch": 0.23144328848366222,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.963837047475742e-06,
+      "loss": 17.7574,
+      "step": 12800
+    },
+    {
+      "epoch": 0.2316241035527901,
+      "grad_norm": 37.375,
+      "learning_rate": 9.963808795169082e-06,
+      "loss": 17.421,
+      "step": 12810
+    },
+    {
+      "epoch": 0.23180491862191796,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.963780542862423e-06,
+      "loss": 17.1152,
+      "step": 12820
+    },
+    {
+      "epoch": 0.2319857336910458,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.963752290555763e-06,
+      "loss": 17.2219,
+      "step": 12830
+    },
+    {
+      "epoch": 0.23216654876017367,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.963724038249104e-06,
+      "loss": 17.1605,
+      "step": 12840
+    },
+    {
+      "epoch": 0.23234736382930155,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.963695785942445e-06,
+      "loss": 17.1662,
+      "step": 12850
+    },
+    {
+      "epoch": 0.2325281788984294,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.963667533635785e-06,
+      "loss": 17.369,
+      "step": 12860
+    },
+    {
+      "epoch": 0.23270899396755726,
+      "grad_norm": 43.375,
+      "learning_rate": 9.963639281329124e-06,
+      "loss": 17.798,
+      "step": 12870
+    },
+    {
+      "epoch": 0.23288980903668513,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.963611029022465e-06,
+      "loss": 17.1347,
+      "step": 12880
+    },
+    {
+      "epoch": 0.23307062410581297,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.963582776715805e-06,
+      "loss": 17.398,
+      "step": 12890
+    },
+    {
+      "epoch": 0.23325143917494084,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.963554524409146e-06,
+      "loss": 17.3474,
+      "step": 12900
+    },
+    {
+      "epoch": 0.2334322542440687,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.963526272102487e-06,
+      "loss": 17.1073,
+      "step": 12910
+    },
+    {
+      "epoch": 0.23361306931319656,
+      "grad_norm": 41.625,
+      "learning_rate": 9.963498019795827e-06,
+      "loss": 17.2598,
+      "step": 12920
+    },
+    {
+      "epoch": 0.23379388438232443,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.963469767489168e-06,
+      "loss": 17.114,
+      "step": 12930
+    },
+    {
+      "epoch": 0.23397469945145227,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.963441515182509e-06,
+      "loss": 17.214,
+      "step": 12940
+    },
+    {
+      "epoch": 0.23415551452058014,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.96341326287585e-06,
+      "loss": 17.0908,
+      "step": 12950
+    },
+    {
+      "epoch": 0.23433632958970801,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.963385010569188e-06,
+      "loss": 16.9238,
+      "step": 12960
+    },
+    {
+      "epoch": 0.23451714465883586,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.963356758262529e-06,
+      "loss": 17.0022,
+      "step": 12970
+    },
+    {
+      "epoch": 0.23469795972796373,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.96332850595587e-06,
+      "loss": 17.0961,
+      "step": 12980
+    },
+    {
+      "epoch": 0.2348787747970916,
+      "grad_norm": 38.53125,
+      "learning_rate": 9.96330025364921e-06,
+      "loss": 17.5029,
+      "step": 12990
+    },
+    {
+      "epoch": 0.23505958986621944,
+      "grad_norm": 38.3125,
+      "learning_rate": 9.96327200134255e-06,
+      "loss": 17.3472,
+      "step": 13000
+    },
+    {
+      "epoch": 0.2352404049353473,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.963243749035891e-06,
+      "loss": 17.238,
+      "step": 13010
+    },
+    {
+      "epoch": 0.23542122000447518,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.963215496729232e-06,
+      "loss": 17.2826,
+      "step": 13020
+    },
+    {
+      "epoch": 0.23560203507360303,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.963187244422573e-06,
+      "loss": 16.9839,
+      "step": 13030
+    },
+    {
+      "epoch": 0.2357828501427309,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.963158992115911e-06,
+      "loss": 17.0687,
+      "step": 13040
+    },
+    {
+      "epoch": 0.23596366521185877,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.963130739809252e-06,
+      "loss": 17.3794,
+      "step": 13050
+    },
+    {
+      "epoch": 0.2361444802809866,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.963102487502593e-06,
+      "loss": 17.0712,
+      "step": 13060
+    },
+    {
+      "epoch": 0.23632529535011448,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.963074235195933e-06,
+      "loss": 16.9269,
+      "step": 13070
+    },
+    {
+      "epoch": 0.23650611041924233,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.963045982889274e-06,
+      "loss": 17.0576,
+      "step": 13080
+    },
+    {
+      "epoch": 0.2366869254883702,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.963017730582613e-06,
+      "loss": 16.9084,
+      "step": 13090
+    },
+    {
+      "epoch": 0.23686774055749807,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.962989478275955e-06,
+      "loss": 17.7958,
+      "step": 13100
+    },
+    {
+      "epoch": 0.2370485556266259,
+      "grad_norm": 40.125,
+      "learning_rate": 9.962961225969296e-06,
+      "loss": 17.2817,
+      "step": 13110
+    },
+    {
+      "epoch": 0.23722937069575378,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.962932973662636e-06,
+      "loss": 17.777,
+      "step": 13120
+    },
+    {
+      "epoch": 0.23741018576488165,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.962904721355975e-06,
+      "loss": 17.2679,
+      "step": 13130
+    },
+    {
+      "epoch": 0.2375910008340095,
+      "grad_norm": 42.375,
+      "learning_rate": 9.962876469049316e-06,
+      "loss": 17.0529,
+      "step": 13140
+    },
+    {
+      "epoch": 0.23777181590313737,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.962848216742657e-06,
+      "loss": 17.0093,
+      "step": 13150
+    },
+    {
+      "epoch": 0.23795263097226524,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.962819964435997e-06,
+      "loss": 17.5174,
+      "step": 13160
+    },
+    {
+      "epoch": 0.23813344604139308,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.962791712129338e-06,
+      "loss": 17.5967,
+      "step": 13170
+    },
+    {
+      "epoch": 0.23831426111052095,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.962763459822678e-06,
+      "loss": 17.1299,
+      "step": 13180
+    },
+    {
+      "epoch": 0.23849507617964882,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.962735207516019e-06,
+      "loss": 17.0755,
+      "step": 13190
+    },
+    {
+      "epoch": 0.23867589124877667,
+      "grad_norm": 39.375,
+      "learning_rate": 9.96270695520936e-06,
+      "loss": 17.0905,
+      "step": 13200
+    },
+    {
+      "epoch": 0.23885670631790454,
+      "grad_norm": 41.25,
+      "learning_rate": 9.962678702902699e-06,
+      "loss": 17.5971,
+      "step": 13210
+    },
+    {
+      "epoch": 0.2390375213870324,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.96265045059604e-06,
+      "loss": 17.156,
+      "step": 13220
+    },
+    {
+      "epoch": 0.23921833645616025,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.96262219828938e-06,
+      "loss": 17.2076,
+      "step": 13230
+    },
+    {
+      "epoch": 0.23939915152528812,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.96259394598272e-06,
+      "loss": 17.1039,
+      "step": 13240
+    },
+    {
+      "epoch": 0.23957996659441597,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.962565693676061e-06,
+      "loss": 17.3425,
+      "step": 13250
+    },
+    {
+      "epoch": 0.23976078166354384,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.962537441369402e-06,
+      "loss": 17.5018,
+      "step": 13260
+    },
+    {
+      "epoch": 0.2399415967326717,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.962509189062742e-06,
+      "loss": 17.3076,
+      "step": 13270
+    },
+    {
+      "epoch": 0.24012241180179955,
+      "grad_norm": 38.34375,
+      "learning_rate": 9.962480936756083e-06,
+      "loss": 17.1876,
+      "step": 13280
+    },
+    {
+      "epoch": 0.24030322687092742,
+      "grad_norm": 36.90625,
+      "learning_rate": 9.962452684449424e-06,
+      "loss": 16.7979,
+      "step": 13290
+    },
+    {
+      "epoch": 0.2404840419400553,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.962424432142763e-06,
+      "loss": 17.3982,
+      "step": 13300
+    },
+    {
+      "epoch": 0.24066485700918314,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.962396179836103e-06,
+      "loss": 17.331,
+      "step": 13310
+    },
+    {
+      "epoch": 0.240845672078311,
+      "grad_norm": 39.5,
+      "learning_rate": 9.962367927529444e-06,
+      "loss": 16.9671,
+      "step": 13320
+    },
+    {
+      "epoch": 0.24102648714743888,
+      "grad_norm": 39.0,
+      "learning_rate": 9.962339675222784e-06,
+      "loss": 17.5099,
+      "step": 13330
+    },
+    {
+      "epoch": 0.24120730221656672,
+      "grad_norm": 41.0,
+      "learning_rate": 9.962311422916125e-06,
+      "loss": 17.1498,
+      "step": 13340
+    },
+    {
+      "epoch": 0.2413881172856946,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.962283170609464e-06,
+      "loss": 17.5852,
+      "step": 13350
+    },
+    {
+      "epoch": 0.24156893235482246,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.962254918302806e-06,
+      "loss": 17.0536,
+      "step": 13360
+    },
+    {
+      "epoch": 0.2417497474239503,
+      "grad_norm": 40.875,
+      "learning_rate": 9.962226665996147e-06,
+      "loss": 17.2196,
+      "step": 13370
+    },
+    {
+      "epoch": 0.24193056249307818,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.962198413689488e-06,
+      "loss": 17.5202,
+      "step": 13380
+    },
+    {
+      "epoch": 0.24211137756220602,
+      "grad_norm": 41.625,
+      "learning_rate": 9.962170161382826e-06,
+      "loss": 17.5732,
+      "step": 13390
+    },
+    {
+      "epoch": 0.2422921926313339,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.962141909076167e-06,
+      "loss": 17.365,
+      "step": 13400
+    },
+    {
+      "epoch": 0.24247300770046176,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.962113656769508e-06,
+      "loss": 17.4699,
+      "step": 13410
+    },
+    {
+      "epoch": 0.2426538227695896,
+      "grad_norm": 37.46875,
+      "learning_rate": 9.962085404462848e-06,
+      "loss": 17.4083,
+      "step": 13420
+    },
+    {
+      "epoch": 0.24283463783871748,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.962057152156189e-06,
+      "loss": 16.9631,
+      "step": 13430
+    },
+    {
+      "epoch": 0.24301545290784535,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.962028899849528e-06,
+      "loss": 17.1987,
+      "step": 13440
+    },
+    {
+      "epoch": 0.2431962679769732,
+      "grad_norm": 42.125,
+      "learning_rate": 9.96200064754287e-06,
+      "loss": 17.1089,
+      "step": 13450
+    },
+    {
+      "epoch": 0.24337708304610106,
+      "grad_norm": 39.875,
+      "learning_rate": 9.96197239523621e-06,
+      "loss": 17.1529,
+      "step": 13460
+    },
+    {
+      "epoch": 0.24355789811522893,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.96194414292955e-06,
+      "loss": 17.3545,
+      "step": 13470
+    },
+    {
+      "epoch": 0.24373871318435678,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.96191589062289e-06,
+      "loss": 17.4552,
+      "step": 13480
+    },
+    {
+      "epoch": 0.24391952825348465,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.961887638316231e-06,
+      "loss": 17.1754,
+      "step": 13490
+    },
+    {
+      "epoch": 0.24410034332261252,
+      "grad_norm": 37.53125,
+      "learning_rate": 9.961859386009572e-06,
+      "loss": 16.8383,
+      "step": 13500
+    },
+    {
+      "epoch": 0.24428115839174036,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.961831133702912e-06,
+      "loss": 17.8071,
+      "step": 13510
+    },
+    {
+      "epoch": 0.24446197346086823,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.961802881396251e-06,
+      "loss": 17.6229,
+      "step": 13520
+    },
+    {
+      "epoch": 0.2446427885299961,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.961774629089593e-06,
+      "loss": 17.1821,
+      "step": 13530
+    },
+    {
+      "epoch": 0.24482360359912395,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.961746376782934e-06,
+      "loss": 17.358,
+      "step": 13540
+    },
+    {
+      "epoch": 0.24500441866825182,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.961718124476275e-06,
+      "loss": 17.3703,
+      "step": 13550
+    },
+    {
+      "epoch": 0.24518523373737966,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.961689872169614e-06,
+      "loss": 17.2847,
+      "step": 13560
+    },
+    {
+      "epoch": 0.24536604880650753,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.961661619862954e-06,
+      "loss": 16.8204,
+      "step": 13570
+    },
+    {
+      "epoch": 0.2455468638756354,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.961633367556295e-06,
+      "loss": 17.361,
+      "step": 13580
+    },
+    {
+      "epoch": 0.24572767894476324,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.961605115249636e-06,
+      "loss": 17.4287,
+      "step": 13590
+    },
+    {
+      "epoch": 0.24590849401389112,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.961576862942976e-06,
+      "loss": 17.5482,
+      "step": 13600
+    },
+    {
+      "epoch": 0.246089309083019,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.961548610636315e-06,
+      "loss": 17.2304,
+      "step": 13610
+    },
+    {
+      "epoch": 0.24627012415214683,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.961520358329657e-06,
+      "loss": 17.3173,
+      "step": 13620
+    },
+    {
+      "epoch": 0.2464509392212747,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.961492106022998e-06,
+      "loss": 17.6542,
+      "step": 13630
+    },
+    {
+      "epoch": 0.24663175429040257,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.961463853716337e-06,
+      "loss": 17.1626,
+      "step": 13640
+    },
+    {
+      "epoch": 0.24681256935953041,
+      "grad_norm": 40.5,
+      "learning_rate": 9.961435601409678e-06,
+      "loss": 17.2235,
+      "step": 13650
+    },
+    {
+      "epoch": 0.24699338442865829,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.961407349103018e-06,
+      "loss": 17.0925,
+      "step": 13660
+    },
+    {
+      "epoch": 0.24717419949778616,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.961379096796359e-06,
+      "loss": 16.9831,
+      "step": 13670
+    },
+    {
+      "epoch": 0.247355014566914,
+      "grad_norm": 36.78125,
+      "learning_rate": 9.9613508444897e-06,
+      "loss": 16.8113,
+      "step": 13680
+    },
+    {
+      "epoch": 0.24753582963604187,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.96132259218304e-06,
+      "loss": 17.2586,
+      "step": 13690
+    },
+    {
+      "epoch": 0.24771664470516974,
+      "grad_norm": 39.625,
+      "learning_rate": 9.961294339876379e-06,
+      "loss": 17.7591,
+      "step": 13700
+    },
+    {
+      "epoch": 0.24789745977429758,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.961266087569721e-06,
+      "loss": 17.3166,
+      "step": 13710
+    },
+    {
+      "epoch": 0.24807827484342546,
+      "grad_norm": 39.375,
+      "learning_rate": 9.961237835263062e-06,
+      "loss": 17.3201,
+      "step": 13720
+    },
+    {
+      "epoch": 0.2482590899125533,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.9612095829564e-06,
+      "loss": 16.998,
+      "step": 13730
+    },
+    {
+      "epoch": 0.24843990498168117,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.961181330649741e-06,
+      "loss": 17.3526,
+      "step": 13740
+    },
+    {
+      "epoch": 0.24862072005080904,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.961153078343082e-06,
+      "loss": 17.0846,
+      "step": 13750
+    },
+    {
+      "epoch": 0.24880153511993688,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.961124826036423e-06,
+      "loss": 17.3245,
+      "step": 13760
+    },
+    {
+      "epoch": 0.24898235018906476,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.961096573729763e-06,
+      "loss": 17.525,
+      "step": 13770
+    },
+    {
+      "epoch": 0.24916316525819263,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.961068321423102e-06,
+      "loss": 16.7753,
+      "step": 13780
+    },
+    {
+      "epoch": 0.24934398032732047,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.961040069116443e-06,
+      "loss": 17.5064,
+      "step": 13790
+    },
+    {
+      "epoch": 0.24952479539644834,
+      "grad_norm": 41.0,
+      "learning_rate": 9.961011816809785e-06,
+      "loss": 16.8966,
+      "step": 13800
+    },
+    {
+      "epoch": 0.2497056104655762,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.960983564503126e-06,
+      "loss": 17.5192,
+      "step": 13810
+    },
+    {
+      "epoch": 0.24988642553470405,
+      "grad_norm": 41.875,
+      "learning_rate": 9.960955312196465e-06,
+      "loss": 17.1112,
+      "step": 13820
+    },
+    {
+      "epoch": 0.2500672406038319,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.960927059889805e-06,
+      "loss": 17.2937,
+      "step": 13830
+    },
+    {
+      "epoch": 0.25024805567295977,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.960898807583146e-06,
+      "loss": 17.4666,
+      "step": 13840
+    },
+    {
+      "epoch": 0.25042887074208764,
+      "grad_norm": 38.875,
+      "learning_rate": 9.960870555276487e-06,
+      "loss": 17.4754,
+      "step": 13850
+    },
+    {
+      "epoch": 0.2506096858112155,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.960842302969827e-06,
+      "loss": 17.234,
+      "step": 13860
+    },
+    {
+      "epoch": 0.2507905008803434,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.960814050663166e-06,
+      "loss": 17.5303,
+      "step": 13870
+    },
+    {
+      "epoch": 0.25097131594947125,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.960785798356508e-06,
+      "loss": 17.6465,
+      "step": 13880
+    },
+    {
+      "epoch": 0.25115213101859907,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.960757546049849e-06,
+      "loss": 17.1546,
+      "step": 13890
+    },
+    {
+      "epoch": 0.25133294608772694,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.960729293743188e-06,
+      "loss": 17.2995,
+      "step": 13900
+    },
+    {
+      "epoch": 0.2515137611568548,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.960701041436529e-06,
+      "loss": 17.6509,
+      "step": 13910
+    },
+    {
+      "epoch": 0.2516945762259827,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.96067278912987e-06,
+      "loss": 17.4684,
+      "step": 13920
+    },
+    {
+      "epoch": 0.25187539129511055,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.96064453682321e-06,
+      "loss": 17.4557,
+      "step": 13930
+    },
+    {
+      "epoch": 0.2520562063642384,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.96061628451655e-06,
+      "loss": 17.2042,
+      "step": 13940
+    },
+    {
+      "epoch": 0.25223702143336624,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.96058803220989e-06,
+      "loss": 17.3159,
+      "step": 13950
+    },
+    {
+      "epoch": 0.2524178365024941,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.96055977990323e-06,
+      "loss": 17.0903,
+      "step": 13960
+    },
+    {
+      "epoch": 0.252598651571622,
+      "grad_norm": 38.28125,
+      "learning_rate": 9.960531527596572e-06,
+      "loss": 17.1521,
+      "step": 13970
+    },
+    {
+      "epoch": 0.25277946664074985,
+      "grad_norm": 40.875,
+      "learning_rate": 9.960503275289913e-06,
+      "loss": 16.9971,
+      "step": 13980
+    },
+    {
+      "epoch": 0.2529602817098777,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.960475022983252e-06,
+      "loss": 17.4463,
+      "step": 13990
+    },
+    {
+      "epoch": 0.25314109677900554,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.960446770676593e-06,
+      "loss": 17.1802,
+      "step": 14000
+    },
+    {
+      "epoch": 0.2533219118481334,
+      "grad_norm": 41.75,
+      "learning_rate": 9.960418518369933e-06,
+      "loss": 17.2515,
+      "step": 14010
+    },
+    {
+      "epoch": 0.2535027269172613,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.960390266063274e-06,
+      "loss": 17.5549,
+      "step": 14020
+    },
+    {
+      "epoch": 0.25368354198638915,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.960362013756614e-06,
+      "loss": 17.5882,
+      "step": 14030
+    },
+    {
+      "epoch": 0.253864357055517,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.960333761449953e-06,
+      "loss": 17.462,
+      "step": 14040
+    },
+    {
+      "epoch": 0.2540451721246449,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.960305509143294e-06,
+      "loss": 17.2972,
+      "step": 14050
+    },
+    {
+      "epoch": 0.2542259871937727,
+      "grad_norm": 41.875,
+      "learning_rate": 9.960277256836636e-06,
+      "loss": 17.5263,
+      "step": 14060
+    },
+    {
+      "epoch": 0.2544068022629006,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.960249004529975e-06,
+      "loss": 17.6281,
+      "step": 14070
+    },
+    {
+      "epoch": 0.25458761733202845,
+      "grad_norm": 37.40625,
+      "learning_rate": 9.960220752223316e-06,
+      "loss": 17.341,
+      "step": 14080
+    },
+    {
+      "epoch": 0.2547684324011563,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.960192499916656e-06,
+      "loss": 17.308,
+      "step": 14090
+    },
+    {
+      "epoch": 0.2549492474702842,
+      "grad_norm": 37.71875,
+      "learning_rate": 9.960164247609997e-06,
+      "loss": 17.3255,
+      "step": 14100
+    },
+    {
+      "epoch": 0.25513006253941206,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.960135995303338e-06,
+      "loss": 17.5287,
+      "step": 14110
+    },
+    {
+      "epoch": 0.2553108776085399,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.960107742996678e-06,
+      "loss": 17.3311,
+      "step": 14120
+    },
+    {
+      "epoch": 0.25549169267766775,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.960079490690017e-06,
+      "loss": 16.9866,
+      "step": 14130
+    },
+    {
+      "epoch": 0.2556725077467956,
+      "grad_norm": 41.25,
+      "learning_rate": 9.960051238383358e-06,
+      "loss": 17.3274,
+      "step": 14140
+    },
+    {
+      "epoch": 0.2558533228159235,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.9600229860767e-06,
+      "loss": 17.3333,
+      "step": 14150
+    },
+    {
+      "epoch": 0.25603413788505136,
+      "grad_norm": 37.75,
+      "learning_rate": 9.959994733770039e-06,
+      "loss": 17.3042,
+      "step": 14160
+    },
+    {
+      "epoch": 0.2562149529541792,
+      "grad_norm": 40.875,
+      "learning_rate": 9.95996648146338e-06,
+      "loss": 17.2533,
+      "step": 14170
+    },
+    {
+      "epoch": 0.25639576802330705,
+      "grad_norm": 36.25,
+      "learning_rate": 9.95993822915672e-06,
+      "loss": 17.3347,
+      "step": 14180
+    },
+    {
+      "epoch": 0.2565765830924349,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.959909976850061e-06,
+      "loss": 17.7013,
+      "step": 14190
+    },
+    {
+      "epoch": 0.2567573981615628,
+      "grad_norm": 41.25,
+      "learning_rate": 9.959881724543402e-06,
+      "loss": 17.4491,
+      "step": 14200
+    },
+    {
+      "epoch": 0.25693821323069066,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.95985347223674e-06,
+      "loss": 16.9943,
+      "step": 14210
+    },
+    {
+      "epoch": 0.25711902829981853,
+      "grad_norm": 39.25,
+      "learning_rate": 9.959825219930081e-06,
+      "loss": 17.5549,
+      "step": 14220
+    },
+    {
+      "epoch": 0.25729984336894635,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.959796967623423e-06,
+      "loss": 17.2006,
+      "step": 14230
+    },
+    {
+      "epoch": 0.2574806584380742,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.959768715316764e-06,
+      "loss": 17.0657,
+      "step": 14240
+    },
+    {
+      "epoch": 0.2576614735072021,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.959740463010103e-06,
+      "loss": 16.9687,
+      "step": 14250
+    },
+    {
+      "epoch": 0.25784228857632996,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.959712210703444e-06,
+      "loss": 17.2125,
+      "step": 14260
+    },
+    {
+      "epoch": 0.25802310364545783,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.959683958396784e-06,
+      "loss": 17.1497,
+      "step": 14270
+    },
+    {
+      "epoch": 0.2582039187145857,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.959655706090125e-06,
+      "loss": 16.7245,
+      "step": 14280
+    },
+    {
+      "epoch": 0.2583847337837135,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.959627453783466e-06,
+      "loss": 17.2191,
+      "step": 14290
+    },
+    {
+      "epoch": 0.2585655488528414,
+      "grad_norm": 38.09375,
+      "learning_rate": 9.959599201476804e-06,
+      "loss": 17.0521,
+      "step": 14300
+    },
+    {
+      "epoch": 0.25874636392196926,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.959570949170145e-06,
+      "loss": 17.1253,
+      "step": 14310
+    },
+    {
+      "epoch": 0.25892717899109713,
+      "grad_norm": 39.125,
+      "learning_rate": 9.959542696863487e-06,
+      "loss": 16.9786,
+      "step": 14320
+    },
+    {
+      "epoch": 0.259107994060225,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.959514444556826e-06,
+      "loss": 17.5312,
+      "step": 14330
+    },
+    {
+      "epoch": 0.2592888091293528,
+      "grad_norm": 36.75,
+      "learning_rate": 9.959486192250167e-06,
+      "loss": 17.2473,
+      "step": 14340
+    },
+    {
+      "epoch": 0.2594696241984807,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.959457939943508e-06,
+      "loss": 16.8098,
+      "step": 14350
+    },
+    {
+      "epoch": 0.25965043926760856,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.959429687636848e-06,
+      "loss": 17.3783,
+      "step": 14360
+    },
+    {
+      "epoch": 0.25983125433673643,
+      "grad_norm": 39.625,
+      "learning_rate": 9.959401435330189e-06,
+      "loss": 17.593,
+      "step": 14370
+    },
+    {
+      "epoch": 0.2600120694058643,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.959373183023528e-06,
+      "loss": 17.406,
+      "step": 14380
+    },
+    {
+      "epoch": 0.26019288447499217,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.959344930716868e-06,
+      "loss": 17.3751,
+      "step": 14390
+    },
+    {
+      "epoch": 0.26037369954412,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.959316678410209e-06,
+      "loss": 17.2173,
+      "step": 14400
+    },
+    {
+      "epoch": 0.26055451461324786,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.959288426103551e-06,
+      "loss": 17.1151,
+      "step": 14410
+    },
+    {
+      "epoch": 0.2607353296823757,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.95926017379689e-06,
+      "loss": 17.3042,
+      "step": 14420
+    },
+    {
+      "epoch": 0.2609161447515036,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.95923192149023e-06,
+      "loss": 17.3013,
+      "step": 14430
+    },
+    {
+      "epoch": 0.26109695982063147,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.959203669183571e-06,
+      "loss": 17.1391,
+      "step": 14440
+    },
+    {
+      "epoch": 0.26127777488975934,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.959175416876912e-06,
+      "loss": 17.412,
+      "step": 14450
+    },
+    {
+      "epoch": 0.26145858995888716,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.959147164570253e-06,
+      "loss": 17.0796,
+      "step": 14460
+    },
+    {
+      "epoch": 0.261639405028015,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.959118912263592e-06,
+      "loss": 17.1053,
+      "step": 14470
+    },
+    {
+      "epoch": 0.2618202200971429,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.959090659956932e-06,
+      "loss": 17.2328,
+      "step": 14480
+    },
+    {
+      "epoch": 0.26200103516627077,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.959062407650273e-06,
+      "loss": 17.2747,
+      "step": 14490
+    },
+    {
+      "epoch": 0.26218185023539864,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.959034155343614e-06,
+      "loss": 17.1025,
+      "step": 14500
+    },
+    {
+      "epoch": 0.26236266530452645,
+      "grad_norm": 40.25,
+      "learning_rate": 9.959005903036954e-06,
+      "loss": 17.8269,
+      "step": 14510
+    },
+    {
+      "epoch": 0.2625434803736543,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.958977650730295e-06,
+      "loss": 17.1326,
+      "step": 14520
+    },
+    {
+      "epoch": 0.2627242954427822,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.958949398423635e-06,
+      "loss": 17.4717,
+      "step": 14530
+    },
+    {
+      "epoch": 0.26290511051191007,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.958921146116976e-06,
+      "loss": 16.7587,
+      "step": 14540
+    },
+    {
+      "epoch": 0.26308592558103794,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.958892893810315e-06,
+      "loss": 17.5066,
+      "step": 14550
+    },
+    {
+      "epoch": 0.2632667406501658,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.958864641503656e-06,
+      "loss": 17.0883,
+      "step": 14560
+    },
+    {
+      "epoch": 0.2634475557192936,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.958836389196996e-06,
+      "loss": 17.5933,
+      "step": 14570
+    },
+    {
+      "epoch": 0.2636283707884215,
+      "grad_norm": 42.0,
+      "learning_rate": 9.958808136890338e-06,
+      "loss": 17.4965,
+      "step": 14580
+    },
+    {
+      "epoch": 0.26380918585754937,
+      "grad_norm": 38.625,
+      "learning_rate": 9.958779884583677e-06,
+      "loss": 17.28,
+      "step": 14590
+    },
+    {
+      "epoch": 0.26399000092667724,
+      "grad_norm": 40.5,
+      "learning_rate": 9.958751632277018e-06,
+      "loss": 17.4132,
+      "step": 14600
+    },
+    {
+      "epoch": 0.2641708159958051,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.958723379970359e-06,
+      "loss": 17.2562,
+      "step": 14610
+    },
+    {
+      "epoch": 0.264351631064933,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.9586951276637e-06,
+      "loss": 17.5402,
+      "step": 14620
+    },
+    {
+      "epoch": 0.2645324461340608,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.95866687535704e-06,
+      "loss": 17.2135,
+      "step": 14630
+    },
+    {
+      "epoch": 0.26471326120318867,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.958638623050379e-06,
+      "loss": 17.0918,
+      "step": 14640
+    },
+    {
+      "epoch": 0.26489407627231654,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.95861037074372e-06,
+      "loss": 17.3941,
+      "step": 14650
+    },
+    {
+      "epoch": 0.2650748913414444,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.95858211843706e-06,
+      "loss": 17.3198,
+      "step": 14660
+    },
+    {
+      "epoch": 0.2652557064105723,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.9585538661304e-06,
+      "loss": 16.8643,
+      "step": 14670
+    },
+    {
+      "epoch": 0.2654365214797001,
+      "grad_norm": 41.0,
+      "learning_rate": 9.958525613823741e-06,
+      "loss": 17.2676,
+      "step": 14680
+    },
+    {
+      "epoch": 0.26561733654882796,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.958497361517082e-06,
+      "loss": 17.462,
+      "step": 14690
+    },
+    {
+      "epoch": 0.26579815161795584,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.958469109210423e-06,
+      "loss": 17.0876,
+      "step": 14700
+    },
+    {
+      "epoch": 0.2659789666870837,
+      "grad_norm": 37.8125,
+      "learning_rate": 9.958440856903763e-06,
+      "loss": 17.2175,
+      "step": 14710
+    },
+    {
+      "epoch": 0.2661597817562116,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.958412604597104e-06,
+      "loss": 17.4817,
+      "step": 14720
+    },
+    {
+      "epoch": 0.26634059682533945,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.958384352290443e-06,
+      "loss": 17.4135,
+      "step": 14730
+    },
+    {
+      "epoch": 0.26652141189446726,
+      "grad_norm": 40.375,
+      "learning_rate": 9.958356099983783e-06,
+      "loss": 17.1609,
+      "step": 14740
+    },
+    {
+      "epoch": 0.26670222696359513,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.958327847677124e-06,
+      "loss": 17.1004,
+      "step": 14750
+    },
+    {
+      "epoch": 0.266883042032723,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.958299595370465e-06,
+      "loss": 17.1141,
+      "step": 14760
+    },
+    {
+      "epoch": 0.2670638571018509,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.958271343063805e-06,
+      "loss": 17.2939,
+      "step": 14770
+    },
+    {
+      "epoch": 0.26724467217097875,
+      "grad_norm": 41.625,
+      "learning_rate": 9.958243090757146e-06,
+      "loss": 17.0918,
+      "step": 14780
+    },
+    {
+      "epoch": 0.26742548724010656,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.958214838450486e-06,
+      "loss": 17.1987,
+      "step": 14790
+    },
+    {
+      "epoch": 0.26760630230923443,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.958186586143827e-06,
+      "loss": 16.9342,
+      "step": 14800
+    },
+    {
+      "epoch": 0.2677871173783623,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.958158333837166e-06,
+      "loss": 17.0782,
+      "step": 14810
+    },
+    {
+      "epoch": 0.2679679324474902,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.958130081530507e-06,
+      "loss": 17.7101,
+      "step": 14820
+    },
+    {
+      "epoch": 0.26814874751661805,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.958101829223847e-06,
+      "loss": 17.6526,
+      "step": 14830
+    },
+    {
+      "epoch": 0.2683295625857459,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.958073576917188e-06,
+      "loss": 17.1204,
+      "step": 14840
+    },
+    {
+      "epoch": 0.26851037765487373,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.958045324610529e-06,
+      "loss": 17.2844,
+      "step": 14850
+    },
+    {
+      "epoch": 0.2686911927240016,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.958017072303869e-06,
+      "loss": 16.8951,
+      "step": 14860
+    },
+    {
+      "epoch": 0.2688720077931295,
+      "grad_norm": 39.25,
+      "learning_rate": 9.95798881999721e-06,
+      "loss": 17.281,
+      "step": 14870
+    },
+    {
+      "epoch": 0.26905282286225735,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.95796056769055e-06,
+      "loss": 17.4636,
+      "step": 14880
+    },
+    {
+      "epoch": 0.2692336379313852,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.957932315383891e-06,
+      "loss": 17.1833,
+      "step": 14890
+    },
+    {
+      "epoch": 0.2694144530005131,
+      "grad_norm": 41.375,
+      "learning_rate": 9.95790406307723e-06,
+      "loss": 17.3453,
+      "step": 14900
+    },
+    {
+      "epoch": 0.2695952680696409,
+      "grad_norm": 42.125,
+      "learning_rate": 9.95787581077057e-06,
+      "loss": 17.3075,
+      "step": 14910
+    },
+    {
+      "epoch": 0.2697760831387688,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.957847558463911e-06,
+      "loss": 17.2296,
+      "step": 14920
+    },
+    {
+      "epoch": 0.26995689820789665,
+      "grad_norm": 37.8125,
+      "learning_rate": 9.957819306157252e-06,
+      "loss": 17.2295,
+      "step": 14930
+    },
+    {
+      "epoch": 0.2701377132770245,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.957791053850592e-06,
+      "loss": 17.1168,
+      "step": 14940
+    },
+    {
+      "epoch": 0.2703185283461524,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.957762801543933e-06,
+      "loss": 16.8162,
+      "step": 14950
+    },
+    {
+      "epoch": 0.2704993434152802,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.957734549237274e-06,
+      "loss": 17.2402,
+      "step": 14960
+    },
+    {
+      "epoch": 0.2706801584844081,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.957706296930614e-06,
+      "loss": 17.6101,
+      "step": 14970
+    },
+    {
+      "epoch": 0.27086097355353594,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.957678044623953e-06,
+      "loss": 17.2303,
+      "step": 14980
+    },
+    {
+      "epoch": 0.2710417886226638,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.957649792317294e-06,
+      "loss": 17.1591,
+      "step": 14990
+    },
+    {
+      "epoch": 0.2712226036917917,
+      "grad_norm": 39.25,
+      "learning_rate": 9.957621540010634e-06,
+      "loss": 17.2613,
+      "step": 15000
+    },
+    {
+      "epoch": 0.2712226036917917,
+      "eval_loss": 2.1613612174987793,
+      "eval_runtime": 229.1402,
+      "eval_samples_per_second": 3168.623,
+      "eval_steps_per_second": 49.511,
+      "step": 15000
+    },
+    {
+      "epoch": 0.27140341876091956,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.957593287703975e-06,
+      "loss": 17.1637,
+      "step": 15010
+    },
+    {
+      "epoch": 0.2715842338300474,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.957565035397316e-06,
+      "loss": 17.2246,
+      "step": 15020
+    },
+    {
+      "epoch": 0.27176504889917524,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.957536783090656e-06,
+      "loss": 17.1489,
+      "step": 15030
+    },
+    {
+      "epoch": 0.2719458639683031,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.957508530783997e-06,
+      "loss": 17.6059,
+      "step": 15040
+    },
+    {
+      "epoch": 0.272126679037431,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.957480278477338e-06,
+      "loss": 17.1587,
+      "step": 15050
+    },
+    {
+      "epoch": 0.27230749410655886,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.957452026170678e-06,
+      "loss": 17.5717,
+      "step": 15060
+    },
+    {
+      "epoch": 0.2724883091756867,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.957423773864017e-06,
+      "loss": 17.1256,
+      "step": 15070
+    },
+    {
+      "epoch": 0.27266912424481454,
+      "grad_norm": 42.25,
+      "learning_rate": 9.957395521557358e-06,
+      "loss": 17.153,
+      "step": 15080
+    },
+    {
+      "epoch": 0.2728499393139424,
+      "grad_norm": 39.5,
+      "learning_rate": 9.957367269250698e-06,
+      "loss": 16.9621,
+      "step": 15090
+    },
+    {
+      "epoch": 0.2730307543830703,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.957339016944039e-06,
+      "loss": 17.2758,
+      "step": 15100
+    },
+    {
+      "epoch": 0.27321156945219816,
+      "grad_norm": 41.5,
+      "learning_rate": 9.95731076463738e-06,
+      "loss": 17.0574,
+      "step": 15110
+    },
+    {
+      "epoch": 0.273392384521326,
+      "grad_norm": 41.625,
+      "learning_rate": 9.95728251233072e-06,
+      "loss": 17.3584,
+      "step": 15120
+    },
+    {
+      "epoch": 0.27357319959045384,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.957254260024061e-06,
+      "loss": 17.1221,
+      "step": 15130
+    },
+    {
+      "epoch": 0.2737540146595817,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.957226007717401e-06,
+      "loss": 17.4213,
+      "step": 15140
+    },
+    {
+      "epoch": 0.2739348297287096,
+      "grad_norm": 38.59375,
+      "learning_rate": 9.957197755410742e-06,
+      "loss": 17.5993,
+      "step": 15150
+    },
+    {
+      "epoch": 0.27411564479783745,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.957169503104081e-06,
+      "loss": 17.2159,
+      "step": 15160
+    },
+    {
+      "epoch": 0.2742964598669653,
+      "grad_norm": 41.25,
+      "learning_rate": 9.957141250797422e-06,
+      "loss": 17.6288,
+      "step": 15170
+    },
+    {
+      "epoch": 0.2744772749360932,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.957112998490762e-06,
+      "loss": 17.0387,
+      "step": 15180
+    },
+    {
+      "epoch": 0.274658090005221,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.957084746184103e-06,
+      "loss": 17.3209,
+      "step": 15190
+    },
+    {
+      "epoch": 0.2748389050743489,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.957056493877444e-06,
+      "loss": 16.9922,
+      "step": 15200
+    },
+    {
+      "epoch": 0.27501972014347675,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.957028241570784e-06,
+      "loss": 17.4538,
+      "step": 15210
+    },
+    {
+      "epoch": 0.2752005352126046,
+      "grad_norm": 41.875,
+      "learning_rate": 9.956999989264125e-06,
+      "loss": 17.3762,
+      "step": 15220
+    },
+    {
+      "epoch": 0.2753813502817325,
+      "grad_norm": 38.875,
+      "learning_rate": 9.956971736957465e-06,
+      "loss": 17.343,
+      "step": 15230
+    },
+    {
+      "epoch": 0.27556216535086037,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.956943484650804e-06,
+      "loss": 17.5539,
+      "step": 15240
+    },
+    {
+      "epoch": 0.2757429804199882,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.956915232344145e-06,
+      "loss": 16.9692,
+      "step": 15250
+    },
+    {
+      "epoch": 0.27592379548911605,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.956886980037486e-06,
+      "loss": 17.3419,
+      "step": 15260
+    },
+    {
+      "epoch": 0.2761046105582439,
+      "grad_norm": 42.5,
+      "learning_rate": 9.956858727730826e-06,
+      "loss": 16.8046,
+      "step": 15270
+    },
+    {
+      "epoch": 0.2762854256273718,
+      "grad_norm": 38.53125,
+      "learning_rate": 9.956830475424167e-06,
+      "loss": 17.2915,
+      "step": 15280
+    },
+    {
+      "epoch": 0.27646624069649967,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.956802223117507e-06,
+      "loss": 17.1039,
+      "step": 15290
+    },
+    {
+      "epoch": 0.2766470557656275,
+      "grad_norm": 40.0,
+      "learning_rate": 9.956773970810848e-06,
+      "loss": 17.0082,
+      "step": 15300
+    },
+    {
+      "epoch": 0.27682787083475535,
+      "grad_norm": 41.25,
+      "learning_rate": 9.956745718504189e-06,
+      "loss": 17.4862,
+      "step": 15310
+    },
+    {
+      "epoch": 0.2770086859038832,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.95671746619753e-06,
+      "loss": 17.2742,
+      "step": 15320
+    },
+    {
+      "epoch": 0.2771895009730111,
+      "grad_norm": 42.0,
+      "learning_rate": 9.956689213890868e-06,
+      "loss": 17.3996,
+      "step": 15330
+    },
+    {
+      "epoch": 0.27737031604213896,
+      "grad_norm": 41.625,
+      "learning_rate": 9.956660961584209e-06,
+      "loss": 17.3848,
+      "step": 15340
+    },
+    {
+      "epoch": 0.27755113111126684,
+      "grad_norm": 39.75,
+      "learning_rate": 9.95663270927755e-06,
+      "loss": 17.3582,
+      "step": 15350
+    },
+    {
+      "epoch": 0.27773194618039465,
+      "grad_norm": 44.0,
+      "learning_rate": 9.95660445697089e-06,
+      "loss": 17.5821,
+      "step": 15360
+    },
+    {
+      "epoch": 0.2779127612495225,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.95657620466423e-06,
+      "loss": 17.4158,
+      "step": 15370
+    },
+    {
+      "epoch": 0.2780935763186504,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.956547952357571e-06,
+      "loss": 17.2659,
+      "step": 15380
+    },
+    {
+      "epoch": 0.27827439138777826,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.956519700050912e-06,
+      "loss": 17.218,
+      "step": 15390
+    },
+    {
+      "epoch": 0.27845520645690613,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.956491447744253e-06,
+      "loss": 17.1191,
+      "step": 15400
+    },
+    {
+      "epoch": 0.278636021526034,
+      "grad_norm": 43.625,
+      "learning_rate": 9.956463195437592e-06,
+      "loss": 17.2365,
+      "step": 15410
+    },
+    {
+      "epoch": 0.2788168365951618,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.956434943130932e-06,
+      "loss": 17.5236,
+      "step": 15420
+    },
+    {
+      "epoch": 0.2789976516642897,
+      "grad_norm": 42.625,
+      "learning_rate": 9.956406690824273e-06,
+      "loss": 17.0254,
+      "step": 15430
+    },
+    {
+      "epoch": 0.27917846673341756,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.956378438517613e-06,
+      "loss": 17.2221,
+      "step": 15440
+    },
+    {
+      "epoch": 0.27935928180254543,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.956350186210954e-06,
+      "loss": 17.5745,
+      "step": 15450
+    },
+    {
+      "epoch": 0.2795400968716733,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.956321933904295e-06,
+      "loss": 16.9634,
+      "step": 15460
+    },
+    {
+      "epoch": 0.2797209119408011,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.956293681597635e-06,
+      "loss": 16.9116,
+      "step": 15470
+    },
+    {
+      "epoch": 0.279901727009929,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.956265429290976e-06,
+      "loss": 17.1836,
+      "step": 15480
+    },
+    {
+      "epoch": 0.28008254207905686,
+      "grad_norm": 40.5,
+      "learning_rate": 9.956237176984316e-06,
+      "loss": 17.068,
+      "step": 15490
+    },
+    {
+      "epoch": 0.28026335714818473,
+      "grad_norm": 41.0,
+      "learning_rate": 9.956208924677655e-06,
+      "loss": 17.2735,
+      "step": 15500
+    },
+    {
+      "epoch": 0.2804441722173126,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.956180672370996e-06,
+      "loss": 17.0922,
+      "step": 15510
+    },
+    {
+      "epoch": 0.2806249872864405,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.956152420064337e-06,
+      "loss": 16.9675,
+      "step": 15520
+    },
+    {
+      "epoch": 0.2808058023555683,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.956124167757677e-06,
+      "loss": 17.1198,
+      "step": 15530
+    },
+    {
+      "epoch": 0.28098661742469616,
+      "grad_norm": 42.5,
+      "learning_rate": 9.956095915451018e-06,
+      "loss": 16.9641,
+      "step": 15540
+    },
+    {
+      "epoch": 0.28116743249382403,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.956067663144359e-06,
+      "loss": 17.2347,
+      "step": 15550
+    },
+    {
+      "epoch": 0.2813482475629519,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.956039410837699e-06,
+      "loss": 16.6774,
+      "step": 15560
+    },
+    {
+      "epoch": 0.2815290626320798,
+      "grad_norm": 37.3125,
+      "learning_rate": 9.95601115853104e-06,
+      "loss": 17.3081,
+      "step": 15570
+    },
+    {
+      "epoch": 0.28170987770120764,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.95598290622438e-06,
+      "loss": 17.3257,
+      "step": 15580
+    },
+    {
+      "epoch": 0.28189069277033546,
+      "grad_norm": 40.5,
+      "learning_rate": 9.95595465391772e-06,
+      "loss": 16.8696,
+      "step": 15590
+    },
+    {
+      "epoch": 0.28207150783946333,
+      "grad_norm": 40.25,
+      "learning_rate": 9.95592640161106e-06,
+      "loss": 17.0273,
+      "step": 15600
+    },
+    {
+      "epoch": 0.2822523229085912,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.9558981493044e-06,
+      "loss": 17.3795,
+      "step": 15610
+    },
+    {
+      "epoch": 0.2824331379777191,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.955869896997741e-06,
+      "loss": 17.113,
+      "step": 15620
+    },
+    {
+      "epoch": 0.28261395304684694,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.955841644691082e-06,
+      "loss": 17.3427,
+      "step": 15630
+    },
+    {
+      "epoch": 0.28279476811597476,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.955813392384422e-06,
+      "loss": 17.609,
+      "step": 15640
+    },
+    {
+      "epoch": 0.28297558318510263,
+      "grad_norm": 39.5,
+      "learning_rate": 9.955785140077763e-06,
+      "loss": 17.6259,
+      "step": 15650
+    },
+    {
+      "epoch": 0.2831563982542305,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.955756887771104e-06,
+      "loss": 17.0857,
+      "step": 15660
+    },
+    {
+      "epoch": 0.2833372133233584,
+      "grad_norm": 42.125,
+      "learning_rate": 9.955728635464443e-06,
+      "loss": 17.0976,
+      "step": 15670
+    },
+    {
+      "epoch": 0.28351802839248624,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.955700383157783e-06,
+      "loss": 17.3903,
+      "step": 15680
+    },
+    {
+      "epoch": 0.2836988434616141,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.955672130851124e-06,
+      "loss": 17.3015,
+      "step": 15690
+    },
+    {
+      "epoch": 0.28387965853074193,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.955643878544464e-06,
+      "loss": 16.9424,
+      "step": 15700
+    },
+    {
+      "epoch": 0.2840604735998698,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.955615626237805e-06,
+      "loss": 17.3459,
+      "step": 15710
+    },
+    {
+      "epoch": 0.28424128866899767,
+      "grad_norm": 44.75,
+      "learning_rate": 9.955587373931146e-06,
+      "loss": 17.1487,
+      "step": 15720
+    },
+    {
+      "epoch": 0.28442210373812554,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.955559121624486e-06,
+      "loss": 17.3224,
+      "step": 15730
+    },
+    {
+      "epoch": 0.2846029188072534,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.955530869317827e-06,
+      "loss": 17.1909,
+      "step": 15740
+    },
+    {
+      "epoch": 0.2847837338763813,
+      "grad_norm": 41.0,
+      "learning_rate": 9.955502617011168e-06,
+      "loss": 16.998,
+      "step": 15750
+    },
+    {
+      "epoch": 0.2849645489455091,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.955474364704507e-06,
+      "loss": 17.4409,
+      "step": 15760
+    },
+    {
+      "epoch": 0.28514536401463697,
+      "grad_norm": 38.0625,
+      "learning_rate": 9.955446112397847e-06,
+      "loss": 17.0526,
+      "step": 15770
+    },
+    {
+      "epoch": 0.28532617908376484,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.955417860091188e-06,
+      "loss": 16.7261,
+      "step": 15780
+    },
+    {
+      "epoch": 0.2855069941528927,
+      "grad_norm": 41.375,
+      "learning_rate": 9.955389607784528e-06,
+      "loss": 17.4373,
+      "step": 15790
+    },
+    {
+      "epoch": 0.2856878092220206,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.955361355477869e-06,
+      "loss": 17.2315,
+      "step": 15800
+    },
+    {
+      "epoch": 0.2858686242911484,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.95533310317121e-06,
+      "loss": 17.264,
+      "step": 15810
+    },
+    {
+      "epoch": 0.28604943936027627,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.95530485086455e-06,
+      "loss": 17.5945,
+      "step": 15820
+    },
+    {
+      "epoch": 0.28623025442940414,
+      "grad_norm": 40.0,
+      "learning_rate": 9.955276598557891e-06,
+      "loss": 17.2339,
+      "step": 15830
+    },
+    {
+      "epoch": 0.286411069498532,
+      "grad_norm": 37.9375,
+      "learning_rate": 9.95524834625123e-06,
+      "loss": 17.3516,
+      "step": 15840
+    },
+    {
+      "epoch": 0.2865918845676599,
+      "grad_norm": 39.125,
+      "learning_rate": 9.95522009394457e-06,
+      "loss": 17.5479,
+      "step": 15850
+    },
+    {
+      "epoch": 0.28677269963678775,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.955191841637911e-06,
+      "loss": 17.3622,
+      "step": 15860
+    },
+    {
+      "epoch": 0.28695351470591557,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.955163589331252e-06,
+      "loss": 17.4275,
+      "step": 15870
+    },
+    {
+      "epoch": 0.28713432977504344,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.955135337024592e-06,
+      "loss": 17.2381,
+      "step": 15880
+    },
+    {
+      "epoch": 0.2873151448441713,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.955107084717933e-06,
+      "loss": 17.163,
+      "step": 15890
+    },
+    {
+      "epoch": 0.2874959599132992,
+      "grad_norm": 40.875,
+      "learning_rate": 9.955078832411274e-06,
+      "loss": 17.093,
+      "step": 15900
+    },
+    {
+      "epoch": 0.28767677498242705,
+      "grad_norm": 39.25,
+      "learning_rate": 9.955050580104614e-06,
+      "loss": 17.1249,
+      "step": 15910
+    },
+    {
+      "epoch": 0.2878575900515549,
+      "grad_norm": 39.625,
+      "learning_rate": 9.955022327797955e-06,
+      "loss": 17.3523,
+      "step": 15920
+    },
+    {
+      "epoch": 0.28803840512068274,
+      "grad_norm": 41.875,
+      "learning_rate": 9.954994075491294e-06,
+      "loss": 17.2994,
+      "step": 15930
+    },
+    {
+      "epoch": 0.2882192201898106,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.954965823184634e-06,
+      "loss": 17.3658,
+      "step": 15940
+    },
+    {
+      "epoch": 0.2884000352589385,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.954937570877975e-06,
+      "loss": 17.1942,
+      "step": 15950
+    },
+    {
+      "epoch": 0.28858085032806635,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.954909318571316e-06,
+      "loss": 17.3323,
+      "step": 15960
+    },
+    {
+      "epoch": 0.2887616653971942,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.954881066264656e-06,
+      "loss": 17.148,
+      "step": 15970
+    },
+    {
+      "epoch": 0.28894248046632204,
+      "grad_norm": 38.375,
+      "learning_rate": 9.954852813957997e-06,
+      "loss": 17.149,
+      "step": 15980
+    },
+    {
+      "epoch": 0.2891232955354499,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.954824561651337e-06,
+      "loss": 17.4588,
+      "step": 15990
+    },
+    {
+      "epoch": 0.2893041106045778,
+      "grad_norm": 40.75,
+      "learning_rate": 9.954796309344678e-06,
+      "loss": 17.116,
+      "step": 16000
+    },
+    {
+      "epoch": 0.28948492567370565,
+      "grad_norm": 38.5625,
+      "learning_rate": 9.954768057038019e-06,
+      "loss": 16.7949,
+      "step": 16010
+    },
+    {
+      "epoch": 0.2896657407428335,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.954739804731358e-06,
+      "loss": 17.0813,
+      "step": 16020
+    },
+    {
+      "epoch": 0.2898465558119614,
+      "grad_norm": 39.625,
+      "learning_rate": 9.954711552424698e-06,
+      "loss": 17.6624,
+      "step": 16030
+    },
+    {
+      "epoch": 0.2900273708810892,
+      "grad_norm": 40.0,
+      "learning_rate": 9.954683300118039e-06,
+      "loss": 17.3748,
+      "step": 16040
+    },
+    {
+      "epoch": 0.2902081859502171,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.95465504781138e-06,
+      "loss": 17.3348,
+      "step": 16050
+    },
+    {
+      "epoch": 0.29038900101934495,
+      "grad_norm": 38.28125,
+      "learning_rate": 9.95462679550472e-06,
+      "loss": 16.9131,
+      "step": 16060
+    },
+    {
+      "epoch": 0.2905698160884728,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.95459854319806e-06,
+      "loss": 17.5051,
+      "step": 16070
+    },
+    {
+      "epoch": 0.2907506311576007,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.954570290891401e-06,
+      "loss": 17.3849,
+      "step": 16080
+    },
+    {
+      "epoch": 0.29093144622672856,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.954542038584742e-06,
+      "loss": 17.4643,
+      "step": 16090
+    },
+    {
+      "epoch": 0.2911122612958564,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.954513786278081e-06,
+      "loss": 17.4884,
+      "step": 16100
+    },
+    {
+      "epoch": 0.29129307636498425,
+      "grad_norm": 39.875,
+      "learning_rate": 9.954485533971422e-06,
+      "loss": 17.3067,
+      "step": 16110
+    },
+    {
+      "epoch": 0.2914738914341121,
+      "grad_norm": 38.53125,
+      "learning_rate": 9.954457281664762e-06,
+      "loss": 17.2178,
+      "step": 16120
+    },
+    {
+      "epoch": 0.29165470650324,
+      "grad_norm": 42.875,
+      "learning_rate": 9.954429029358103e-06,
+      "loss": 17.2724,
+      "step": 16130
+    },
+    {
+      "epoch": 0.29183552157236786,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.954400777051443e-06,
+      "loss": 17.2351,
+      "step": 16140
+    },
+    {
+      "epoch": 0.2920163366414957,
+      "grad_norm": 38.59375,
+      "learning_rate": 9.954372524744782e-06,
+      "loss": 17.3964,
+      "step": 16150
+    },
+    {
+      "epoch": 0.29219715171062355,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.954344272438125e-06,
+      "loss": 17.1551,
+      "step": 16160
+    },
+    {
+      "epoch": 0.2923779667797514,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.954316020131465e-06,
+      "loss": 17.0422,
+      "step": 16170
+    },
+    {
+      "epoch": 0.2925587818488793,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.954287767824806e-06,
+      "loss": 17.5605,
+      "step": 16180
+    },
+    {
+      "epoch": 0.29273959691800716,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.954259515518145e-06,
+      "loss": 17.3039,
+      "step": 16190
+    },
+    {
+      "epoch": 0.29292041198713503,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.954231263211485e-06,
+      "loss": 17.0426,
+      "step": 16200
+    },
+    {
+      "epoch": 0.29310122705626285,
+      "grad_norm": 42.0,
+      "learning_rate": 9.954203010904826e-06,
+      "loss": 17.1135,
+      "step": 16210
+    },
+    {
+      "epoch": 0.2932820421253907,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.954174758598167e-06,
+      "loss": 17.2362,
+      "step": 16220
+    },
+    {
+      "epoch": 0.2934628571945186,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.954146506291507e-06,
+      "loss": 17.3939,
+      "step": 16230
+    },
+    {
+      "epoch": 0.29364367226364646,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.954118253984846e-06,
+      "loss": 17.2928,
+      "step": 16240
+    },
+    {
+      "epoch": 0.29382448733277433,
+      "grad_norm": 42.125,
+      "learning_rate": 9.954090001678189e-06,
+      "loss": 17.3053,
+      "step": 16250
+    },
+    {
+      "epoch": 0.2940053024019022,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.95406174937153e-06,
+      "loss": 17.1759,
+      "step": 16260
+    },
+    {
+      "epoch": 0.29418611747103,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.954033497064868e-06,
+      "loss": 17.2078,
+      "step": 16270
+    },
+    {
+      "epoch": 0.2943669325401579,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.954005244758209e-06,
+      "loss": 17.2217,
+      "step": 16280
+    },
+    {
+      "epoch": 0.29454774760928576,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.95397699245155e-06,
+      "loss": 17.5906,
+      "step": 16290
+    },
+    {
+      "epoch": 0.29472856267841363,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.95394874014489e-06,
+      "loss": 17.3592,
+      "step": 16300
+    },
+    {
+      "epoch": 0.2949093777475415,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.95392048783823e-06,
+      "loss": 17.7649,
+      "step": 16310
+    },
+    {
+      "epoch": 0.2950901928166693,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.953892235531571e-06,
+      "loss": 17.0514,
+      "step": 16320
+    },
+    {
+      "epoch": 0.2952710078857972,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.953863983224912e-06,
+      "loss": 17.0892,
+      "step": 16330
+    },
+    {
+      "epoch": 0.29545182295492506,
+      "grad_norm": 38.75,
+      "learning_rate": 9.953835730918252e-06,
+      "loss": 17.6396,
+      "step": 16340
+    },
+    {
+      "epoch": 0.29563263802405293,
+      "grad_norm": 40.75,
+      "learning_rate": 9.953807478611593e-06,
+      "loss": 17.1078,
+      "step": 16350
+    },
+    {
+      "epoch": 0.2958134530931808,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.953779226304932e-06,
+      "loss": 17.0602,
+      "step": 16360
+    },
+    {
+      "epoch": 0.29599426816230867,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.953750973998273e-06,
+      "loss": 17.2024,
+      "step": 16370
+    },
+    {
+      "epoch": 0.2961750832314365,
+      "grad_norm": 44.0,
+      "learning_rate": 9.953722721691613e-06,
+      "loss": 17.265,
+      "step": 16380
+    },
+    {
+      "epoch": 0.29635589830056436,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.953694469384954e-06,
+      "loss": 16.996,
+      "step": 16390
+    },
+    {
+      "epoch": 0.29653671336969223,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.953666217078294e-06,
+      "loss": 16.7257,
+      "step": 16400
+    },
+    {
+      "epoch": 0.2967175284388201,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.953637964771633e-06,
+      "loss": 17.1792,
+      "step": 16410
+    },
+    {
+      "epoch": 0.29689834350794797,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.953609712464976e-06,
+      "loss": 17.4675,
+      "step": 16420
+    },
+    {
+      "epoch": 0.2970791585770758,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.953581460158316e-06,
+      "loss": 17.298,
+      "step": 16430
+    },
+    {
+      "epoch": 0.29725997364620366,
+      "grad_norm": 38.09375,
+      "learning_rate": 9.953553207851657e-06,
+      "loss": 17.1823,
+      "step": 16440
+    },
+    {
+      "epoch": 0.2974407887153315,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.953524955544996e-06,
+      "loss": 17.3904,
+      "step": 16450
+    },
+    {
+      "epoch": 0.2976216037844594,
+      "grad_norm": 40.5,
+      "learning_rate": 9.953496703238337e-06,
+      "loss": 17.1084,
+      "step": 16460
+    },
+    {
+      "epoch": 0.29780241885358727,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.953468450931677e-06,
+      "loss": 17.1752,
+      "step": 16470
+    },
+    {
+      "epoch": 0.29798323392271514,
+      "grad_norm": 38.0,
+      "learning_rate": 9.953440198625018e-06,
+      "loss": 17.1169,
+      "step": 16480
+    },
+    {
+      "epoch": 0.29816404899184296,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.953411946318358e-06,
+      "loss": 17.6988,
+      "step": 16490
+    },
+    {
+      "epoch": 0.2983448640609708,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.953383694011697e-06,
+      "loss": 17.0947,
+      "step": 16500
+    },
+    {
+      "epoch": 0.2985256791300987,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.95335544170504e-06,
+      "loss": 17.1475,
+      "step": 16510
+    },
+    {
+      "epoch": 0.29870649419922657,
+      "grad_norm": 40.25,
+      "learning_rate": 9.95332718939838e-06,
+      "loss": 17.1444,
+      "step": 16520
+    },
+    {
+      "epoch": 0.29888730926835444,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.95329893709172e-06,
+      "loss": 17.1142,
+      "step": 16530
+    },
+    {
+      "epoch": 0.2990681243374823,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.95327068478506e-06,
+      "loss": 16.9027,
+      "step": 16540
+    },
+    {
+      "epoch": 0.2992489394066101,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.9532424324784e-06,
+      "loss": 17.0472,
+      "step": 16550
+    },
+    {
+      "epoch": 0.299429754475738,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.953214180171741e-06,
+      "loss": 16.765,
+      "step": 16560
+    },
+    {
+      "epoch": 0.29961056954486587,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.953185927865082e-06,
+      "loss": 16.9042,
+      "step": 16570
+    },
+    {
+      "epoch": 0.29979138461399374,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.95315767555842e-06,
+      "loss": 16.9619,
+      "step": 16580
+    },
+    {
+      "epoch": 0.2999721996831216,
+      "grad_norm": 42.25,
+      "learning_rate": 9.953129423251761e-06,
+      "loss": 17.0627,
+      "step": 16590
+    },
+    {
+      "epoch": 0.3001530147522494,
+      "grad_norm": 37.75,
+      "learning_rate": 9.953101170945104e-06,
+      "loss": 17.3962,
+      "step": 16600
+    },
+    {
+      "epoch": 0.3003338298213773,
+      "grad_norm": 40.5,
+      "learning_rate": 9.953072918638444e-06,
+      "loss": 17.0685,
+      "step": 16610
+    },
+    {
+      "epoch": 0.30051464489050517,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.953044666331783e-06,
+      "loss": 17.456,
+      "step": 16620
+    },
+    {
+      "epoch": 0.30069545995963304,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.953016414025124e-06,
+      "loss": 17.1949,
+      "step": 16630
+    },
+    {
+      "epoch": 0.3008762750287609,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.952988161718464e-06,
+      "loss": 17.527,
+      "step": 16640
+    },
+    {
+      "epoch": 0.3010570900978888,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.952959909411805e-06,
+      "loss": 17.1713,
+      "step": 16650
+    },
+    {
+      "epoch": 0.3012379051670166,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.952931657105146e-06,
+      "loss": 16.9757,
+      "step": 16660
+    },
+    {
+      "epoch": 0.30141872023614447,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.952903404798485e-06,
+      "loss": 17.2477,
+      "step": 16670
+    },
+    {
+      "epoch": 0.30159953530527234,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.952875152491827e-06,
+      "loss": 17.256,
+      "step": 16680
+    },
+    {
+      "epoch": 0.3017803503744002,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.952846900185167e-06,
+      "loss": 17.1216,
+      "step": 16690
+    },
+    {
+      "epoch": 0.3019611654435281,
+      "grad_norm": 38.625,
+      "learning_rate": 9.952818647878506e-06,
+      "loss": 17.2615,
+      "step": 16700
+    },
+    {
+      "epoch": 0.30214198051265595,
+      "grad_norm": 39.625,
+      "learning_rate": 9.952790395571847e-06,
+      "loss": 17.2899,
+      "step": 16710
+    },
+    {
+      "epoch": 0.30232279558178377,
+      "grad_norm": 41.375,
+      "learning_rate": 9.952762143265188e-06,
+      "loss": 17.8263,
+      "step": 16720
+    },
+    {
+      "epoch": 0.30250361065091164,
+      "grad_norm": 42.5,
+      "learning_rate": 9.952733890958528e-06,
+      "loss": 17.4072,
+      "step": 16730
+    },
+    {
+      "epoch": 0.3026844257200395,
+      "grad_norm": 44.25,
+      "learning_rate": 9.952705638651869e-06,
+      "loss": 17.9111,
+      "step": 16740
+    },
+    {
+      "epoch": 0.3028652407891674,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.952677386345208e-06,
+      "loss": 17.2915,
+      "step": 16750
+    },
+    {
+      "epoch": 0.30304605585829525,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.952649134038548e-06,
+      "loss": 17.3275,
+      "step": 16760
+    },
+    {
+      "epoch": 0.30322687092742306,
+      "grad_norm": 39.125,
+      "learning_rate": 9.95262088173189e-06,
+      "loss": 17.4011,
+      "step": 16770
+    },
+    {
+      "epoch": 0.30340768599655094,
+      "grad_norm": 41.25,
+      "learning_rate": 9.952592629425231e-06,
+      "loss": 17.0106,
+      "step": 16780
+    },
+    {
+      "epoch": 0.3035885010656788,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.95256437711857e-06,
+      "loss": 17.0403,
+      "step": 16790
+    },
+    {
+      "epoch": 0.3037693161348067,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.952536124811911e-06,
+      "loss": 17.1771,
+      "step": 16800
+    },
+    {
+      "epoch": 0.30395013120393455,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.952507872505252e-06,
+      "loss": 17.4583,
+      "step": 16810
+    },
+    {
+      "epoch": 0.3041309462730624,
+      "grad_norm": 38.09375,
+      "learning_rate": 9.952479620198592e-06,
+      "loss": 17.1125,
+      "step": 16820
+    },
+    {
+      "epoch": 0.30431176134219023,
+      "grad_norm": 40.25,
+      "learning_rate": 9.952451367891933e-06,
+      "loss": 17.0927,
+      "step": 16830
+    },
+    {
+      "epoch": 0.3044925764113181,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.952423115585272e-06,
+      "loss": 17.3882,
+      "step": 16840
+    },
+    {
+      "epoch": 0.304673391480446,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.952394863278612e-06,
+      "loss": 16.8834,
+      "step": 16850
+    },
+    {
+      "epoch": 0.30485420654957385,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.952366610971955e-06,
+      "loss": 17.6202,
+      "step": 16860
+    },
+    {
+      "epoch": 0.3050350216187017,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.952338358665294e-06,
+      "loss": 17.3385,
+      "step": 16870
+    },
+    {
+      "epoch": 0.3052158366878296,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.952310106358634e-06,
+      "loss": 17.5984,
+      "step": 16880
+    },
+    {
+      "epoch": 0.3053966517569574,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.952281854051975e-06,
+      "loss": 17.3033,
+      "step": 16890
+    },
+    {
+      "epoch": 0.3055774668260853,
+      "grad_norm": 43.0,
+      "learning_rate": 9.952253601745315e-06,
+      "loss": 17.1927,
+      "step": 16900
+    },
+    {
+      "epoch": 0.30575828189521315,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.952225349438656e-06,
+      "loss": 17.7446,
+      "step": 16910
+    },
+    {
+      "epoch": 0.305939096964341,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.952197097131997e-06,
+      "loss": 17.2445,
+      "step": 16920
+    },
+    {
+      "epoch": 0.3061199120334689,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.952168844825336e-06,
+      "loss": 17.4012,
+      "step": 16930
+    },
+    {
+      "epoch": 0.3063007271025967,
+      "grad_norm": 41.5,
+      "learning_rate": 9.952140592518676e-06,
+      "loss": 17.2224,
+      "step": 16940
+    },
+    {
+      "epoch": 0.3064815421717246,
+      "grad_norm": 44.75,
+      "learning_rate": 9.952112340212019e-06,
+      "loss": 17.4899,
+      "step": 16950
+    },
+    {
+      "epoch": 0.30666235724085245,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.952084087905357e-06,
+      "loss": 16.8993,
+      "step": 16960
+    },
+    {
+      "epoch": 0.3068431723099803,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.952055835598698e-06,
+      "loss": 17.5821,
+      "step": 16970
+    },
+    {
+      "epoch": 0.3070239873791082,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.952027583292039e-06,
+      "loss": 17.31,
+      "step": 16980
+    },
+    {
+      "epoch": 0.30720480244823606,
+      "grad_norm": 43.0,
+      "learning_rate": 9.95199933098538e-06,
+      "loss": 17.4335,
+      "step": 16990
+    },
+    {
+      "epoch": 0.3073856175173639,
+      "grad_norm": 44.5,
+      "learning_rate": 9.95197107867872e-06,
+      "loss": 17.3602,
+      "step": 17000
+    },
+    {
+      "epoch": 0.30756643258649174,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.951942826372059e-06,
+      "loss": 16.9662,
+      "step": 17010
+    },
+    {
+      "epoch": 0.3077472476556196,
+      "grad_norm": 39.625,
+      "learning_rate": 9.9519145740654e-06,
+      "loss": 17.0458,
+      "step": 17020
+    },
+    {
+      "epoch": 0.3079280627247475,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.95188632175874e-06,
+      "loss": 17.7801,
+      "step": 17030
+    },
+    {
+      "epoch": 0.30810887779387536,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.951858069452082e-06,
+      "loss": 17.4092,
+      "step": 17040
+    },
+    {
+      "epoch": 0.30828969286300323,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.951829817145421e-06,
+      "loss": 17.5656,
+      "step": 17050
+    },
+    {
+      "epoch": 0.30847050793213104,
+      "grad_norm": 41.375,
+      "learning_rate": 9.951801564838762e-06,
+      "loss": 17.5283,
+      "step": 17060
+    },
+    {
+      "epoch": 0.3086513230012589,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.951773312532103e-06,
+      "loss": 17.3261,
+      "step": 17070
+    },
+    {
+      "epoch": 0.3088321380703868,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.951745060225443e-06,
+      "loss": 17.403,
+      "step": 17080
+    },
+    {
+      "epoch": 0.30901295313951466,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.951716807918784e-06,
+      "loss": 17.1059,
+      "step": 17090
+    },
+    {
+      "epoch": 0.3091937682086425,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.951688555612123e-06,
+      "loss": 17.2102,
+      "step": 17100
+    },
+    {
+      "epoch": 0.30937458327777034,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.951660303305463e-06,
+      "loss": 16.8268,
+      "step": 17110
+    },
+    {
+      "epoch": 0.3095553983468982,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.951632050998806e-06,
+      "loss": 16.661,
+      "step": 17120
+    },
+    {
+      "epoch": 0.3097362134160261,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.951603798692145e-06,
+      "loss": 17.2557,
+      "step": 17130
+    },
+    {
+      "epoch": 0.30991702848515396,
+      "grad_norm": 38.5,
+      "learning_rate": 9.951575546385485e-06,
+      "loss": 17.03,
+      "step": 17140
+    },
+    {
+      "epoch": 0.3100978435542818,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.951547294078826e-06,
+      "loss": 16.9144,
+      "step": 17150
+    },
+    {
+      "epoch": 0.3102786586234097,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.951519041772167e-06,
+      "loss": 16.892,
+      "step": 17160
+    },
+    {
+      "epoch": 0.3104594736925375,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.951490789465507e-06,
+      "loss": 17.3718,
+      "step": 17170
+    },
+    {
+      "epoch": 0.3106402887616654,
+      "grad_norm": 43.625,
+      "learning_rate": 9.951462537158846e-06,
+      "loss": 16.9441,
+      "step": 17180
+    },
+    {
+      "epoch": 0.31082110383079325,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.951434284852187e-06,
+      "loss": 16.8048,
+      "step": 17190
+    },
+    {
+      "epoch": 0.3110019188999211,
+      "grad_norm": 40.5,
+      "learning_rate": 9.951406032545527e-06,
+      "loss": 17.1232,
+      "step": 17200
+    },
+    {
+      "epoch": 0.311182733969049,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.95137778023887e-06,
+      "loss": 16.9251,
+      "step": 17210
+    },
+    {
+      "epoch": 0.31136354903817687,
+      "grad_norm": 39.5,
+      "learning_rate": 9.951349527932209e-06,
+      "loss": 17.2344,
+      "step": 17220
+    },
+    {
+      "epoch": 0.3115443641073047,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.95132127562555e-06,
+      "loss": 16.9758,
+      "step": 17230
+    },
+    {
+      "epoch": 0.31172517917643255,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.95129302331889e-06,
+      "loss": 17.0863,
+      "step": 17240
+    },
+    {
+      "epoch": 0.3119059942455604,
+      "grad_norm": 36.375,
+      "learning_rate": 9.95126477101223e-06,
+      "loss": 17.3485,
+      "step": 17250
+    },
+    {
+      "epoch": 0.3120868093146883,
+      "grad_norm": 39.875,
+      "learning_rate": 9.951236518705571e-06,
+      "loss": 17.0591,
+      "step": 17260
+    },
+    {
+      "epoch": 0.31226762438381617,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.95120826639891e-06,
+      "loss": 17.2459,
+      "step": 17270
+    },
+    {
+      "epoch": 0.312448439452944,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.95118001409225e-06,
+      "loss": 17.2446,
+      "step": 17280
+    },
+    {
+      "epoch": 0.31262925452207185,
+      "grad_norm": 40.0,
+      "learning_rate": 9.951151761785591e-06,
+      "loss": 16.8575,
+      "step": 17290
+    },
+    {
+      "epoch": 0.3128100695911997,
+      "grad_norm": 39.875,
+      "learning_rate": 9.951123509478932e-06,
+      "loss": 17.4136,
+      "step": 17300
+    },
+    {
+      "epoch": 0.3129908846603276,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.951095257172272e-06,
+      "loss": 17.2629,
+      "step": 17310
+    },
+    {
+      "epoch": 0.31317169972945547,
+      "grad_norm": 37.375,
+      "learning_rate": 9.951067004865613e-06,
+      "loss": 17.0746,
+      "step": 17320
+    },
+    {
+      "epoch": 0.31335251479858334,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.951038752558954e-06,
+      "loss": 17.2394,
+      "step": 17330
+    },
+    {
+      "epoch": 0.31353332986771115,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.951010500252294e-06,
+      "loss": 17.4258,
+      "step": 17340
+    },
+    {
+      "epoch": 0.313714144936839,
+      "grad_norm": 39.75,
+      "learning_rate": 9.950982247945635e-06,
+      "loss": 17.1237,
+      "step": 17350
+    },
+    {
+      "epoch": 0.3138949600059669,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.950953995638974e-06,
+      "loss": 16.8779,
+      "step": 17360
+    },
+    {
+      "epoch": 0.31407577507509477,
+      "grad_norm": 38.59375,
+      "learning_rate": 9.950925743332315e-06,
+      "loss": 17.5397,
+      "step": 17370
+    },
+    {
+      "epoch": 0.31425659014422264,
+      "grad_norm": 41.75,
+      "learning_rate": 9.950897491025655e-06,
+      "loss": 17.3225,
+      "step": 17380
+    },
+    {
+      "epoch": 0.3144374052133505,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.950869238718996e-06,
+      "loss": 17.1693,
+      "step": 17390
+    },
+    {
+      "epoch": 0.3146182202824783,
+      "grad_norm": 38.5625,
+      "learning_rate": 9.950840986412336e-06,
+      "loss": 17.3698,
+      "step": 17400
+    },
+    {
+      "epoch": 0.3147990353516062,
+      "grad_norm": 39.875,
+      "learning_rate": 9.950812734105677e-06,
+      "loss": 17.0733,
+      "step": 17410
+    },
+    {
+      "epoch": 0.31497985042073406,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.950784481799018e-06,
+      "loss": 16.9899,
+      "step": 17420
+    },
+    {
+      "epoch": 0.31516066548986194,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.950756229492358e-06,
+      "loss": 17.282,
+      "step": 17430
+    },
+    {
+      "epoch": 0.3153414805589898,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.950727977185697e-06,
+      "loss": 17.2829,
+      "step": 17440
+    },
+    {
+      "epoch": 0.3155222956281176,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.950699724879038e-06,
+      "loss": 17.432,
+      "step": 17450
+    },
+    {
+      "epoch": 0.3157031106972455,
+      "grad_norm": 36.0,
+      "learning_rate": 9.950671472572378e-06,
+      "loss": 17.3085,
+      "step": 17460
+    },
+    {
+      "epoch": 0.31588392576637336,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.95064322026572e-06,
+      "loss": 17.0396,
+      "step": 17470
+    },
+    {
+      "epoch": 0.31606474083550123,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.95061496795906e-06,
+      "loss": 17.3051,
+      "step": 17480
+    },
+    {
+      "epoch": 0.3162455559046291,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.9505867156524e-06,
+      "loss": 17.087,
+      "step": 17490
+    },
+    {
+      "epoch": 0.316426370973757,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.950558463345741e-06,
+      "loss": 17.0963,
+      "step": 17500
+    },
+    {
+      "epoch": 0.3166071860428848,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.950530211039082e-06,
+      "loss": 17.3186,
+      "step": 17510
+    },
+    {
+      "epoch": 0.31678800111201266,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.950501958732422e-06,
+      "loss": 17.0498,
+      "step": 17520
+    },
+    {
+      "epoch": 0.31696881618114053,
+      "grad_norm": 40.375,
+      "learning_rate": 9.950473706425761e-06,
+      "loss": 17.2796,
+      "step": 17530
+    },
+    {
+      "epoch": 0.3171496312502684,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.950445454119102e-06,
+      "loss": 16.9422,
+      "step": 17540
+    },
+    {
+      "epoch": 0.3173304463193963,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.950417201812442e-06,
+      "loss": 17.1075,
+      "step": 17550
+    },
+    {
+      "epoch": 0.31751126138852415,
+      "grad_norm": 41.0,
+      "learning_rate": 9.950388949505783e-06,
+      "loss": 17.3609,
+      "step": 17560
+    },
+    {
+      "epoch": 0.31769207645765196,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.950360697199124e-06,
+      "loss": 17.1011,
+      "step": 17570
+    },
+    {
+      "epoch": 0.31787289152677983,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.950332444892464e-06,
+      "loss": 17.5317,
+      "step": 17580
+    },
+    {
+      "epoch": 0.3180537065959077,
+      "grad_norm": 37.0,
+      "learning_rate": 9.950304192585805e-06,
+      "loss": 17.2281,
+      "step": 17590
+    },
+    {
+      "epoch": 0.3182345216650356,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.950275940279145e-06,
+      "loss": 16.8669,
+      "step": 17600
+    },
+    {
+      "epoch": 0.31841533673416345,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.950247687972484e-06,
+      "loss": 16.9876,
+      "step": 17610
+    },
+    {
+      "epoch": 0.31859615180329126,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.950219435665825e-06,
+      "loss": 17.7563,
+      "step": 17620
+    },
+    {
+      "epoch": 0.31877696687241913,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.950191183359166e-06,
+      "loss": 17.0608,
+      "step": 17630
+    },
+    {
+      "epoch": 0.318957781941547,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.950162931052506e-06,
+      "loss": 17.4674,
+      "step": 17640
+    },
+    {
+      "epoch": 0.3191385970106749,
+      "grad_norm": 40.5,
+      "learning_rate": 9.950134678745847e-06,
+      "loss": 16.9903,
+      "step": 17650
+    },
+    {
+      "epoch": 0.31931941207980274,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.950106426439187e-06,
+      "loss": 17.3048,
+      "step": 17660
+    },
+    {
+      "epoch": 0.3195002271489306,
+      "grad_norm": 39.25,
+      "learning_rate": 9.950078174132528e-06,
+      "loss": 17.2867,
+      "step": 17670
+    },
+    {
+      "epoch": 0.31968104221805843,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.950049921825869e-06,
+      "loss": 17.6517,
+      "step": 17680
+    },
+    {
+      "epoch": 0.3198618572871863,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.95002166951921e-06,
+      "loss": 17.5702,
+      "step": 17690
+    },
+    {
+      "epoch": 0.3200426723563142,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.949993417212548e-06,
+      "loss": 17.4709,
+      "step": 17700
+    },
+    {
+      "epoch": 0.32022348742544204,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.949965164905889e-06,
+      "loss": 17.372,
+      "step": 17710
+    },
+    {
+      "epoch": 0.3204043024945699,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.94993691259923e-06,
+      "loss": 17.1449,
+      "step": 17720
+    },
+    {
+      "epoch": 0.3205851175636978,
+      "grad_norm": 39.875,
+      "learning_rate": 9.94990866029257e-06,
+      "loss": 16.9721,
+      "step": 17730
+    },
+    {
+      "epoch": 0.3207659326328256,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.94988040798591e-06,
+      "loss": 17.058,
+      "step": 17740
+    },
+    {
+      "epoch": 0.32094674770195347,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.949852155679251e-06,
+      "loss": 17.1496,
+      "step": 17750
+    },
+    {
+      "epoch": 0.32112756277108134,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.949823903372592e-06,
+      "loss": 17.3002,
+      "step": 17760
+    },
+    {
+      "epoch": 0.3213083778402092,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.949795651065933e-06,
+      "loss": 17.1647,
+      "step": 17770
+    },
+    {
+      "epoch": 0.3214891929093371,
+      "grad_norm": 41.625,
+      "learning_rate": 9.949767398759273e-06,
+      "loss": 17.1988,
+      "step": 17780
+    },
+    {
+      "epoch": 0.3216700079784649,
+      "grad_norm": 43.875,
+      "learning_rate": 9.949739146452612e-06,
+      "loss": 17.3825,
+      "step": 17790
+    },
+    {
+      "epoch": 0.32185082304759277,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.949710894145953e-06,
+      "loss": 17.05,
+      "step": 17800
+    },
+    {
+      "epoch": 0.32203163811672064,
+      "grad_norm": 43.59375,
+      "learning_rate": 9.949682641839293e-06,
+      "loss": 17.0607,
+      "step": 17810
+    },
+    {
+      "epoch": 0.3222124531858485,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.949654389532634e-06,
+      "loss": 16.5678,
+      "step": 17820
+    },
+    {
+      "epoch": 0.3223932682549764,
+      "grad_norm": 42.0,
+      "learning_rate": 9.949626137225975e-06,
+      "loss": 17.0074,
+      "step": 17830
+    },
+    {
+      "epoch": 0.32257408332410425,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.949597884919315e-06,
+      "loss": 17.5844,
+      "step": 17840
+    },
+    {
+      "epoch": 0.32275489839323207,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.949569632612656e-06,
+      "loss": 17.0044,
+      "step": 17850
+    },
+    {
+      "epoch": 0.32293571346235994,
+      "grad_norm": 40.375,
+      "learning_rate": 9.949541380305997e-06,
+      "loss": 17.272,
+      "step": 17860
+    },
+    {
+      "epoch": 0.3231165285314878,
+      "grad_norm": 41.75,
+      "learning_rate": 9.949513127999335e-06,
+      "loss": 17.2767,
+      "step": 17870
+    },
+    {
+      "epoch": 0.3232973436006157,
+      "grad_norm": 40.125,
+      "learning_rate": 9.949484875692676e-06,
+      "loss": 16.9778,
+      "step": 17880
+    },
+    {
+      "epoch": 0.32347815866974355,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.949456623386017e-06,
+      "loss": 17.1859,
+      "step": 17890
+    },
+    {
+      "epoch": 0.3236589737388714,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.949428371079357e-06,
+      "loss": 16.9518,
+      "step": 17900
+    },
+    {
+      "epoch": 0.32383978880799924,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.949400118772698e-06,
+      "loss": 17.2181,
+      "step": 17910
+    },
+    {
+      "epoch": 0.3240206038771271,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.949371866466039e-06,
+      "loss": 17.0839,
+      "step": 17920
+    },
+    {
+      "epoch": 0.324201418946255,
+      "grad_norm": 41.0,
+      "learning_rate": 9.94934361415938e-06,
+      "loss": 16.9002,
+      "step": 17930
+    },
+    {
+      "epoch": 0.32438223401538285,
+      "grad_norm": 37.875,
+      "learning_rate": 9.94931536185272e-06,
+      "loss": 17.3268,
+      "step": 17940
+    },
+    {
+      "epoch": 0.3245630490845107,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.94928710954606e-06,
+      "loss": 17.4407,
+      "step": 17950
+    },
+    {
+      "epoch": 0.32474386415363854,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.9492588572394e-06,
+      "loss": 17.2294,
+      "step": 17960
+    },
+    {
+      "epoch": 0.3249246792227664,
+      "grad_norm": 41.625,
+      "learning_rate": 9.94923060493274e-06,
+      "loss": 17.0065,
+      "step": 17970
+    },
+    {
+      "epoch": 0.3251054942918943,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.94920235262608e-06,
+      "loss": 17.3333,
+      "step": 17980
+    },
+    {
+      "epoch": 0.32528630936102215,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.949174100319421e-06,
+      "loss": 17.2219,
+      "step": 17990
+    },
+    {
+      "epoch": 0.32546712443015,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.949145848012762e-06,
+      "loss": 17.4031,
+      "step": 18000
+    },
+    {
+      "epoch": 0.3256479394992779,
+      "grad_norm": 36.53125,
+      "learning_rate": 9.949117595706102e-06,
+      "loss": 17.2132,
+      "step": 18010
+    },
+    {
+      "epoch": 0.3258287545684057,
+      "grad_norm": 37.28125,
+      "learning_rate": 9.949089343399443e-06,
+      "loss": 17.2204,
+      "step": 18020
+    },
+    {
+      "epoch": 0.3260095696375336,
+      "grad_norm": 42.375,
+      "learning_rate": 9.949061091092784e-06,
+      "loss": 17.3199,
+      "step": 18030
+    },
+    {
+      "epoch": 0.32619038470666145,
+      "grad_norm": 38.5,
+      "learning_rate": 9.949032838786123e-06,
+      "loss": 17.6373,
+      "step": 18040
+    },
+    {
+      "epoch": 0.3263711997757893,
+      "grad_norm": 42.375,
+      "learning_rate": 9.949004586479463e-06,
+      "loss": 17.1415,
+      "step": 18050
+    },
+    {
+      "epoch": 0.3265520148449172,
+      "grad_norm": 41.0,
+      "learning_rate": 9.948976334172804e-06,
+      "loss": 17.1106,
+      "step": 18060
+    },
+    {
+      "epoch": 0.32673282991404506,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.948948081866145e-06,
+      "loss": 17.5328,
+      "step": 18070
+    },
+    {
+      "epoch": 0.3269136449831729,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.948919829559485e-06,
+      "loss": 17.2637,
+      "step": 18080
+    },
+    {
+      "epoch": 0.32709446005230075,
+      "grad_norm": 41.375,
+      "learning_rate": 9.948891577252826e-06,
+      "loss": 16.6566,
+      "step": 18090
+    },
+    {
+      "epoch": 0.3272752751214286,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.948863324946166e-06,
+      "loss": 17.204,
+      "step": 18100
+    },
+    {
+      "epoch": 0.3274560901905565,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.948835072639507e-06,
+      "loss": 17.4579,
+      "step": 18110
+    },
+    {
+      "epoch": 0.32763690525968436,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.948806820332848e-06,
+      "loss": 17.6162,
+      "step": 18120
+    },
+    {
+      "epoch": 0.3278177203288122,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.948778568026187e-06,
+      "loss": 17.0227,
+      "step": 18130
+    },
+    {
+      "epoch": 0.32799853539794005,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.948750315719527e-06,
+      "loss": 17.1382,
+      "step": 18140
+    },
+    {
+      "epoch": 0.3281793504670679,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.948722063412868e-06,
+      "loss": 17.3412,
+      "step": 18150
+    },
+    {
+      "epoch": 0.3283601655361958,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.948693811106208e-06,
+      "loss": 17.1184,
+      "step": 18160
+    },
+    {
+      "epoch": 0.32854098060532366,
+      "grad_norm": 42.5,
+      "learning_rate": 9.948665558799549e-06,
+      "loss": 16.8094,
+      "step": 18170
+    },
+    {
+      "epoch": 0.32872179567445153,
+      "grad_norm": 39.5,
+      "learning_rate": 9.94863730649289e-06,
+      "loss": 16.9699,
+      "step": 18180
+    },
+    {
+      "epoch": 0.32890261074357935,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.94860905418623e-06,
+      "loss": 17.0162,
+      "step": 18190
+    },
+    {
+      "epoch": 0.3290834258127072,
+      "grad_norm": 39.875,
+      "learning_rate": 9.948580801879571e-06,
+      "loss": 17.3723,
+      "step": 18200
+    },
+    {
+      "epoch": 0.3292642408818351,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.948552549572912e-06,
+      "loss": 17.3077,
+      "step": 18210
+    },
+    {
+      "epoch": 0.32944505595096296,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.94852429726625e-06,
+      "loss": 16.9637,
+      "step": 18220
+    },
+    {
+      "epoch": 0.32962587102009083,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.948496044959591e-06,
+      "loss": 17.1387,
+      "step": 18230
+    },
+    {
+      "epoch": 0.32980668608921865,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.948467792652932e-06,
+      "loss": 17.3337,
+      "step": 18240
+    },
+    {
+      "epoch": 0.3299875011583465,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.948439540346272e-06,
+      "loss": 17.097,
+      "step": 18250
+    },
+    {
+      "epoch": 0.3301683162274744,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.948411288039613e-06,
+      "loss": 17.0158,
+      "step": 18260
+    },
+    {
+      "epoch": 0.33034913129660226,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.948383035732954e-06,
+      "loss": 17.2627,
+      "step": 18270
+    },
+    {
+      "epoch": 0.33052994636573013,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.948354783426294e-06,
+      "loss": 16.8991,
+      "step": 18280
+    },
+    {
+      "epoch": 0.330710761434858,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.948326531119635e-06,
+      "loss": 17.0619,
+      "step": 18290
+    },
+    {
+      "epoch": 0.3308915765039858,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.948298278812974e-06,
+      "loss": 17.3769,
+      "step": 18300
+    },
+    {
+      "epoch": 0.3310723915731137,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.948270026506314e-06,
+      "loss": 17.4126,
+      "step": 18310
+    },
+    {
+      "epoch": 0.33125320664224156,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.948241774199655e-06,
+      "loss": 17.0779,
+      "step": 18320
+    },
+    {
+      "epoch": 0.33143402171136943,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.948213521892996e-06,
+      "loss": 16.984,
+      "step": 18330
+    },
+    {
+      "epoch": 0.3316148367804973,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.948185269586336e-06,
+      "loss": 16.7094,
+      "step": 18340
+    },
+    {
+      "epoch": 0.3317956518496252,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.948157017279677e-06,
+      "loss": 17.0435,
+      "step": 18350
+    },
+    {
+      "epoch": 0.331976466918753,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.948128764973017e-06,
+      "loss": 17.2927,
+      "step": 18360
+    },
+    {
+      "epoch": 0.33215728198788086,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.948100512666358e-06,
+      "loss": 17.579,
+      "step": 18370
+    },
+    {
+      "epoch": 0.33233809705700873,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.948072260359699e-06,
+      "loss": 17.1167,
+      "step": 18380
+    },
+    {
+      "epoch": 0.3325189121261366,
+      "grad_norm": 40.875,
+      "learning_rate": 9.948044008053038e-06,
+      "loss": 17.3039,
+      "step": 18390
+    },
+    {
+      "epoch": 0.33269972719526447,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.948015755746378e-06,
+      "loss": 17.0618,
+      "step": 18400
+    },
+    {
+      "epoch": 0.3328805422643923,
+      "grad_norm": 38.0625,
+      "learning_rate": 9.947987503439719e-06,
+      "loss": 17.0005,
+      "step": 18410
+    },
+    {
+      "epoch": 0.33306135733352016,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.94795925113306e-06,
+      "loss": 17.2511,
+      "step": 18420
+    },
+    {
+      "epoch": 0.33324217240264803,
+      "grad_norm": 37.9375,
+      "learning_rate": 9.9479309988264e-06,
+      "loss": 16.8679,
+      "step": 18430
+    },
+    {
+      "epoch": 0.3334229874717759,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.94790274651974e-06,
+      "loss": 17.434,
+      "step": 18440
+    },
+    {
+      "epoch": 0.33360380254090377,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.947874494213081e-06,
+      "loss": 17.287,
+      "step": 18450
+    },
+    {
+      "epoch": 0.33378461761003164,
+      "grad_norm": 37.5625,
+      "learning_rate": 9.947846241906422e-06,
+      "loss": 17.7589,
+      "step": 18460
+    },
+    {
+      "epoch": 0.33396543267915946,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.947817989599761e-06,
+      "loss": 16.7847,
+      "step": 18470
+    },
+    {
+      "epoch": 0.33414624774828733,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.947789737293102e-06,
+      "loss": 16.9083,
+      "step": 18480
+    },
+    {
+      "epoch": 0.3343270628174152,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.947761484986442e-06,
+      "loss": 17.2558,
+      "step": 18490
+    },
+    {
+      "epoch": 0.33450787788654307,
+      "grad_norm": 39.75,
+      "learning_rate": 9.947733232679783e-06,
+      "loss": 16.9924,
+      "step": 18500
+    },
+    {
+      "epoch": 0.33468869295567094,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.947704980373123e-06,
+      "loss": 17.0651,
+      "step": 18510
+    },
+    {
+      "epoch": 0.3348695080247988,
+      "grad_norm": 42.875,
+      "learning_rate": 9.947676728066464e-06,
+      "loss": 16.8699,
+      "step": 18520
+    },
+    {
+      "epoch": 0.3350503230939266,
+      "grad_norm": 37.375,
+      "learning_rate": 9.947648475759805e-06,
+      "loss": 17.208,
+      "step": 18530
+    },
+    {
+      "epoch": 0.3352311381630545,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.947620223453145e-06,
+      "loss": 17.3355,
+      "step": 18540
+    },
+    {
+      "epoch": 0.33541195323218237,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.947591971146486e-06,
+      "loss": 16.9956,
+      "step": 18550
+    },
+    {
+      "epoch": 0.33559276830131024,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.947563718839825e-06,
+      "loss": 16.8433,
+      "step": 18560
+    },
+    {
+      "epoch": 0.3357735833704381,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.947535466533165e-06,
+      "loss": 17.1173,
+      "step": 18570
+    },
+    {
+      "epoch": 0.3359543984395659,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.947507214226506e-06,
+      "loss": 17.3725,
+      "step": 18580
+    },
+    {
+      "epoch": 0.3361352135086938,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.947478961919847e-06,
+      "loss": 17.1542,
+      "step": 18590
+    },
+    {
+      "epoch": 0.33631602857782167,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.947450709613187e-06,
+      "loss": 17.6042,
+      "step": 18600
+    },
+    {
+      "epoch": 0.33649684364694954,
+      "grad_norm": 43.5,
+      "learning_rate": 9.947422457306528e-06,
+      "loss": 17.1111,
+      "step": 18610
+    },
+    {
+      "epoch": 0.3366776587160774,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.947394204999869e-06,
+      "loss": 16.7568,
+      "step": 18620
+    },
+    {
+      "epoch": 0.3368584737852053,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.94736595269321e-06,
+      "loss": 16.7065,
+      "step": 18630
+    },
+    {
+      "epoch": 0.3370392888543331,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.94733770038655e-06,
+      "loss": 16.9225,
+      "step": 18640
+    },
+    {
+      "epoch": 0.33722010392346097,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.947309448079889e-06,
+      "loss": 17.0526,
+      "step": 18650
+    },
+    {
+      "epoch": 0.33740091899258884,
+      "grad_norm": 38.625,
+      "learning_rate": 9.94728119577323e-06,
+      "loss": 17.6703,
+      "step": 18660
+    },
+    {
+      "epoch": 0.3375817340617167,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.94725294346657e-06,
+      "loss": 17.1176,
+      "step": 18670
+    },
+    {
+      "epoch": 0.3377625491308446,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.94722469115991e-06,
+      "loss": 17.3024,
+      "step": 18680
+    },
+    {
+      "epoch": 0.33794336419997245,
+      "grad_norm": 40.25,
+      "learning_rate": 9.947196438853251e-06,
+      "loss": 17.4143,
+      "step": 18690
+    },
+    {
+      "epoch": 0.33812417926910027,
+      "grad_norm": 38.5,
+      "learning_rate": 9.947168186546592e-06,
+      "loss": 17.0131,
+      "step": 18700
+    },
+    {
+      "epoch": 0.33830499433822814,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.947139934239932e-06,
+      "loss": 17.4532,
+      "step": 18710
+    },
+    {
+      "epoch": 0.338485809407356,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.947111681933273e-06,
+      "loss": 17.7063,
+      "step": 18720
+    },
+    {
+      "epoch": 0.3386666244764839,
+      "grad_norm": 44.6875,
+      "learning_rate": 9.947083429626612e-06,
+      "loss": 17.6791,
+      "step": 18730
+    },
+    {
+      "epoch": 0.33884743954561175,
+      "grad_norm": 37.96875,
+      "learning_rate": 9.947055177319953e-06,
+      "loss": 17.2725,
+      "step": 18740
+    },
+    {
+      "epoch": 0.33902825461473957,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.947026925013293e-06,
+      "loss": 17.5307,
+      "step": 18750
+    },
+    {
+      "epoch": 0.33920906968386744,
+      "grad_norm": 38.25,
+      "learning_rate": 9.946998672706634e-06,
+      "loss": 17.2355,
+      "step": 18760
+    },
+    {
+      "epoch": 0.3393898847529953,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.946970420399975e-06,
+      "loss": 17.0688,
+      "step": 18770
+    },
+    {
+      "epoch": 0.3395706998221232,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.946942168093313e-06,
+      "loss": 17.016,
+      "step": 18780
+    },
+    {
+      "epoch": 0.33975151489125105,
+      "grad_norm": 41.125,
+      "learning_rate": 9.946913915786656e-06,
+      "loss": 17.0653,
+      "step": 18790
+    },
+    {
+      "epoch": 0.3399323299603789,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.946885663479996e-06,
+      "loss": 17.0459,
+      "step": 18800
+    },
+    {
+      "epoch": 0.34011314502950674,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.946857411173337e-06,
+      "loss": 17.3077,
+      "step": 18810
+    },
+    {
+      "epoch": 0.3402939600986346,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.946829158866676e-06,
+      "loss": 17.0437,
+      "step": 18820
+    },
+    {
+      "epoch": 0.3404747751677625,
+      "grad_norm": 40.25,
+      "learning_rate": 9.946800906560017e-06,
+      "loss": 17.9247,
+      "step": 18830
+    },
+    {
+      "epoch": 0.34065559023689035,
+      "grad_norm": 41.625,
+      "learning_rate": 9.946772654253357e-06,
+      "loss": 17.3083,
+      "step": 18840
+    },
+    {
+      "epoch": 0.3408364053060182,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.946744401946698e-06,
+      "loss": 16.8552,
+      "step": 18850
+    },
+    {
+      "epoch": 0.3410172203751461,
+      "grad_norm": 42.875,
+      "learning_rate": 9.946716149640038e-06,
+      "loss": 17.5379,
+      "step": 18860
+    },
+    {
+      "epoch": 0.3411980354442739,
+      "grad_norm": 38.0625,
+      "learning_rate": 9.946687897333379e-06,
+      "loss": 17.1698,
+      "step": 18870
+    },
+    {
+      "epoch": 0.3413788505134018,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.94665964502672e-06,
+      "loss": 17.2349,
+      "step": 18880
+    },
+    {
+      "epoch": 0.34155966558252965,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.94663139272006e-06,
+      "loss": 17.123,
+      "step": 18890
+    },
+    {
+      "epoch": 0.3417404806516575,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.9466031404134e-06,
+      "loss": 17.0436,
+      "step": 18900
+    },
+    {
+      "epoch": 0.3419212957207854,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.94657488810674e-06,
+      "loss": 17.3219,
+      "step": 18910
+    },
+    {
+      "epoch": 0.3421021107899132,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.94654663580008e-06,
+      "loss": 17.252,
+      "step": 18920
+    },
+    {
+      "epoch": 0.3422829258590411,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.946518383493421e-06,
+      "loss": 17.3506,
+      "step": 18930
+    },
+    {
+      "epoch": 0.34246374092816895,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.946490131186762e-06,
+      "loss": 17.1539,
+      "step": 18940
+    },
+    {
+      "epoch": 0.3426445559972968,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.946461878880102e-06,
+      "loss": 17.2454,
+      "step": 18950
+    },
+    {
+      "epoch": 0.3428253710664247,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.946433626573443e-06,
+      "loss": 17.1604,
+      "step": 18960
+    },
+    {
+      "epoch": 0.34300618613555256,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.946405374266784e-06,
+      "loss": 17.216,
+      "step": 18970
+    },
+    {
+      "epoch": 0.3431870012046804,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.946377121960124e-06,
+      "loss": 17.0927,
+      "step": 18980
+    },
+    {
+      "epoch": 0.34336781627380825,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.946348869653463e-06,
+      "loss": 17.1234,
+      "step": 18990
+    },
+    {
+      "epoch": 0.3435486313429361,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.946320617346804e-06,
+      "loss": 17.0432,
+      "step": 19000
+    },
+    {
+      "epoch": 0.343729446412064,
+      "grad_norm": 39.75,
+      "learning_rate": 9.946292365040144e-06,
+      "loss": 17.5622,
+      "step": 19010
+    },
+    {
+      "epoch": 0.34391026148119186,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.946264112733485e-06,
+      "loss": 17.1439,
+      "step": 19020
+    },
+    {
+      "epoch": 0.34409107655031973,
+      "grad_norm": 42.125,
+      "learning_rate": 9.946235860426826e-06,
+      "loss": 16.7243,
+      "step": 19030
+    },
+    {
+      "epoch": 0.34427189161944755,
+      "grad_norm": 36.53125,
+      "learning_rate": 9.946207608120165e-06,
+      "loss": 17.1458,
+      "step": 19040
+    },
+    {
+      "epoch": 0.3444527066885754,
+      "grad_norm": 41.25,
+      "learning_rate": 9.946179355813507e-06,
+      "loss": 17.2407,
+      "step": 19050
+    },
+    {
+      "epoch": 0.3446335217577033,
+      "grad_norm": 39.125,
+      "learning_rate": 9.946151103506848e-06,
+      "loss": 17.1062,
+      "step": 19060
+    },
+    {
+      "epoch": 0.34481433682683116,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.946122851200188e-06,
+      "loss": 17.1634,
+      "step": 19070
+    },
+    {
+      "epoch": 0.34499515189595903,
+      "grad_norm": 41.125,
+      "learning_rate": 9.946094598893527e-06,
+      "loss": 17.13,
+      "step": 19080
+    },
+    {
+      "epoch": 0.34517596696508684,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.946066346586868e-06,
+      "loss": 17.0695,
+      "step": 19090
+    },
+    {
+      "epoch": 0.3453567820342147,
+      "grad_norm": 39.625,
+      "learning_rate": 9.946038094280208e-06,
+      "loss": 16.5874,
+      "step": 19100
+    },
+    {
+      "epoch": 0.3455375971033426,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.946009841973549e-06,
+      "loss": 17.1617,
+      "step": 19110
+    },
+    {
+      "epoch": 0.34571841217247046,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.94598158966689e-06,
+      "loss": 17.3538,
+      "step": 19120
+    },
+    {
+      "epoch": 0.34589922724159833,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.945953337360228e-06,
+      "loss": 17.2054,
+      "step": 19130
+    },
+    {
+      "epoch": 0.3460800423107262,
+      "grad_norm": 42.25,
+      "learning_rate": 9.94592508505357e-06,
+      "loss": 17.5536,
+      "step": 19140
+    },
+    {
+      "epoch": 0.346260857379854,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.945896832746911e-06,
+      "loss": 17.1485,
+      "step": 19150
+    },
+    {
+      "epoch": 0.3464416724489819,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.94586858044025e-06,
+      "loss": 17.3381,
+      "step": 19160
+    },
+    {
+      "epoch": 0.34662248751810976,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.945840328133591e-06,
+      "loss": 16.9011,
+      "step": 19170
+    },
+    {
+      "epoch": 0.3468033025872376,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.945812075826932e-06,
+      "loss": 16.9682,
+      "step": 19180
+    },
+    {
+      "epoch": 0.3469841176563655,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.945783823520272e-06,
+      "loss": 16.9983,
+      "step": 19190
+    },
+    {
+      "epoch": 0.34716493272549337,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.945755571213613e-06,
+      "loss": 17.5587,
+      "step": 19200
+    },
+    {
+      "epoch": 0.3473457477946212,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.945727318906952e-06,
+      "loss": 16.8602,
+      "step": 19210
+    },
+    {
+      "epoch": 0.34752656286374906,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.945699066600294e-06,
+      "loss": 17.4146,
+      "step": 19220
+    },
+    {
+      "epoch": 0.3477073779328769,
+      "grad_norm": 42.375,
+      "learning_rate": 9.945670814293635e-06,
+      "loss": 16.7323,
+      "step": 19230
+    },
+    {
+      "epoch": 0.3478881930020048,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.945642561986975e-06,
+      "loss": 17.0783,
+      "step": 19240
+    },
+    {
+      "epoch": 0.34806900807113267,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.945614309680314e-06,
+      "loss": 16.9396,
+      "step": 19250
+    },
+    {
+      "epoch": 0.3482498231402605,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.945586057373655e-06,
+      "loss": 17.1608,
+      "step": 19260
+    },
+    {
+      "epoch": 0.34843063820938835,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.945557805066995e-06,
+      "loss": 17.3588,
+      "step": 19270
+    },
+    {
+      "epoch": 0.3486114532785162,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.945529552760336e-06,
+      "loss": 17.2244,
+      "step": 19280
+    },
+    {
+      "epoch": 0.3487922683476441,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.945501300453677e-06,
+      "loss": 17.6382,
+      "step": 19290
+    },
+    {
+      "epoch": 0.34897308341677197,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.945473048147016e-06,
+      "loss": 17.0822,
+      "step": 19300
+    },
+    {
+      "epoch": 0.34915389848589984,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.945444795840358e-06,
+      "loss": 17.418,
+      "step": 19310
+    },
+    {
+      "epoch": 0.34933471355502765,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.945416543533699e-06,
+      "loss": 17.2986,
+      "step": 19320
+    },
+    {
+      "epoch": 0.3495155286241555,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.945388291227038e-06,
+      "loss": 16.9621,
+      "step": 19330
+    },
+    {
+      "epoch": 0.3496963436932834,
+      "grad_norm": 41.5,
+      "learning_rate": 9.945360038920378e-06,
+      "loss": 17.1927,
+      "step": 19340
+    },
+    {
+      "epoch": 0.34987715876241127,
+      "grad_norm": 40.75,
+      "learning_rate": 9.945331786613719e-06,
+      "loss": 17.5507,
+      "step": 19350
+    },
+    {
+      "epoch": 0.35005797383153914,
+      "grad_norm": 41.375,
+      "learning_rate": 9.94530353430706e-06,
+      "loss": 17.3923,
+      "step": 19360
+    },
+    {
+      "epoch": 0.350238788900667,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.9452752820004e-06,
+      "loss": 17.129,
+      "step": 19370
+    },
+    {
+      "epoch": 0.3504196039697948,
+      "grad_norm": 39.375,
+      "learning_rate": 9.945247029693739e-06,
+      "loss": 17.591,
+      "step": 19380
+    },
+    {
+      "epoch": 0.3506004190389227,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.94521877738708e-06,
+      "loss": 17.3433,
+      "step": 19390
+    },
+    {
+      "epoch": 0.35078123410805057,
+      "grad_norm": 38.09375,
+      "learning_rate": 9.945190525080422e-06,
+      "loss": 17.116,
+      "step": 19400
+    },
+    {
+      "epoch": 0.35096204917717844,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.945162272773763e-06,
+      "loss": 17.407,
+      "step": 19410
+    },
+    {
+      "epoch": 0.3511428642463063,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.945134020467101e-06,
+      "loss": 16.9265,
+      "step": 19420
+    },
+    {
+      "epoch": 0.3513236793154341,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.945105768160442e-06,
+      "loss": 16.8851,
+      "step": 19430
+    },
+    {
+      "epoch": 0.351504494384562,
+      "grad_norm": 39.5,
+      "learning_rate": 9.945077515853783e-06,
+      "loss": 17.0815,
+      "step": 19440
+    },
+    {
+      "epoch": 0.35168530945368986,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.945049263547123e-06,
+      "loss": 17.43,
+      "step": 19450
+    },
+    {
+      "epoch": 0.35186612452281774,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.945021011240464e-06,
+      "loss": 16.987,
+      "step": 19460
+    },
+    {
+      "epoch": 0.3520469395919456,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.944992758933803e-06,
+      "loss": 16.6883,
+      "step": 19470
+    },
+    {
+      "epoch": 0.3522277546610735,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.944964506627143e-06,
+      "loss": 17.3894,
+      "step": 19480
+    },
+    {
+      "epoch": 0.3524085697302013,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.944936254320486e-06,
+      "loss": 17.1256,
+      "step": 19490
+    },
+    {
+      "epoch": 0.35258938479932916,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.944908002013825e-06,
+      "loss": 17.2746,
+      "step": 19500
+    },
+    {
+      "epoch": 0.35277019986845704,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.944879749707165e-06,
+      "loss": 17.3932,
+      "step": 19510
+    },
+    {
+      "epoch": 0.3529510149375849,
+      "grad_norm": 38.0625,
+      "learning_rate": 9.944851497400506e-06,
+      "loss": 16.6536,
+      "step": 19520
+    },
+    {
+      "epoch": 0.3531318300067128,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.944823245093847e-06,
+      "loss": 17.1047,
+      "step": 19530
+    },
+    {
+      "epoch": 0.35331264507584065,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.944794992787187e-06,
+      "loss": 17.5294,
+      "step": 19540
+    },
+    {
+      "epoch": 0.35349346014496846,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.944766740480528e-06,
+      "loss": 16.927,
+      "step": 19550
+    },
+    {
+      "epoch": 0.35367427521409633,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.944738488173867e-06,
+      "loss": 17.1587,
+      "step": 19560
+    },
+    {
+      "epoch": 0.3538550902832242,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.944710235867209e-06,
+      "loss": 17.4372,
+      "step": 19570
+    },
+    {
+      "epoch": 0.3540359053523521,
+      "grad_norm": 41.0,
+      "learning_rate": 9.94468198356055e-06,
+      "loss": 17.637,
+      "step": 19580
+    },
+    {
+      "epoch": 0.35421672042147995,
+      "grad_norm": 42.125,
+      "learning_rate": 9.944653731253889e-06,
+      "loss": 17.1499,
+      "step": 19590
+    },
+    {
+      "epoch": 0.35439753549060776,
+      "grad_norm": 37.25,
+      "learning_rate": 9.94462547894723e-06,
+      "loss": 17.0127,
+      "step": 19600
+    },
+    {
+      "epoch": 0.35457835055973563,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.94459722664057e-06,
+      "loss": 17.0926,
+      "step": 19610
+    },
+    {
+      "epoch": 0.3547591656288635,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.94456897433391e-06,
+      "loss": 17.227,
+      "step": 19620
+    },
+    {
+      "epoch": 0.3549399806979914,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.944540722027251e-06,
+      "loss": 16.7545,
+      "step": 19630
+    },
+    {
+      "epoch": 0.35512079576711925,
+      "grad_norm": 40.625,
+      "learning_rate": 9.94451246972059e-06,
+      "loss": 17.1527,
+      "step": 19640
+    },
+    {
+      "epoch": 0.3553016108362471,
+      "grad_norm": 39.5,
+      "learning_rate": 9.94448421741393e-06,
+      "loss": 17.1649,
+      "step": 19650
+    },
+    {
+      "epoch": 0.35548242590537493,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.944455965107273e-06,
+      "loss": 17.5156,
+      "step": 19660
+    },
+    {
+      "epoch": 0.3556632409745028,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.944427712800614e-06,
+      "loss": 16.9445,
+      "step": 19670
+    },
+    {
+      "epoch": 0.3558440560436307,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.944399460493953e-06,
+      "loss": 16.997,
+      "step": 19680
+    },
+    {
+      "epoch": 0.35602487111275855,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.944371208187293e-06,
+      "loss": 17.3185,
+      "step": 19690
+    },
+    {
+      "epoch": 0.3562056861818864,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.944342955880634e-06,
+      "loss": 17.1195,
+      "step": 19700
+    },
+    {
+      "epoch": 0.3563865012510143,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.944314703573974e-06,
+      "loss": 17.2379,
+      "step": 19710
+    },
+    {
+      "epoch": 0.3565673163201421,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.944286451267315e-06,
+      "loss": 17.1858,
+      "step": 19720
+    },
+    {
+      "epoch": 0.35674813138927,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.944258198960654e-06,
+      "loss": 17.3016,
+      "step": 19730
+    },
+    {
+      "epoch": 0.35692894645839784,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.944229946653995e-06,
+      "loss": 17.0505,
+      "step": 19740
+    },
+    {
+      "epoch": 0.3571097615275257,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.944201694347337e-06,
+      "loss": 16.6689,
+      "step": 19750
+    },
+    {
+      "epoch": 0.3572905765966536,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.944173442040676e-06,
+      "loss": 16.7929,
+      "step": 19760
+    },
+    {
+      "epoch": 0.3574713916657814,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.944145189734016e-06,
+      "loss": 17.0547,
+      "step": 19770
+    },
+    {
+      "epoch": 0.3576522067349093,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.944116937427357e-06,
+      "loss": 17.0405,
+      "step": 19780
+    },
+    {
+      "epoch": 0.35783302180403714,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.944088685120698e-06,
+      "loss": 17.3822,
+      "step": 19790
+    },
+    {
+      "epoch": 0.358013836873165,
+      "grad_norm": 43.125,
+      "learning_rate": 9.944060432814038e-06,
+      "loss": 17.0451,
+      "step": 19800
+    },
+    {
+      "epoch": 0.3581946519422929,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.944032180507377e-06,
+      "loss": 17.159,
+      "step": 19810
+    },
+    {
+      "epoch": 0.35837546701142076,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.944003928200718e-06,
+      "loss": 16.8105,
+      "step": 19820
+    },
+    {
+      "epoch": 0.35855628208054857,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.943975675894058e-06,
+      "loss": 17.3335,
+      "step": 19830
+    },
+    {
+      "epoch": 0.35873709714967644,
+      "grad_norm": 39.25,
+      "learning_rate": 9.9439474235874e-06,
+      "loss": 17.1033,
+      "step": 19840
+    },
+    {
+      "epoch": 0.3589179122188043,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.94391917128074e-06,
+      "loss": 17.2213,
+      "step": 19850
+    },
+    {
+      "epoch": 0.3590987272879322,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.94389091897408e-06,
+      "loss": 16.6604,
+      "step": 19860
+    },
+    {
+      "epoch": 0.35927954235706006,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.943862666667421e-06,
+      "loss": 17.556,
+      "step": 19870
+    },
+    {
+      "epoch": 0.35946035742618787,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.943834414360762e-06,
+      "loss": 16.9048,
+      "step": 19880
+    },
+    {
+      "epoch": 0.35964117249531574,
+      "grad_norm": 41.625,
+      "learning_rate": 9.943806162054102e-06,
+      "loss": 17.0686,
+      "step": 19890
+    },
+    {
+      "epoch": 0.3598219875644436,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.943777909747441e-06,
+      "loss": 17.6033,
+      "step": 19900
+    },
+    {
+      "epoch": 0.3600028026335715,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.943749657440782e-06,
+      "loss": 16.8213,
+      "step": 19910
+    },
+    {
+      "epoch": 0.36018361770269935,
+      "grad_norm": 43.375,
+      "learning_rate": 9.943721405134124e-06,
+      "loss": 17.6674,
+      "step": 19920
+    },
+    {
+      "epoch": 0.3603644327718272,
+      "grad_norm": 39.0,
+      "learning_rate": 9.943693152827463e-06,
+      "loss": 17.1387,
+      "step": 19930
+    },
+    {
+      "epoch": 0.36054524784095504,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.943664900520804e-06,
+      "loss": 17.6029,
+      "step": 19940
+    },
+    {
+      "epoch": 0.3607260629100829,
+      "grad_norm": 40.5,
+      "learning_rate": 9.943636648214144e-06,
+      "loss": 17.2961,
+      "step": 19950
+    },
+    {
+      "epoch": 0.3609068779792108,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.943608395907485e-06,
+      "loss": 17.1464,
+      "step": 19960
+    },
+    {
+      "epoch": 0.36108769304833865,
+      "grad_norm": 41.625,
+      "learning_rate": 9.943580143600825e-06,
+      "loss": 17.2045,
+      "step": 19970
+    },
+    {
+      "epoch": 0.3612685081174665,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.943551891294166e-06,
+      "loss": 17.0232,
+      "step": 19980
+    },
+    {
+      "epoch": 0.3614493231865944,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.943523638987505e-06,
+      "loss": 16.7419,
+      "step": 19990
+    },
+    {
+      "epoch": 0.3616301382557222,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.943495386680846e-06,
+      "loss": 17.3945,
+      "step": 20000
+    },
+    {
+      "epoch": 0.3616301382557222,
+      "eval_loss": 2.1501667499542236,
+      "eval_runtime": 228.7367,
+      "eval_samples_per_second": 3174.213,
+      "eval_steps_per_second": 49.599,
+      "step": 20000
+    },
+    {
+      "epoch": 0.3618109533248501,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.943467134374188e-06,
+      "loss": 17.2655,
+      "step": 20010
+    },
+    {
+      "epoch": 0.36199176839397795,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.943438882067527e-06,
+      "loss": 17.4808,
+      "step": 20020
+    },
+    {
+      "epoch": 0.3621725834631058,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.943410629760868e-06,
+      "loss": 17.0917,
+      "step": 20030
+    },
+    {
+      "epoch": 0.3623533985322337,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.943382377454208e-06,
+      "loss": 16.8377,
+      "step": 20040
+    },
+    {
+      "epoch": 0.3625342136013615,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.943354125147549e-06,
+      "loss": 17.6013,
+      "step": 20050
+    },
+    {
+      "epoch": 0.3627150286704894,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.94332587284089e-06,
+      "loss": 17.57,
+      "step": 20060
+    },
+    {
+      "epoch": 0.36289584373961725,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.943297620534228e-06,
+      "loss": 17.0332,
+      "step": 20070
+    },
+    {
+      "epoch": 0.3630766588087451,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.943269368227569e-06,
+      "loss": 17.2701,
+      "step": 20080
+    },
+    {
+      "epoch": 0.363257473877873,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.94324111592091e-06,
+      "loss": 17.0229,
+      "step": 20090
+    },
+    {
+      "epoch": 0.36343828894700086,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.943212863614252e-06,
+      "loss": 17.3704,
+      "step": 20100
+    },
+    {
+      "epoch": 0.3636191040161287,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.94318461130759e-06,
+      "loss": 16.8609,
+      "step": 20110
+    },
+    {
+      "epoch": 0.36379991908525655,
+      "grad_norm": 43.625,
+      "learning_rate": 9.943156359000931e-06,
+      "loss": 17.259,
+      "step": 20120
+    },
+    {
+      "epoch": 0.3639807341543844,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.943128106694272e-06,
+      "loss": 17.0671,
+      "step": 20130
+    },
+    {
+      "epoch": 0.3641615492235123,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.943099854387613e-06,
+      "loss": 17.6245,
+      "step": 20140
+    },
+    {
+      "epoch": 0.36434236429264016,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.943071602080953e-06,
+      "loss": 17.1875,
+      "step": 20150
+    },
+    {
+      "epoch": 0.36452317936176803,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.943043349774292e-06,
+      "loss": 17.0023,
+      "step": 20160
+    },
+    {
+      "epoch": 0.36470399443089585,
+      "grad_norm": 40.0,
+      "learning_rate": 9.943015097467633e-06,
+      "loss": 16.8232,
+      "step": 20170
+    },
+    {
+      "epoch": 0.3648848095000237,
+      "grad_norm": 38.3125,
+      "learning_rate": 9.942986845160973e-06,
+      "loss": 17.224,
+      "step": 20180
+    },
+    {
+      "epoch": 0.3650656245691516,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.942958592854314e-06,
+      "loss": 17.2585,
+      "step": 20190
+    },
+    {
+      "epoch": 0.36524643963827946,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.942930340547655e-06,
+      "loss": 17.0163,
+      "step": 20200
+    },
+    {
+      "epoch": 0.36542725470740733,
+      "grad_norm": 44.5,
+      "learning_rate": 9.942902088240995e-06,
+      "loss": 16.949,
+      "step": 20210
+    },
+    {
+      "epoch": 0.36560806977653515,
+      "grad_norm": 37.96875,
+      "learning_rate": 9.942873835934336e-06,
+      "loss": 17.4766,
+      "step": 20220
+    },
+    {
+      "epoch": 0.365788884845663,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.942845583627677e-06,
+      "loss": 17.3368,
+      "step": 20230
+    },
+    {
+      "epoch": 0.3659696999147909,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.942817331321016e-06,
+      "loss": 16.8237,
+      "step": 20240
+    },
+    {
+      "epoch": 0.36615051498391876,
+      "grad_norm": 40.125,
+      "learning_rate": 9.942789079014356e-06,
+      "loss": 17.1638,
+      "step": 20250
+    },
+    {
+      "epoch": 0.36633133005304663,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.942760826707697e-06,
+      "loss": 17.4119,
+      "step": 20260
+    },
+    {
+      "epoch": 0.3665121451221745,
+      "grad_norm": 39.5,
+      "learning_rate": 9.942732574401039e-06,
+      "loss": 16.9666,
+      "step": 20270
+    },
+    {
+      "epoch": 0.3666929601913023,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.942704322094378e-06,
+      "loss": 17.1098,
+      "step": 20280
+    },
+    {
+      "epoch": 0.3668737752604302,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.942676069787719e-06,
+      "loss": 17.1769,
+      "step": 20290
+    },
+    {
+      "epoch": 0.36705459032955806,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.94264781748106e-06,
+      "loss": 16.8182,
+      "step": 20300
+    },
+    {
+      "epoch": 0.36723540539868593,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.9426195651744e-06,
+      "loss": 17.4049,
+      "step": 20310
+    },
+    {
+      "epoch": 0.3674162204678138,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.94259131286774e-06,
+      "loss": 17.2833,
+      "step": 20320
+    },
+    {
+      "epoch": 0.3675970355369417,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.94256306056108e-06,
+      "loss": 17.001,
+      "step": 20330
+    },
+    {
+      "epoch": 0.3677778506060695,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.94253480825442e-06,
+      "loss": 17.0131,
+      "step": 20340
+    },
+    {
+      "epoch": 0.36795866567519736,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.94250655594776e-06,
+      "loss": 16.411,
+      "step": 20350
+    },
+    {
+      "epoch": 0.36813948074432523,
+      "grad_norm": 41.875,
+      "learning_rate": 9.942478303641101e-06,
+      "loss": 17.2263,
+      "step": 20360
+    },
+    {
+      "epoch": 0.3683202958134531,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.942450051334442e-06,
+      "loss": 16.7313,
+      "step": 20370
+    },
+    {
+      "epoch": 0.368501110882581,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.942421799027783e-06,
+      "loss": 16.9421,
+      "step": 20380
+    },
+    {
+      "epoch": 0.3686819259517088,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.942393546721123e-06,
+      "loss": 17.3382,
+      "step": 20390
+    },
+    {
+      "epoch": 0.36886274102083666,
+      "grad_norm": 42.75,
+      "learning_rate": 9.942365294414464e-06,
+      "loss": 17.1704,
+      "step": 20400
+    },
+    {
+      "epoch": 0.36904355608996453,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.942337042107804e-06,
+      "loss": 17.117,
+      "step": 20410
+    },
+    {
+      "epoch": 0.3692243711590924,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.942308789801143e-06,
+      "loss": 16.4973,
+      "step": 20420
+    },
+    {
+      "epoch": 0.3694051862282203,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.942280537494484e-06,
+      "loss": 17.1972,
+      "step": 20430
+    },
+    {
+      "epoch": 0.36958600129734814,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.942252285187825e-06,
+      "loss": 17.1071,
+      "step": 20440
+    },
+    {
+      "epoch": 0.36976681636647596,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.942224032881165e-06,
+      "loss": 16.8985,
+      "step": 20450
+    },
+    {
+      "epoch": 0.36994763143560383,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.942195780574506e-06,
+      "loss": 17.5709,
+      "step": 20460
+    },
+    {
+      "epoch": 0.3701284465047317,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.942167528267846e-06,
+      "loss": 17.0988,
+      "step": 20470
+    },
+    {
+      "epoch": 0.37030926157385957,
+      "grad_norm": 44.625,
+      "learning_rate": 9.942139275961187e-06,
+      "loss": 17.2837,
+      "step": 20480
+    },
+    {
+      "epoch": 0.37049007664298744,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.942111023654528e-06,
+      "loss": 17.2057,
+      "step": 20490
+    },
+    {
+      "epoch": 0.3706708917121153,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.942082771347867e-06,
+      "loss": 16.946,
+      "step": 20500
+    },
+    {
+      "epoch": 0.37085170678124313,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.942054519041207e-06,
+      "loss": 17.1524,
+      "step": 20510
+    },
+    {
+      "epoch": 0.371032521850371,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.942026266734548e-06,
+      "loss": 16.8513,
+      "step": 20520
+    },
+    {
+      "epoch": 0.37121333691949887,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.941998014427888e-06,
+      "loss": 16.7765,
+      "step": 20530
+    },
+    {
+      "epoch": 0.37139415198862674,
+      "grad_norm": 36.78125,
+      "learning_rate": 9.941969762121229e-06,
+      "loss": 17.0085,
+      "step": 20540
+    },
+    {
+      "epoch": 0.3715749670577546,
+      "grad_norm": 37.75,
+      "learning_rate": 9.94194150981457e-06,
+      "loss": 16.9942,
+      "step": 20550
+    },
+    {
+      "epoch": 0.37175578212688243,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.94191325750791e-06,
+      "loss": 17.1815,
+      "step": 20560
+    },
+    {
+      "epoch": 0.3719365971960103,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.941885005201251e-06,
+      "loss": 16.9027,
+      "step": 20570
+    },
+    {
+      "epoch": 0.37211741226513817,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.941856752894592e-06,
+      "loss": 17.4756,
+      "step": 20580
+    },
+    {
+      "epoch": 0.37229822733426604,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.94182850058793e-06,
+      "loss": 17.3078,
+      "step": 20590
+    },
+    {
+      "epoch": 0.3724790424033939,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.941800248281271e-06,
+      "loss": 16.9457,
+      "step": 20600
+    },
+    {
+      "epoch": 0.3726598574725218,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.941771995974612e-06,
+      "loss": 16.969,
+      "step": 20610
+    },
+    {
+      "epoch": 0.3728406725416496,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.941743743667952e-06,
+      "loss": 16.8207,
+      "step": 20620
+    },
+    {
+      "epoch": 0.37302148761077747,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.941715491361293e-06,
+      "loss": 16.9767,
+      "step": 20630
+    },
+    {
+      "epoch": 0.37320230267990534,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.941687239054634e-06,
+      "loss": 16.9558,
+      "step": 20640
+    },
+    {
+      "epoch": 0.3733831177490332,
+      "grad_norm": 38.375,
+      "learning_rate": 9.941658986747974e-06,
+      "loss": 16.9039,
+      "step": 20650
+    },
+    {
+      "epoch": 0.3735639328181611,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.941630734441315e-06,
+      "loss": 17.105,
+      "step": 20660
+    },
+    {
+      "epoch": 0.37374474788728895,
+      "grad_norm": 41.0,
+      "learning_rate": 9.941602482134654e-06,
+      "loss": 16.8125,
+      "step": 20670
+    },
+    {
+      "epoch": 0.37392556295641677,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.941574229827994e-06,
+      "loss": 17.5724,
+      "step": 20680
+    },
+    {
+      "epoch": 0.37410637802554464,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.941545977521335e-06,
+      "loss": 17.3323,
+      "step": 20690
+    },
+    {
+      "epoch": 0.3742871930946725,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.941517725214676e-06,
+      "loss": 17.2595,
+      "step": 20700
+    },
+    {
+      "epoch": 0.3744680081638004,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.941489472908016e-06,
+      "loss": 16.9378,
+      "step": 20710
+    },
+    {
+      "epoch": 0.37464882323292825,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.941461220601357e-06,
+      "loss": 17.1729,
+      "step": 20720
+    },
+    {
+      "epoch": 0.37482963830205607,
+      "grad_norm": 38.5625,
+      "learning_rate": 9.941432968294698e-06,
+      "loss": 17.1281,
+      "step": 20730
+    },
+    {
+      "epoch": 0.37501045337118394,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.941404715988038e-06,
+      "loss": 17.3943,
+      "step": 20740
+    },
+    {
+      "epoch": 0.3751912684403118,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.941376463681379e-06,
+      "loss": 17.0422,
+      "step": 20750
+    },
+    {
+      "epoch": 0.3753720835094397,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.941348211374718e-06,
+      "loss": 17.6543,
+      "step": 20760
+    },
+    {
+      "epoch": 0.37555289857856755,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.941319959068058e-06,
+      "loss": 17.1982,
+      "step": 20770
+    },
+    {
+      "epoch": 0.3757337136476954,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.941291706761399e-06,
+      "loss": 17.2206,
+      "step": 20780
+    },
+    {
+      "epoch": 0.37591452871682324,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.94126345445474e-06,
+      "loss": 16.8636,
+      "step": 20790
+    },
+    {
+      "epoch": 0.3760953437859511,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.94123520214808e-06,
+      "loss": 16.6194,
+      "step": 20800
+    },
+    {
+      "epoch": 0.376276158855079,
+      "grad_norm": 42.25,
+      "learning_rate": 9.94120694984142e-06,
+      "loss": 17.0831,
+      "step": 20810
+    },
+    {
+      "epoch": 0.37645697392420685,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.941178697534761e-06,
+      "loss": 17.2419,
+      "step": 20820
+    },
+    {
+      "epoch": 0.3766377889933347,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.941150445228102e-06,
+      "loss": 17.0365,
+      "step": 20830
+    },
+    {
+      "epoch": 0.3768186040624626,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.941122192921443e-06,
+      "loss": 17.113,
+      "step": 20840
+    },
+    {
+      "epoch": 0.3769994191315904,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.941093940614782e-06,
+      "loss": 17.5729,
+      "step": 20850
+    },
+    {
+      "epoch": 0.3771802342007183,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.941065688308122e-06,
+      "loss": 17.2333,
+      "step": 20860
+    },
+    {
+      "epoch": 0.37736104926984615,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.941037436001463e-06,
+      "loss": 17.4125,
+      "step": 20870
+    },
+    {
+      "epoch": 0.377541864338974,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.941009183694803e-06,
+      "loss": 17.2689,
+      "step": 20880
+    },
+    {
+      "epoch": 0.3777226794081019,
+      "grad_norm": 41.375,
+      "learning_rate": 9.940980931388144e-06,
+      "loss": 17.0753,
+      "step": 20890
+    },
+    {
+      "epoch": 0.3779034944772297,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.940952679081485e-06,
+      "loss": 17.0969,
+      "step": 20900
+    },
+    {
+      "epoch": 0.3780843095463576,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.940924426774825e-06,
+      "loss": 16.9405,
+      "step": 20910
+    },
+    {
+      "epoch": 0.37826512461548545,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.940896174468166e-06,
+      "loss": 16.7523,
+      "step": 20920
+    },
+    {
+      "epoch": 0.3784459396846133,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.940867922161505e-06,
+      "loss": 16.8263,
+      "step": 20930
+    },
+    {
+      "epoch": 0.3786267547537412,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.940839669854846e-06,
+      "loss": 17.2799,
+      "step": 20940
+    },
+    {
+      "epoch": 0.37880756982286906,
+      "grad_norm": 40.25,
+      "learning_rate": 9.940811417548186e-06,
+      "loss": 16.8069,
+      "step": 20950
+    },
+    {
+      "epoch": 0.3789883848919969,
+      "grad_norm": 40.875,
+      "learning_rate": 9.940783165241527e-06,
+      "loss": 17.2283,
+      "step": 20960
+    },
+    {
+      "epoch": 0.37916919996112475,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.940754912934867e-06,
+      "loss": 16.9905,
+      "step": 20970
+    },
+    {
+      "epoch": 0.3793500150302526,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.940726660628208e-06,
+      "loss": 17.0025,
+      "step": 20980
+    },
+    {
+      "epoch": 0.3795308300993805,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.940698408321549e-06,
+      "loss": 17.2225,
+      "step": 20990
+    },
+    {
+      "epoch": 0.37971164516850836,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.94067015601489e-06,
+      "loss": 17.4368,
+      "step": 21000
+    },
+    {
+      "epoch": 0.37989246023763623,
+      "grad_norm": 40.375,
+      "learning_rate": 9.94064190370823e-06,
+      "loss": 17.1831,
+      "step": 21010
+    },
+    {
+      "epoch": 0.38007327530676405,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.940613651401569e-06,
+      "loss": 17.1583,
+      "step": 21020
+    },
+    {
+      "epoch": 0.3802540903758919,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.94058539909491e-06,
+      "loss": 17.0701,
+      "step": 21030
+    },
+    {
+      "epoch": 0.3804349054450198,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.94055714678825e-06,
+      "loss": 17.1009,
+      "step": 21040
+    },
+    {
+      "epoch": 0.38061572051414766,
+      "grad_norm": 40.5,
+      "learning_rate": 9.94052889448159e-06,
+      "loss": 17.0964,
+      "step": 21050
+    },
+    {
+      "epoch": 0.38079653558327553,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.940500642174931e-06,
+      "loss": 17.2598,
+      "step": 21060
+    },
+    {
+      "epoch": 0.38097735065240335,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.940472389868272e-06,
+      "loss": 17.3123,
+      "step": 21070
+    },
+    {
+      "epoch": 0.3811581657215312,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.940444137561613e-06,
+      "loss": 17.112,
+      "step": 21080
+    },
+    {
+      "epoch": 0.3813389807906591,
+      "grad_norm": 41.5,
+      "learning_rate": 9.940415885254953e-06,
+      "loss": 17.1988,
+      "step": 21090
+    },
+    {
+      "epoch": 0.38151979585978696,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.940387632948292e-06,
+      "loss": 17.2093,
+      "step": 21100
+    },
+    {
+      "epoch": 0.38170061092891483,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.940359380641633e-06,
+      "loss": 17.2864,
+      "step": 21110
+    },
+    {
+      "epoch": 0.3818814259980427,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.940331128334973e-06,
+      "loss": 16.7385,
+      "step": 21120
+    },
+    {
+      "epoch": 0.3820622410671705,
+      "grad_norm": 40.0,
+      "learning_rate": 9.940302876028314e-06,
+      "loss": 17.4242,
+      "step": 21130
+    },
+    {
+      "epoch": 0.3822430561362984,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.940274623721655e-06,
+      "loss": 17.2833,
+      "step": 21140
+    },
+    {
+      "epoch": 0.38242387120542626,
+      "grad_norm": 42.0,
+      "learning_rate": 9.940246371414995e-06,
+      "loss": 17.4265,
+      "step": 21150
+    },
+    {
+      "epoch": 0.38260468627455413,
+      "grad_norm": 43.75,
+      "learning_rate": 9.940218119108336e-06,
+      "loss": 17.0023,
+      "step": 21160
+    },
+    {
+      "epoch": 0.382785501343682,
+      "grad_norm": 38.40625,
+      "learning_rate": 9.940189866801676e-06,
+      "loss": 17.121,
+      "step": 21170
+    },
+    {
+      "epoch": 0.38296631641280987,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.940161614495017e-06,
+      "loss": 17.3334,
+      "step": 21180
+    },
+    {
+      "epoch": 0.3831471314819377,
+      "grad_norm": 44.5,
+      "learning_rate": 9.940133362188356e-06,
+      "loss": 16.8784,
+      "step": 21190
+    },
+    {
+      "epoch": 0.38332794655106556,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.940105109881697e-06,
+      "loss": 17.4054,
+      "step": 21200
+    },
+    {
+      "epoch": 0.38350876162019343,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.940076857575037e-06,
+      "loss": 17.2301,
+      "step": 21210
+    },
+    {
+      "epoch": 0.3836895766893213,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.940048605268378e-06,
+      "loss": 17.1037,
+      "step": 21220
+    },
+    {
+      "epoch": 0.38387039175844917,
+      "grad_norm": 40.375,
+      "learning_rate": 9.940020352961719e-06,
+      "loss": 17.263,
+      "step": 21230
+    },
+    {
+      "epoch": 0.384051206827577,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.939992100655059e-06,
+      "loss": 17.4868,
+      "step": 21240
+    },
+    {
+      "epoch": 0.38423202189670486,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.9399638483484e-06,
+      "loss": 17.1353,
+      "step": 21250
+    },
+    {
+      "epoch": 0.3844128369658327,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.93993559604174e-06,
+      "loss": 17.1201,
+      "step": 21260
+    },
+    {
+      "epoch": 0.3845936520349606,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.939907343735081e-06,
+      "loss": 17.1401,
+      "step": 21270
+    },
+    {
+      "epoch": 0.38477446710408847,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.93987909142842e-06,
+      "loss": 17.2789,
+      "step": 21280
+    },
+    {
+      "epoch": 0.38495528217321634,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.93985083912176e-06,
+      "loss": 16.8673,
+      "step": 21290
+    },
+    {
+      "epoch": 0.38513609724234416,
+      "grad_norm": 41.375,
+      "learning_rate": 9.939822586815101e-06,
+      "loss": 16.7684,
+      "step": 21300
+    },
+    {
+      "epoch": 0.385316912311472,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.939794334508442e-06,
+      "loss": 17.5884,
+      "step": 21310
+    },
+    {
+      "epoch": 0.3854977273805999,
+      "grad_norm": 44.71875,
+      "learning_rate": 9.939766082201782e-06,
+      "loss": 17.2755,
+      "step": 21320
+    },
+    {
+      "epoch": 0.38567854244972777,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.939737829895123e-06,
+      "loss": 17.0057,
+      "step": 21330
+    },
+    {
+      "epoch": 0.38585935751885564,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.939709577588464e-06,
+      "loss": 17.2807,
+      "step": 21340
+    },
+    {
+      "epoch": 0.3860401725879835,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.939681325281804e-06,
+      "loss": 17.0537,
+      "step": 21350
+    },
+    {
+      "epoch": 0.3862209876571113,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.939653072975143e-06,
+      "loss": 16.9537,
+      "step": 21360
+    },
+    {
+      "epoch": 0.3864018027262392,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.939624820668484e-06,
+      "loss": 16.8987,
+      "step": 21370
+    },
+    {
+      "epoch": 0.38658261779536707,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.939596568361824e-06,
+      "loss": 17.2109,
+      "step": 21380
+    },
+    {
+      "epoch": 0.38676343286449494,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.939568316055165e-06,
+      "loss": 17.2425,
+      "step": 21390
+    },
+    {
+      "epoch": 0.3869442479336228,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.939540063748506e-06,
+      "loss": 17.6312,
+      "step": 21400
+    },
+    {
+      "epoch": 0.3871250630027506,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.939511811441846e-06,
+      "loss": 16.8594,
+      "step": 21410
+    },
+    {
+      "epoch": 0.3873058780718785,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.939483559135187e-06,
+      "loss": 17.2082,
+      "step": 21420
+    },
+    {
+      "epoch": 0.38748669314100637,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.939455306828528e-06,
+      "loss": 16.9624,
+      "step": 21430
+    },
+    {
+      "epoch": 0.38766750821013424,
+      "grad_norm": 41.625,
+      "learning_rate": 9.939427054521868e-06,
+      "loss": 16.6994,
+      "step": 21440
+    },
+    {
+      "epoch": 0.3878483232792621,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.939398802215207e-06,
+      "loss": 17.3257,
+      "step": 21450
+    },
+    {
+      "epoch": 0.38802913834839,
+      "grad_norm": 43.0,
+      "learning_rate": 9.939370549908548e-06,
+      "loss": 17.1406,
+      "step": 21460
+    },
+    {
+      "epoch": 0.3882099534175178,
+      "grad_norm": 37.6875,
+      "learning_rate": 9.939342297601888e-06,
+      "loss": 17.3857,
+      "step": 21470
+    },
+    {
+      "epoch": 0.38839076848664567,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.939314045295229e-06,
+      "loss": 16.8819,
+      "step": 21480
+    },
+    {
+      "epoch": 0.38857158355577354,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.93928579298857e-06,
+      "loss": 17.316,
+      "step": 21490
+    },
+    {
+      "epoch": 0.3887523986249014,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.93925754068191e-06,
+      "loss": 17.0337,
+      "step": 21500
+    },
+    {
+      "epoch": 0.3889332136940293,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.93922928837525e-06,
+      "loss": 17.1861,
+      "step": 21510
+    },
+    {
+      "epoch": 0.3891140287631571,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.939201036068591e-06,
+      "loss": 16.8454,
+      "step": 21520
+    },
+    {
+      "epoch": 0.38929484383228496,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.93917278376193e-06,
+      "loss": 17.6789,
+      "step": 21530
+    },
+    {
+      "epoch": 0.38947565890141284,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.939144531455271e-06,
+      "loss": 16.8952,
+      "step": 21540
+    },
+    {
+      "epoch": 0.3896564739705407,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.939116279148612e-06,
+      "loss": 17.1093,
+      "step": 21550
+    },
+    {
+      "epoch": 0.3898372890396686,
+      "grad_norm": 42.25,
+      "learning_rate": 9.939088026841952e-06,
+      "loss": 17.1814,
+      "step": 21560
+    },
+    {
+      "epoch": 0.39001810410879645,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.939059774535293e-06,
+      "loss": 17.0269,
+      "step": 21570
+    },
+    {
+      "epoch": 0.39019891917792426,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.939031522228632e-06,
+      "loss": 17.3362,
+      "step": 21580
+    },
+    {
+      "epoch": 0.39037973424705213,
+      "grad_norm": 40.125,
+      "learning_rate": 9.939003269921974e-06,
+      "loss": 16.7692,
+      "step": 21590
+    },
+    {
+      "epoch": 0.39056054931618,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.938975017615315e-06,
+      "loss": 17.1616,
+      "step": 21600
+    },
+    {
+      "epoch": 0.3907413643853079,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.938946765308655e-06,
+      "loss": 17.3161,
+      "step": 21610
+    },
+    {
+      "epoch": 0.39092217945443575,
+      "grad_norm": 39.75,
+      "learning_rate": 9.938918513001994e-06,
+      "loss": 17.4791,
+      "step": 21620
+    },
+    {
+      "epoch": 0.3911029945235636,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.938890260695335e-06,
+      "loss": 17.2422,
+      "step": 21630
+    },
+    {
+      "epoch": 0.39128380959269143,
+      "grad_norm": 42.125,
+      "learning_rate": 9.938862008388676e-06,
+      "loss": 17.1619,
+      "step": 21640
+    },
+    {
+      "epoch": 0.3914646246618193,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.938833756082016e-06,
+      "loss": 17.3982,
+      "step": 21650
+    },
+    {
+      "epoch": 0.3916454397309472,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.938805503775357e-06,
+      "loss": 17.2333,
+      "step": 21660
+    },
+    {
+      "epoch": 0.39182625480007505,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.938777251468696e-06,
+      "loss": 17.1,
+      "step": 21670
+    },
+    {
+      "epoch": 0.3920070698692029,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.938748999162038e-06,
+      "loss": 17.1815,
+      "step": 21680
+    },
+    {
+      "epoch": 0.39218788493833073,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.938720746855379e-06,
+      "loss": 16.9364,
+      "step": 21690
+    },
+    {
+      "epoch": 0.3923687000074586,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.938692494548718e-06,
+      "loss": 17.2628,
+      "step": 21700
+    },
+    {
+      "epoch": 0.3925495150765865,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.938664242242058e-06,
+      "loss": 17.1232,
+      "step": 21710
+    },
+    {
+      "epoch": 0.39273033014571435,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.938635989935399e-06,
+      "loss": 16.7432,
+      "step": 21720
+    },
+    {
+      "epoch": 0.3929111452148422,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.93860773762874e-06,
+      "loss": 17.1859,
+      "step": 21730
+    },
+    {
+      "epoch": 0.3930919602839701,
+      "grad_norm": 43.5,
+      "learning_rate": 9.93857948532208e-06,
+      "loss": 17.0688,
+      "step": 21740
+    },
+    {
+      "epoch": 0.3932727753530979,
+      "grad_norm": 41.875,
+      "learning_rate": 9.93855123301542e-06,
+      "loss": 17.1453,
+      "step": 21750
+    },
+    {
+      "epoch": 0.3934535904222258,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.938522980708761e-06,
+      "loss": 17.4998,
+      "step": 21760
+    },
+    {
+      "epoch": 0.39363440549135364,
+      "grad_norm": 44.71875,
+      "learning_rate": 9.938494728402102e-06,
+      "loss": 17.1017,
+      "step": 21770
+    },
+    {
+      "epoch": 0.3938152205604815,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.938466476095443e-06,
+      "loss": 16.8154,
+      "step": 21780
+    },
+    {
+      "epoch": 0.3939960356296094,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.938438223788781e-06,
+      "loss": 17.0335,
+      "step": 21790
+    },
+    {
+      "epoch": 0.39417685069873726,
+      "grad_norm": 41.375,
+      "learning_rate": 9.938409971482122e-06,
+      "loss": 17.3227,
+      "step": 21800
+    },
+    {
+      "epoch": 0.3943576657678651,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.938381719175463e-06,
+      "loss": 16.984,
+      "step": 21810
+    },
+    {
+      "epoch": 0.39453848083699294,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.938353466868803e-06,
+      "loss": 17.2131,
+      "step": 21820
+    },
+    {
+      "epoch": 0.3947192959061208,
+      "grad_norm": 42.375,
+      "learning_rate": 9.938325214562144e-06,
+      "loss": 17.0346,
+      "step": 21830
+    },
+    {
+      "epoch": 0.3949001109752487,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.938296962255483e-06,
+      "loss": 17.0261,
+      "step": 21840
+    },
+    {
+      "epoch": 0.39508092604437656,
+      "grad_norm": 42.0,
+      "learning_rate": 9.938268709948825e-06,
+      "loss": 17.3454,
+      "step": 21850
+    },
+    {
+      "epoch": 0.3952617411135044,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.938240457642166e-06,
+      "loss": 17.0362,
+      "step": 21860
+    },
+    {
+      "epoch": 0.39544255618263224,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.938212205335506e-06,
+      "loss": 17.1867,
+      "step": 21870
+    },
+    {
+      "epoch": 0.3956233712517601,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.938183953028845e-06,
+      "loss": 17.3699,
+      "step": 21880
+    },
+    {
+      "epoch": 0.395804186320888,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.938155700722186e-06,
+      "loss": 17.1582,
+      "step": 21890
+    },
+    {
+      "epoch": 0.39598500139001586,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.938127448415527e-06,
+      "loss": 17.2781,
+      "step": 21900
+    },
+    {
+      "epoch": 0.3961658164591437,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.938099196108867e-06,
+      "loss": 17.2928,
+      "step": 21910
+    },
+    {
+      "epoch": 0.39634663152827154,
+      "grad_norm": 44.125,
+      "learning_rate": 9.938070943802208e-06,
+      "loss": 17.1416,
+      "step": 21920
+    },
+    {
+      "epoch": 0.3965274465973994,
+      "grad_norm": 40.25,
+      "learning_rate": 9.938042691495547e-06,
+      "loss": 17.1793,
+      "step": 21930
+    },
+    {
+      "epoch": 0.3967082616665273,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.938014439188889e-06,
+      "loss": 17.5213,
+      "step": 21940
+    },
+    {
+      "epoch": 0.39688907673565516,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.93798618688223e-06,
+      "loss": 17.0786,
+      "step": 21950
+    },
+    {
+      "epoch": 0.397069891804783,
+      "grad_norm": 40.625,
+      "learning_rate": 9.937957934575569e-06,
+      "loss": 17.1238,
+      "step": 21960
+    },
+    {
+      "epoch": 0.3972507068739109,
+      "grad_norm": 40.5,
+      "learning_rate": 9.93792968226891e-06,
+      "loss": 17.5342,
+      "step": 21970
+    },
+    {
+      "epoch": 0.3974315219430387,
+      "grad_norm": 43.0,
+      "learning_rate": 9.93790142996225e-06,
+      "loss": 17.2814,
+      "step": 21980
+    },
+    {
+      "epoch": 0.3976123370121666,
+      "grad_norm": 40.75,
+      "learning_rate": 9.93787317765559e-06,
+      "loss": 17.0391,
+      "step": 21990
+    },
+    {
+      "epoch": 0.39779315208129445,
+      "grad_norm": 42.125,
+      "learning_rate": 9.937844925348931e-06,
+      "loss": 17.3825,
+      "step": 22000
+    },
+    {
+      "epoch": 0.3979739671504223,
+      "grad_norm": 40.875,
+      "learning_rate": 9.93781667304227e-06,
+      "loss": 17.0485,
+      "step": 22010
+    },
+    {
+      "epoch": 0.3981547822195502,
+      "grad_norm": 37.71875,
+      "learning_rate": 9.93778842073561e-06,
+      "loss": 16.8926,
+      "step": 22020
+    },
+    {
+      "epoch": 0.398335597288678,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.937760168428953e-06,
+      "loss": 17.0525,
+      "step": 22030
+    },
+    {
+      "epoch": 0.3985164123578059,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.937731916122294e-06,
+      "loss": 16.8813,
+      "step": 22040
+    },
+    {
+      "epoch": 0.39869722742693375,
+      "grad_norm": 41.5,
+      "learning_rate": 9.937703663815633e-06,
+      "loss": 16.917,
+      "step": 22050
+    },
+    {
+      "epoch": 0.3988780424960616,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.937675411508973e-06,
+      "loss": 17.3316,
+      "step": 22060
+    },
+    {
+      "epoch": 0.3990588575651895,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.937647159202314e-06,
+      "loss": 17.4144,
+      "step": 22070
+    },
+    {
+      "epoch": 0.39923967263431737,
+      "grad_norm": 42.25,
+      "learning_rate": 9.937618906895654e-06,
+      "loss": 16.9716,
+      "step": 22080
+    },
+    {
+      "epoch": 0.3994204877034452,
+      "grad_norm": 38.46875,
+      "learning_rate": 9.937590654588995e-06,
+      "loss": 17.2706,
+      "step": 22090
+    },
+    {
+      "epoch": 0.39960130277257305,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.937562402282334e-06,
+      "loss": 17.2526,
+      "step": 22100
+    },
+    {
+      "epoch": 0.3997821178417009,
+      "grad_norm": 38.21875,
+      "learning_rate": 9.937534149975676e-06,
+      "loss": 16.7109,
+      "step": 22110
+    },
+    {
+      "epoch": 0.3999629329108288,
+      "grad_norm": 40.0,
+      "learning_rate": 9.937505897669017e-06,
+      "loss": 16.9083,
+      "step": 22120
+    },
+    {
+      "epoch": 0.40014374797995667,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.937477645362356e-06,
+      "loss": 17.1417,
+      "step": 22130
+    },
+    {
+      "epoch": 0.40032456304908454,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.937449393055696e-06,
+      "loss": 16.9667,
+      "step": 22140
+    },
+    {
+      "epoch": 0.40050537811821235,
+      "grad_norm": 41.375,
+      "learning_rate": 9.937421140749037e-06,
+      "loss": 16.957,
+      "step": 22150
+    },
+    {
+      "epoch": 0.4006861931873402,
+      "grad_norm": 39.625,
+      "learning_rate": 9.937392888442378e-06,
+      "loss": 17.6841,
+      "step": 22160
+    },
+    {
+      "epoch": 0.4008670082564681,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.937364636135718e-06,
+      "loss": 16.9138,
+      "step": 22170
+    },
+    {
+      "epoch": 0.40104782332559596,
+      "grad_norm": 40.625,
+      "learning_rate": 9.937336383829059e-06,
+      "loss": 17.0477,
+      "step": 22180
+    },
+    {
+      "epoch": 0.40122863839472384,
+      "grad_norm": 41.625,
+      "learning_rate": 9.937308131522398e-06,
+      "loss": 17.344,
+      "step": 22190
+    },
+    {
+      "epoch": 0.40140945346385165,
+      "grad_norm": 41.125,
+      "learning_rate": 9.93727987921574e-06,
+      "loss": 16.9657,
+      "step": 22200
+    },
+    {
+      "epoch": 0.4015902685329795,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.937251626909081e-06,
+      "loss": 16.5823,
+      "step": 22210
+    },
+    {
+      "epoch": 0.4017710836021074,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.93722337460242e-06,
+      "loss": 17.1726,
+      "step": 22220
+    },
+    {
+      "epoch": 0.40195189867123526,
+      "grad_norm": 41.375,
+      "learning_rate": 9.93719512229576e-06,
+      "loss": 17.4138,
+      "step": 22230
+    },
+    {
+      "epoch": 0.40213271374036313,
+      "grad_norm": 39.0,
+      "learning_rate": 9.937166869989101e-06,
+      "loss": 17.2267,
+      "step": 22240
+    },
+    {
+      "epoch": 0.402313528809491,
+      "grad_norm": 40.0,
+      "learning_rate": 9.937138617682442e-06,
+      "loss": 17.1238,
+      "step": 22250
+    },
+    {
+      "epoch": 0.4024943438786188,
+      "grad_norm": 46.125,
+      "learning_rate": 9.937110365375782e-06,
+      "loss": 16.9259,
+      "step": 22260
+    },
+    {
+      "epoch": 0.4026751589477467,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.937082113069121e-06,
+      "loss": 17.2861,
+      "step": 22270
+    },
+    {
+      "epoch": 0.40285597401687456,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.937053860762462e-06,
+      "loss": 17.508,
+      "step": 22280
+    },
+    {
+      "epoch": 0.40303678908600243,
+      "grad_norm": 44.125,
+      "learning_rate": 9.937025608455804e-06,
+      "loss": 16.9261,
+      "step": 22290
+    },
+    {
+      "epoch": 0.4032176041551303,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.936997356149145e-06,
+      "loss": 17.3141,
+      "step": 22300
+    },
+    {
+      "epoch": 0.4033984192242582,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.936969103842484e-06,
+      "loss": 17.0033,
+      "step": 22310
+    },
+    {
+      "epoch": 0.403579234293386,
+      "grad_norm": 38.65625,
+      "learning_rate": 9.936940851535824e-06,
+      "loss": 17.1212,
+      "step": 22320
+    },
+    {
+      "epoch": 0.40376004936251386,
+      "grad_norm": 40.5,
+      "learning_rate": 9.936912599229165e-06,
+      "loss": 16.7634,
+      "step": 22330
+    },
+    {
+      "epoch": 0.40394086443164173,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.936884346922506e-06,
+      "loss": 17.2485,
+      "step": 22340
+    },
+    {
+      "epoch": 0.4041216795007696,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.936856094615846e-06,
+      "loss": 17.0581,
+      "step": 22350
+    },
+    {
+      "epoch": 0.4043024945698975,
+      "grad_norm": 43.625,
+      "learning_rate": 9.936827842309185e-06,
+      "loss": 17.339,
+      "step": 22360
+    },
+    {
+      "epoch": 0.4044833096390253,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.936799590002526e-06,
+      "loss": 17.0071,
+      "step": 22370
+    },
+    {
+      "epoch": 0.40466412470815316,
+      "grad_norm": 41.5,
+      "learning_rate": 9.936771337695868e-06,
+      "loss": 16.8362,
+      "step": 22380
+    },
+    {
+      "epoch": 0.40484493977728103,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.936743085389207e-06,
+      "loss": 17.1963,
+      "step": 22390
+    },
+    {
+      "epoch": 0.4050257548464089,
+      "grad_norm": 39.75,
+      "learning_rate": 9.936714833082548e-06,
+      "loss": 16.811,
+      "step": 22400
+    },
+    {
+      "epoch": 0.4052065699155368,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.936686580775888e-06,
+      "loss": 17.4384,
+      "step": 22410
+    },
+    {
+      "epoch": 0.40538738498466464,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.936658328469229e-06,
+      "loss": 17.1916,
+      "step": 22420
+    },
+    {
+      "epoch": 0.40556820005379246,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.93663007616257e-06,
+      "loss": 17.2659,
+      "step": 22430
+    },
+    {
+      "epoch": 0.40574901512292033,
+      "grad_norm": 42.25,
+      "learning_rate": 9.936601823855908e-06,
+      "loss": 17.0808,
+      "step": 22440
+    },
+    {
+      "epoch": 0.4059298301920482,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.936573571549249e-06,
+      "loss": 17.25,
+      "step": 22450
+    },
+    {
+      "epoch": 0.4061106452611761,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.936545319242591e-06,
+      "loss": 17.3886,
+      "step": 22460
+    },
+    {
+      "epoch": 0.40629146033030394,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.936517066935932e-06,
+      "loss": 17.0031,
+      "step": 22470
+    },
+    {
+      "epoch": 0.4064722753994318,
+      "grad_norm": 46.25,
+      "learning_rate": 9.936488814629271e-06,
+      "loss": 17.1316,
+      "step": 22480
+    },
+    {
+      "epoch": 0.40665309046855963,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.936460562322612e-06,
+      "loss": 17.3826,
+      "step": 22490
+    },
+    {
+      "epoch": 0.4068339055376875,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.936432310015952e-06,
+      "loss": 16.9754,
+      "step": 22500
+    },
+    {
+      "epoch": 0.40701472060681537,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.936404057709293e-06,
+      "loss": 17.3494,
+      "step": 22510
+    },
+    {
+      "epoch": 0.40719553567594324,
+      "grad_norm": 41.875,
+      "learning_rate": 9.936375805402633e-06,
+      "loss": 16.3543,
+      "step": 22520
+    },
+    {
+      "epoch": 0.4073763507450711,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.936347553095972e-06,
+      "loss": 17.2515,
+      "step": 22530
+    },
+    {
+      "epoch": 0.40755716581419893,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.936319300789313e-06,
+      "loss": 17.1219,
+      "step": 22540
+    },
+    {
+      "epoch": 0.4077379808833268,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.936291048482655e-06,
+      "loss": 17.0418,
+      "step": 22550
+    },
+    {
+      "epoch": 0.40791879595245467,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.936262796175994e-06,
+      "loss": 17.096,
+      "step": 22560
+    },
+    {
+      "epoch": 0.40809961102158254,
+      "grad_norm": 41.25,
+      "learning_rate": 9.936234543869335e-06,
+      "loss": 17.0234,
+      "step": 22570
+    },
+    {
+      "epoch": 0.4082804260907104,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.936206291562675e-06,
+      "loss": 17.0583,
+      "step": 22580
+    },
+    {
+      "epoch": 0.4084612411598383,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.936178039256016e-06,
+      "loss": 17.0878,
+      "step": 22590
+    },
+    {
+      "epoch": 0.4086420562289661,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.936149786949357e-06,
+      "loss": 16.7648,
+      "step": 22600
+    },
+    {
+      "epoch": 0.40882287129809397,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.936121534642697e-06,
+      "loss": 17.1874,
+      "step": 22610
+    },
+    {
+      "epoch": 0.40900368636722184,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.936093282336036e-06,
+      "loss": 17.5475,
+      "step": 22620
+    },
+    {
+      "epoch": 0.4091845014363497,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.936065030029377e-06,
+      "loss": 17.4482,
+      "step": 22630
+    },
+    {
+      "epoch": 0.4093653165054776,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.936036777722719e-06,
+      "loss": 17.099,
+      "step": 22640
+    },
+    {
+      "epoch": 0.40954613157460545,
+      "grad_norm": 41.0,
+      "learning_rate": 9.936008525416058e-06,
+      "loss": 17.336,
+      "step": 22650
+    },
+    {
+      "epoch": 0.40972694664373327,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.935980273109399e-06,
+      "loss": 17.0403,
+      "step": 22660
+    },
+    {
+      "epoch": 0.40990776171286114,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.93595202080274e-06,
+      "loss": 17.0826,
+      "step": 22670
+    },
+    {
+      "epoch": 0.410088576781989,
+      "grad_norm": 41.625,
+      "learning_rate": 9.93592376849608e-06,
+      "loss": 17.2808,
+      "step": 22680
+    },
+    {
+      "epoch": 0.4102693918511169,
+      "grad_norm": 41.25,
+      "learning_rate": 9.93589551618942e-06,
+      "loss": 16.7892,
+      "step": 22690
+    },
+    {
+      "epoch": 0.41045020692024475,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.93586726388276e-06,
+      "loss": 16.9339,
+      "step": 22700
+    },
+    {
+      "epoch": 0.41063102198937257,
+      "grad_norm": 39.125,
+      "learning_rate": 9.9358390115761e-06,
+      "loss": 16.9299,
+      "step": 22710
+    },
+    {
+      "epoch": 0.41081183705850044,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.93581075926944e-06,
+      "loss": 17.1152,
+      "step": 22720
+    },
+    {
+      "epoch": 0.4109926521276283,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.935782506962783e-06,
+      "loss": 17.313,
+      "step": 22730
+    },
+    {
+      "epoch": 0.4111734671967562,
+      "grad_norm": 43.0,
+      "learning_rate": 9.935754254656122e-06,
+      "loss": 17.5031,
+      "step": 22740
+    },
+    {
+      "epoch": 0.41135428226588405,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.935726002349463e-06,
+      "loss": 17.04,
+      "step": 22750
+    },
+    {
+      "epoch": 0.4115350973350119,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.935697750042803e-06,
+      "loss": 17.2801,
+      "step": 22760
+    },
+    {
+      "epoch": 0.41171591240413974,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.935669497736144e-06,
+      "loss": 17.1183,
+      "step": 22770
+    },
+    {
+      "epoch": 0.4118967274732676,
+      "grad_norm": 38.28125,
+      "learning_rate": 9.935641245429484e-06,
+      "loss": 17.0931,
+      "step": 22780
+    },
+    {
+      "epoch": 0.4120775425423955,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.935612993122823e-06,
+      "loss": 17.5523,
+      "step": 22790
+    },
+    {
+      "epoch": 0.41225835761152335,
+      "grad_norm": 42.5,
+      "learning_rate": 9.935584740816164e-06,
+      "loss": 16.6863,
+      "step": 22800
+    },
+    {
+      "epoch": 0.4124391726806512,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.935556488509506e-06,
+      "loss": 16.9507,
+      "step": 22810
+    },
+    {
+      "epoch": 0.4126199877497791,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.935528236202845e-06,
+      "loss": 16.8283,
+      "step": 22820
+    },
+    {
+      "epoch": 0.4128008028189069,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.935499983896186e-06,
+      "loss": 17.189,
+      "step": 22830
+    },
+    {
+      "epoch": 0.4129816178880348,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.935471731589527e-06,
+      "loss": 17.2389,
+      "step": 22840
+    },
+    {
+      "epoch": 0.41316243295716265,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.935443479282867e-06,
+      "loss": 16.7641,
+      "step": 22850
+    },
+    {
+      "epoch": 0.4133432480262905,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.935415226976208e-06,
+      "loss": 17.1253,
+      "step": 22860
+    },
+    {
+      "epoch": 0.4135240630954184,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.935386974669547e-06,
+      "loss": 17.3765,
+      "step": 22870
+    },
+    {
+      "epoch": 0.4137048781645462,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.935358722362887e-06,
+      "loss": 17.4771,
+      "step": 22880
+    },
+    {
+      "epoch": 0.4138856932336741,
+      "grad_norm": 39.625,
+      "learning_rate": 9.935330470056228e-06,
+      "loss": 16.9489,
+      "step": 22890
+    },
+    {
+      "epoch": 0.41406650830280195,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.93530221774957e-06,
+      "loss": 17.4015,
+      "step": 22900
+    },
+    {
+      "epoch": 0.4142473233719298,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.93527396544291e-06,
+      "loss": 17.2297,
+      "step": 22910
+    },
+    {
+      "epoch": 0.4144281384410577,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.93524571313625e-06,
+      "loss": 17.3229,
+      "step": 22920
+    },
+    {
+      "epoch": 0.41460895351018556,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.93521746082959e-06,
+      "loss": 17.0622,
+      "step": 22930
+    },
+    {
+      "epoch": 0.4147897685793134,
+      "grad_norm": 37.96875,
+      "learning_rate": 9.935189208522931e-06,
+      "loss": 17.4614,
+      "step": 22940
+    },
+    {
+      "epoch": 0.41497058364844125,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.935160956216272e-06,
+      "loss": 17.0732,
+      "step": 22950
+    },
+    {
+      "epoch": 0.4151513987175691,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.93513270390961e-06,
+      "loss": 17.2508,
+      "step": 22960
+    },
+    {
+      "epoch": 0.415332213786697,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.935104451602951e-06,
+      "loss": 16.7658,
+      "step": 22970
+    },
+    {
+      "epoch": 0.41551302885582486,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.935076199296292e-06,
+      "loss": 16.942,
+      "step": 22980
+    },
+    {
+      "epoch": 0.41569384392495273,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.935047946989632e-06,
+      "loss": 17.2235,
+      "step": 22990
+    },
+    {
+      "epoch": 0.41587465899408055,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.935019694682973e-06,
+      "loss": 16.4738,
+      "step": 23000
+    },
+    {
+      "epoch": 0.4160554740632084,
+      "grad_norm": 42.5,
+      "learning_rate": 9.934991442376314e-06,
+      "loss": 17.0574,
+      "step": 23010
+    },
+    {
+      "epoch": 0.4162362891323363,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.934963190069654e-06,
+      "loss": 17.2055,
+      "step": 23020
+    },
+    {
+      "epoch": 0.41641710420146416,
+      "grad_norm": 38.15625,
+      "learning_rate": 9.934934937762995e-06,
+      "loss": 16.9899,
+      "step": 23030
+    },
+    {
+      "epoch": 0.41659791927059203,
+      "grad_norm": 43.375,
+      "learning_rate": 9.934906685456336e-06,
+      "loss": 17.4972,
+      "step": 23040
+    },
+    {
+      "epoch": 0.41677873433971985,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.934878433149674e-06,
+      "loss": 17.2599,
+      "step": 23050
+    },
+    {
+      "epoch": 0.4169595494088477,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.934850180843015e-06,
+      "loss": 17.2226,
+      "step": 23060
+    },
+    {
+      "epoch": 0.4171403644779756,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.934821928536356e-06,
+      "loss": 17.1681,
+      "step": 23070
+    },
+    {
+      "epoch": 0.41732117954710346,
+      "grad_norm": 44.0,
+      "learning_rate": 9.934793676229696e-06,
+      "loss": 17.1996,
+      "step": 23080
+    },
+    {
+      "epoch": 0.41750199461623133,
+      "grad_norm": 42.625,
+      "learning_rate": 9.934765423923037e-06,
+      "loss": 17.3154,
+      "step": 23090
+    },
+    {
+      "epoch": 0.4176828096853592,
+      "grad_norm": 41.125,
+      "learning_rate": 9.934737171616378e-06,
+      "loss": 16.4046,
+      "step": 23100
+    },
+    {
+      "epoch": 0.417863624754487,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.934708919309718e-06,
+      "loss": 17.0241,
+      "step": 23110
+    },
+    {
+      "epoch": 0.4180444398236149,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.934680667003059e-06,
+      "loss": 16.9767,
+      "step": 23120
+    },
+    {
+      "epoch": 0.41822525489274276,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.934652414696398e-06,
+      "loss": 17.1326,
+      "step": 23130
+    },
+    {
+      "epoch": 0.41840606996187063,
+      "grad_norm": 42.625,
+      "learning_rate": 9.934624162389738e-06,
+      "loss": 17.2872,
+      "step": 23140
+    },
+    {
+      "epoch": 0.4185868850309985,
+      "grad_norm": 40.5,
+      "learning_rate": 9.934595910083079e-06,
+      "loss": 16.9717,
+      "step": 23150
+    },
+    {
+      "epoch": 0.41876770010012637,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.934567657776421e-06,
+      "loss": 17.3705,
+      "step": 23160
+    },
+    {
+      "epoch": 0.4189485151692542,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.93453940546976e-06,
+      "loss": 17.2579,
+      "step": 23170
+    },
+    {
+      "epoch": 0.41912933023838206,
+      "grad_norm": 40.875,
+      "learning_rate": 9.934511153163101e-06,
+      "loss": 17.1618,
+      "step": 23180
+    },
+    {
+      "epoch": 0.41931014530750993,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.934482900856442e-06,
+      "loss": 17.0195,
+      "step": 23190
+    },
+    {
+      "epoch": 0.4194909603766378,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.934454648549782e-06,
+      "loss": 17.0593,
+      "step": 23200
+    },
+    {
+      "epoch": 0.41967177544576567,
+      "grad_norm": 42.375,
+      "learning_rate": 9.934426396243123e-06,
+      "loss": 17.0742,
+      "step": 23210
+    },
+    {
+      "epoch": 0.4198525905148935,
+      "grad_norm": 40.875,
+      "learning_rate": 9.934398143936462e-06,
+      "loss": 17.3838,
+      "step": 23220
+    },
+    {
+      "epoch": 0.42003340558402136,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.934369891629802e-06,
+      "loss": 17.1919,
+      "step": 23230
+    },
+    {
+      "epoch": 0.42021422065314923,
+      "grad_norm": 42.75,
+      "learning_rate": 9.934341639323143e-06,
+      "loss": 16.9229,
+      "step": 23240
+    },
+    {
+      "epoch": 0.4203950357222771,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.934313387016484e-06,
+      "loss": 17.1713,
+      "step": 23250
+    },
+    {
+      "epoch": 0.42057585079140497,
+      "grad_norm": 43.125,
+      "learning_rate": 9.934285134709824e-06,
+      "loss": 17.0081,
+      "step": 23260
+    },
+    {
+      "epoch": 0.42075666586053284,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.934256882403165e-06,
+      "loss": 17.02,
+      "step": 23270
+    },
+    {
+      "epoch": 0.42093748092966066,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.934228630096505e-06,
+      "loss": 17.4053,
+      "step": 23280
+    },
+    {
+      "epoch": 0.4211182959987885,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.934200377789846e-06,
+      "loss": 16.9431,
+      "step": 23290
+    },
+    {
+      "epoch": 0.4212991110679164,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.934172125483185e-06,
+      "loss": 16.5166,
+      "step": 23300
+    },
+    {
+      "epoch": 0.42147992613704427,
+      "grad_norm": 37.84375,
+      "learning_rate": 9.934143873176526e-06,
+      "loss": 17.1852,
+      "step": 23310
+    },
+    {
+      "epoch": 0.42166074120617214,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.934115620869866e-06,
+      "loss": 16.831,
+      "step": 23320
+    },
+    {
+      "epoch": 0.42184155627529996,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.934087368563207e-06,
+      "loss": 17.0062,
+      "step": 23330
+    },
+    {
+      "epoch": 0.4220223713444278,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.934059116256547e-06,
+      "loss": 17.0704,
+      "step": 23340
+    },
+    {
+      "epoch": 0.4222031864135557,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.934030863949888e-06,
+      "loss": 17.3631,
+      "step": 23350
+    },
+    {
+      "epoch": 0.42238400148268357,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.934002611643229e-06,
+      "loss": 17.2416,
+      "step": 23360
+    },
+    {
+      "epoch": 0.42256481655181144,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.93397435933657e-06,
+      "loss": 16.8875,
+      "step": 23370
+    },
+    {
+      "epoch": 0.4227456316209393,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.93394610702991e-06,
+      "loss": 17.089,
+      "step": 23380
+    },
+    {
+      "epoch": 0.4229264466900671,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.933917854723249e-06,
+      "loss": 17.174,
+      "step": 23390
+    },
+    {
+      "epoch": 0.423107261759195,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.93388960241659e-06,
+      "loss": 16.8249,
+      "step": 23400
+    },
+    {
+      "epoch": 0.42328807682832287,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.93386135010993e-06,
+      "loss": 16.6637,
+      "step": 23410
+    },
+    {
+      "epoch": 0.42346889189745074,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.93383309780327e-06,
+      "loss": 17.1979,
+      "step": 23420
+    },
+    {
+      "epoch": 0.4236497069665786,
+      "grad_norm": 43.125,
+      "learning_rate": 9.933804845496611e-06,
+      "loss": 16.7769,
+      "step": 23430
+    },
+    {
+      "epoch": 0.4238305220357065,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.933776593189952e-06,
+      "loss": 17.3441,
+      "step": 23440
+    },
+    {
+      "epoch": 0.4240113371048343,
+      "grad_norm": 39.5,
+      "learning_rate": 9.933748340883293e-06,
+      "loss": 17.4445,
+      "step": 23450
+    },
+    {
+      "epoch": 0.42419215217396217,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.933720088576633e-06,
+      "loss": 16.9229,
+      "step": 23460
+    },
+    {
+      "epoch": 0.42437296724309004,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.933691836269974e-06,
+      "loss": 17.0078,
+      "step": 23470
+    },
+    {
+      "epoch": 0.4245537823122179,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.933663583963313e-06,
+      "loss": 16.8134,
+      "step": 23480
+    },
+    {
+      "epoch": 0.4247345973813458,
+      "grad_norm": 42.5,
+      "learning_rate": 9.933635331656653e-06,
+      "loss": 16.873,
+      "step": 23490
+    },
+    {
+      "epoch": 0.4249154124504736,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.933607079349994e-06,
+      "loss": 16.9522,
+      "step": 23500
+    },
+    {
+      "epoch": 0.42509622751960147,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.933578827043335e-06,
+      "loss": 17.1385,
+      "step": 23510
+    },
+    {
+      "epoch": 0.42527704258872934,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.933550574736675e-06,
+      "loss": 17.2821,
+      "step": 23520
+    },
+    {
+      "epoch": 0.4254578576578572,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.933522322430016e-06,
+      "loss": 17.0284,
+      "step": 23530
+    },
+    {
+      "epoch": 0.4256386727269851,
+      "grad_norm": 40.875,
+      "learning_rate": 9.933494070123357e-06,
+      "loss": 17.3555,
+      "step": 23540
+    },
+    {
+      "epoch": 0.42581948779611295,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.933465817816697e-06,
+      "loss": 16.8159,
+      "step": 23550
+    },
+    {
+      "epoch": 0.42600030286524077,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.933437565510036e-06,
+      "loss": 17.1709,
+      "step": 23560
+    },
+    {
+      "epoch": 0.42618111793436864,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.933409313203377e-06,
+      "loss": 17.1132,
+      "step": 23570
+    },
+    {
+      "epoch": 0.4263619330034965,
+      "grad_norm": 40.875,
+      "learning_rate": 9.933381060896717e-06,
+      "loss": 16.997,
+      "step": 23580
+    },
+    {
+      "epoch": 0.4265427480726244,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.933352808590058e-06,
+      "loss": 17.2656,
+      "step": 23590
+    },
+    {
+      "epoch": 0.42672356314175225,
+      "grad_norm": 38.875,
+      "learning_rate": 9.933324556283399e-06,
+      "loss": 17.3424,
+      "step": 23600
+    },
+    {
+      "epoch": 0.4269043782108801,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.93329630397674e-06,
+      "loss": 17.2155,
+      "step": 23610
+    },
+    {
+      "epoch": 0.42708519328000794,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.93326805167008e-06,
+      "loss": 17.0139,
+      "step": 23620
+    },
+    {
+      "epoch": 0.4272660083491358,
+      "grad_norm": 42.125,
+      "learning_rate": 9.93323979936342e-06,
+      "loss": 17.278,
+      "step": 23630
+    },
+    {
+      "epoch": 0.4274468234182637,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.933211547056761e-06,
+      "loss": 17.0852,
+      "step": 23640
+    },
+    {
+      "epoch": 0.42762763848739155,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.9331832947501e-06,
+      "loss": 17.0743,
+      "step": 23650
+    },
+    {
+      "epoch": 0.4278084535565194,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.93315504244344e-06,
+      "loss": 17.3539,
+      "step": 23660
+    },
+    {
+      "epoch": 0.42798926862564723,
+      "grad_norm": 39.5,
+      "learning_rate": 9.933126790136781e-06,
+      "loss": 16.9718,
+      "step": 23670
+    },
+    {
+      "epoch": 0.4281700836947751,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.933098537830122e-06,
+      "loss": 17.266,
+      "step": 23680
+    },
+    {
+      "epoch": 0.428350898763903,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.933070285523462e-06,
+      "loss": 17.0662,
+      "step": 23690
+    },
+    {
+      "epoch": 0.42853171383303085,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.933042033216803e-06,
+      "loss": 16.9782,
+      "step": 23700
+    },
+    {
+      "epoch": 0.4287125289021587,
+      "grad_norm": 40.625,
+      "learning_rate": 9.933013780910144e-06,
+      "loss": 17.1737,
+      "step": 23710
+    },
+    {
+      "epoch": 0.4288933439712866,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.932985528603484e-06,
+      "loss": 17.0811,
+      "step": 23720
+    },
+    {
+      "epoch": 0.4290741590404144,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.932957276296823e-06,
+      "loss": 16.8908,
+      "step": 23730
+    },
+    {
+      "epoch": 0.4292549741095423,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.932929023990164e-06,
+      "loss": 17.0634,
+      "step": 23740
+    },
+    {
+      "epoch": 0.42943578917867015,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.932900771683505e-06,
+      "loss": 17.1466,
+      "step": 23750
+    },
+    {
+      "epoch": 0.429616604247798,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.932872519376845e-06,
+      "loss": 16.7681,
+      "step": 23760
+    },
+    {
+      "epoch": 0.4297974193169259,
+      "grad_norm": 38.34375,
+      "learning_rate": 9.932844267070186e-06,
+      "loss": 17.4063,
+      "step": 23770
+    },
+    {
+      "epoch": 0.42997823438605376,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.932816014763526e-06,
+      "loss": 17.1721,
+      "step": 23780
+    },
+    {
+      "epoch": 0.4301590494551816,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.932787762456867e-06,
+      "loss": 17.095,
+      "step": 23790
+    },
+    {
+      "epoch": 0.43033986452430945,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.932759510150208e-06,
+      "loss": 17.3986,
+      "step": 23800
+    },
+    {
+      "epoch": 0.4305206795934373,
+      "grad_norm": 44.25,
+      "learning_rate": 9.932731257843548e-06,
+      "loss": 17.5244,
+      "step": 23810
+    },
+    {
+      "epoch": 0.4307014946625652,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.932703005536887e-06,
+      "loss": 17.2414,
+      "step": 23820
+    },
+    {
+      "epoch": 0.43088230973169306,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.932674753230228e-06,
+      "loss": 16.83,
+      "step": 23830
+    },
+    {
+      "epoch": 0.4310631248008209,
+      "grad_norm": 38.875,
+      "learning_rate": 9.932646500923568e-06,
+      "loss": 17.2673,
+      "step": 23840
+    },
+    {
+      "epoch": 0.43124393986994874,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.932618248616909e-06,
+      "loss": 17.1928,
+      "step": 23850
+    },
+    {
+      "epoch": 0.4314247549390766,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.93258999631025e-06,
+      "loss": 17.1712,
+      "step": 23860
+    },
+    {
+      "epoch": 0.4316055700082045,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.93256174400359e-06,
+      "loss": 16.9979,
+      "step": 23870
+    },
+    {
+      "epoch": 0.43178638507733236,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.932533491696931e-06,
+      "loss": 17.0087,
+      "step": 23880
+    },
+    {
+      "epoch": 0.43196720014646023,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.932505239390272e-06,
+      "loss": 16.8195,
+      "step": 23890
+    },
+    {
+      "epoch": 0.43214801521558804,
+      "grad_norm": 38.15625,
+      "learning_rate": 9.932476987083612e-06,
+      "loss": 16.6816,
+      "step": 23900
+    },
+    {
+      "epoch": 0.4323288302847159,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.932448734776951e-06,
+      "loss": 16.6358,
+      "step": 23910
+    },
+    {
+      "epoch": 0.4325096453538438,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.932420482470292e-06,
+      "loss": 16.9947,
+      "step": 23920
+    },
+    {
+      "epoch": 0.43269046042297166,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.932392230163632e-06,
+      "loss": 17.3373,
+      "step": 23930
+    },
+    {
+      "epoch": 0.4328712754920995,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.932363977856973e-06,
+      "loss": 17.1195,
+      "step": 23940
+    },
+    {
+      "epoch": 0.4330520905612274,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.932335725550314e-06,
+      "loss": 17.2644,
+      "step": 23950
+    },
+    {
+      "epoch": 0.4332329056303552,
+      "grad_norm": 40.5,
+      "learning_rate": 9.932307473243654e-06,
+      "loss": 16.8849,
+      "step": 23960
+    },
+    {
+      "epoch": 0.4334137206994831,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.932279220936995e-06,
+      "loss": 17.2072,
+      "step": 23970
+    },
+    {
+      "epoch": 0.43359453576861096,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.932250968630335e-06,
+      "loss": 16.8773,
+      "step": 23980
+    },
+    {
+      "epoch": 0.4337753508377388,
+      "grad_norm": 42.375,
+      "learning_rate": 9.932222716323674e-06,
+      "loss": 17.1238,
+      "step": 23990
+    },
+    {
+      "epoch": 0.4339561659068667,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.932194464017015e-06,
+      "loss": 16.9992,
+      "step": 24000
+    },
+    {
+      "epoch": 0.4341369809759945,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.932166211710356e-06,
+      "loss": 17.4453,
+      "step": 24010
+    },
+    {
+      "epoch": 0.4343177960451224,
+      "grad_norm": 39.875,
+      "learning_rate": 9.932137959403696e-06,
+      "loss": 16.7997,
+      "step": 24020
+    },
+    {
+      "epoch": 0.43449861111425025,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.932109707097037e-06,
+      "loss": 17.2229,
+      "step": 24030
+    },
+    {
+      "epoch": 0.4346794261833781,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.932081454790377e-06,
+      "loss": 17.3416,
+      "step": 24040
+    },
+    {
+      "epoch": 0.434860241252506,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.932053202483718e-06,
+      "loss": 17.0329,
+      "step": 24050
+    },
+    {
+      "epoch": 0.43504105632163387,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.932024950177059e-06,
+      "loss": 16.8602,
+      "step": 24060
+    },
+    {
+      "epoch": 0.4352218713907617,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.9319966978704e-06,
+      "loss": 17.5934,
+      "step": 24070
+    },
+    {
+      "epoch": 0.43540268645988955,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.931968445563738e-06,
+      "loss": 16.8142,
+      "step": 24080
+    },
+    {
+      "epoch": 0.4355835015290174,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.931940193257079e-06,
+      "loss": 17.3197,
+      "step": 24090
+    },
+    {
+      "epoch": 0.4357643165981453,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.93191194095042e-06,
+      "loss": 17.4021,
+      "step": 24100
+    },
+    {
+      "epoch": 0.43594513166727317,
+      "grad_norm": 41.875,
+      "learning_rate": 9.93188368864376e-06,
+      "loss": 16.9198,
+      "step": 24110
+    },
+    {
+      "epoch": 0.43612594673640104,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.9318554363371e-06,
+      "loss": 17.3164,
+      "step": 24120
+    },
+    {
+      "epoch": 0.43630676180552885,
+      "grad_norm": 42.25,
+      "learning_rate": 9.931827184030441e-06,
+      "loss": 17.3602,
+      "step": 24130
+    },
+    {
+      "epoch": 0.4364875768746567,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.931798931723782e-06,
+      "loss": 17.1367,
+      "step": 24140
+    },
+    {
+      "epoch": 0.4366683919437846,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.931770679417123e-06,
+      "loss": 17.2549,
+      "step": 24150
+    },
+    {
+      "epoch": 0.43684920701291247,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.931742427110462e-06,
+      "loss": 17.0768,
+      "step": 24160
+    },
+    {
+      "epoch": 0.43703002208204034,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.931714174803802e-06,
+      "loss": 16.9547,
+      "step": 24170
+    },
+    {
+      "epoch": 0.43721083715116815,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.931685922497143e-06,
+      "loss": 16.8944,
+      "step": 24180
+    },
+    {
+      "epoch": 0.437391652220296,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.931657670190483e-06,
+      "loss": 17.1201,
+      "step": 24190
+    },
+    {
+      "epoch": 0.4375724672894239,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.931629417883824e-06,
+      "loss": 16.9747,
+      "step": 24200
+    },
+    {
+      "epoch": 0.43775328235855177,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.931601165577163e-06,
+      "loss": 16.6857,
+      "step": 24210
+    },
+    {
+      "epoch": 0.43793409742767964,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.931572913270505e-06,
+      "loss": 17.3643,
+      "step": 24220
+    },
+    {
+      "epoch": 0.4381149124968075,
+      "grad_norm": 43.375,
+      "learning_rate": 9.931544660963846e-06,
+      "loss": 17.2106,
+      "step": 24230
+    },
+    {
+      "epoch": 0.4382957275659353,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.931516408657187e-06,
+      "loss": 16.833,
+      "step": 24240
+    },
+    {
+      "epoch": 0.4384765426350632,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.931488156350525e-06,
+      "loss": 17.0473,
+      "step": 24250
+    },
+    {
+      "epoch": 0.43865735770419106,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.931459904043866e-06,
+      "loss": 16.9058,
+      "step": 24260
+    },
+    {
+      "epoch": 0.43883817277331894,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.931431651737207e-06,
+      "loss": 16.9209,
+      "step": 24270
+    },
+    {
+      "epoch": 0.4390189878424468,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.931403399430547e-06,
+      "loss": 17.0468,
+      "step": 24280
+    },
+    {
+      "epoch": 0.4391998029115747,
+      "grad_norm": 40.25,
+      "learning_rate": 9.931375147123888e-06,
+      "loss": 17.075,
+      "step": 24290
+    },
+    {
+      "epoch": 0.4393806179807025,
+      "grad_norm": 39.25,
+      "learning_rate": 9.931346894817229e-06,
+      "loss": 16.9849,
+      "step": 24300
+    },
+    {
+      "epoch": 0.43956143304983036,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.93131864251057e-06,
+      "loss": 17.4202,
+      "step": 24310
+    },
+    {
+      "epoch": 0.43974224811895823,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.93129039020391e-06,
+      "loss": 17.1817,
+      "step": 24320
+    },
+    {
+      "epoch": 0.4399230631880861,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.931262137897249e-06,
+      "loss": 17.1151,
+      "step": 24330
+    },
+    {
+      "epoch": 0.440103878257214,
+      "grad_norm": 38.375,
+      "learning_rate": 9.93123388559059e-06,
+      "loss": 16.8332,
+      "step": 24340
+    },
+    {
+      "epoch": 0.4402846933263418,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.93120563328393e-06,
+      "loss": 17.2657,
+      "step": 24350
+    },
+    {
+      "epoch": 0.44046550839546966,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.93117738097727e-06,
+      "loss": 17.5942,
+      "step": 24360
+    },
+    {
+      "epoch": 0.44064632346459753,
+      "grad_norm": 39.375,
+      "learning_rate": 9.931149128670611e-06,
+      "loss": 16.6795,
+      "step": 24370
+    },
+    {
+      "epoch": 0.4408271385337254,
+      "grad_norm": 42.25,
+      "learning_rate": 9.931120876363952e-06,
+      "loss": 16.5306,
+      "step": 24380
+    },
+    {
+      "epoch": 0.4410079536028533,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.931092624057292e-06,
+      "loss": 16.996,
+      "step": 24390
+    },
+    {
+      "epoch": 0.44118876867198115,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.931064371750633e-06,
+      "loss": 17.0845,
+      "step": 24400
+    },
+    {
+      "epoch": 0.44136958374110896,
+      "grad_norm": 40.5,
+      "learning_rate": 9.931036119443974e-06,
+      "loss": 16.9601,
+      "step": 24410
+    },
+    {
+      "epoch": 0.44155039881023683,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.931007867137313e-06,
+      "loss": 17.1927,
+      "step": 24420
+    },
+    {
+      "epoch": 0.4417312138793647,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.930979614830653e-06,
+      "loss": 17.0799,
+      "step": 24430
+    },
+    {
+      "epoch": 0.4419120289484926,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.930951362523994e-06,
+      "loss": 17.2743,
+      "step": 24440
+    },
+    {
+      "epoch": 0.44209284401762045,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.930923110217335e-06,
+      "loss": 17.1115,
+      "step": 24450
+    },
+    {
+      "epoch": 0.4422736590867483,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.930894857910675e-06,
+      "loss": 17.0342,
+      "step": 24460
+    },
+    {
+      "epoch": 0.44245447415587613,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.930866605604014e-06,
+      "loss": 17.1333,
+      "step": 24470
+    },
+    {
+      "epoch": 0.442635289225004,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.930838353297356e-06,
+      "loss": 17.1999,
+      "step": 24480
+    },
+    {
+      "epoch": 0.4428161042941319,
+      "grad_norm": 38.75,
+      "learning_rate": 9.930810100990697e-06,
+      "loss": 17.1589,
+      "step": 24490
+    },
+    {
+      "epoch": 0.44299691936325974,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.930781848684038e-06,
+      "loss": 17.0211,
+      "step": 24500
+    },
+    {
+      "epoch": 0.4431777344323876,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.930753596377377e-06,
+      "loss": 16.4922,
+      "step": 24510
+    },
+    {
+      "epoch": 0.44335854950151543,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.930725344070717e-06,
+      "loss": 16.8577,
+      "step": 24520
+    },
+    {
+      "epoch": 0.4435393645706433,
+      "grad_norm": 38.875,
+      "learning_rate": 9.930697091764058e-06,
+      "loss": 17.4463,
+      "step": 24530
+    },
+    {
+      "epoch": 0.4437201796397712,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.930668839457398e-06,
+      "loss": 17.0105,
+      "step": 24540
+    },
+    {
+      "epoch": 0.44390099470889904,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.930640587150739e-06,
+      "loss": 17.0146,
+      "step": 24550
+    },
+    {
+      "epoch": 0.4440818097780269,
+      "grad_norm": 39.25,
+      "learning_rate": 9.930612334844078e-06,
+      "loss": 17.0366,
+      "step": 24560
+    },
+    {
+      "epoch": 0.4442626248471548,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.93058408253742e-06,
+      "loss": 16.7516,
+      "step": 24570
+    },
+    {
+      "epoch": 0.4444434399162826,
+      "grad_norm": 40.125,
+      "learning_rate": 9.930555830230761e-06,
+      "loss": 16.9998,
+      "step": 24580
+    },
+    {
+      "epoch": 0.44462425498541047,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.9305275779241e-06,
+      "loss": 16.992,
+      "step": 24590
+    },
+    {
+      "epoch": 0.44480507005453834,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.93049932561744e-06,
+      "loss": 17.2084,
+      "step": 24600
+    },
+    {
+      "epoch": 0.4449858851236662,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.930471073310781e-06,
+      "loss": 17.1132,
+      "step": 24610
+    },
+    {
+      "epoch": 0.4451667001927941,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.930442821004122e-06,
+      "loss": 17.2638,
+      "step": 24620
+    },
+    {
+      "epoch": 0.44534751526192196,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.930414568697462e-06,
+      "loss": 17.0395,
+      "step": 24630
+    },
+    {
+      "epoch": 0.44552833033104977,
+      "grad_norm": 40.0,
+      "learning_rate": 9.930386316390801e-06,
+      "loss": 17.0863,
+      "step": 24640
+    },
+    {
+      "epoch": 0.44570914540017764,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.930358064084144e-06,
+      "loss": 16.9601,
+      "step": 24650
+    },
+    {
+      "epoch": 0.4458899604693055,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.930329811777484e-06,
+      "loss": 17.36,
+      "step": 24660
+    },
+    {
+      "epoch": 0.4460707755384334,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.930301559470825e-06,
+      "loss": 17.5581,
+      "step": 24670
+    },
+    {
+      "epoch": 0.44625159060756125,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.930273307164164e-06,
+      "loss": 17.0764,
+      "step": 24680
+    },
+    {
+      "epoch": 0.44643240567668907,
+      "grad_norm": 37.71875,
+      "learning_rate": 9.930245054857504e-06,
+      "loss": 17.2964,
+      "step": 24690
+    },
+    {
+      "epoch": 0.44661322074581694,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.930216802550845e-06,
+      "loss": 16.8406,
+      "step": 24700
+    },
+    {
+      "epoch": 0.4467940358149448,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.930188550244186e-06,
+      "loss": 16.8859,
+      "step": 24710
+    },
+    {
+      "epoch": 0.4469748508840727,
+      "grad_norm": 39.375,
+      "learning_rate": 9.930160297937526e-06,
+      "loss": 17.282,
+      "step": 24720
+    },
+    {
+      "epoch": 0.44715566595320055,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.930132045630865e-06,
+      "loss": 16.5238,
+      "step": 24730
+    },
+    {
+      "epoch": 0.4473364810223284,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.930103793324207e-06,
+      "loss": 17.1698,
+      "step": 24740
+    },
+    {
+      "epoch": 0.44751729609145624,
+      "grad_norm": 38.5625,
+      "learning_rate": 9.930075541017548e-06,
+      "loss": 17.1585,
+      "step": 24750
+    },
+    {
+      "epoch": 0.4476981111605841,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.930047288710887e-06,
+      "loss": 17.5576,
+      "step": 24760
+    },
+    {
+      "epoch": 0.447878926229712,
+      "grad_norm": 39.875,
+      "learning_rate": 9.930019036404228e-06,
+      "loss": 17.3201,
+      "step": 24770
+    },
+    {
+      "epoch": 0.44805974129883985,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.929990784097568e-06,
+      "loss": 16.9252,
+      "step": 24780
+    },
+    {
+      "epoch": 0.4482405563679677,
+      "grad_norm": 42.125,
+      "learning_rate": 9.929962531790909e-06,
+      "loss": 17.2519,
+      "step": 24790
+    },
+    {
+      "epoch": 0.4484213714370956,
+      "grad_norm": 38.40625,
+      "learning_rate": 9.92993427948425e-06,
+      "loss": 17.0709,
+      "step": 24800
+    },
+    {
+      "epoch": 0.4486021865062234,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.92990602717759e-06,
+      "loss": 17.0579,
+      "step": 24810
+    },
+    {
+      "epoch": 0.4487830015753513,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.929877774870929e-06,
+      "loss": 16.8308,
+      "step": 24820
+    },
+    {
+      "epoch": 0.44896381664447915,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.929849522564271e-06,
+      "loss": 17.0908,
+      "step": 24830
+    },
+    {
+      "epoch": 0.449144631713607,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.929821270257612e-06,
+      "loss": 16.7871,
+      "step": 24840
+    },
+    {
+      "epoch": 0.4493254467827349,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.929793017950951e-06,
+      "loss": 17.3479,
+      "step": 24850
+    },
+    {
+      "epoch": 0.4495062618518627,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.929764765644292e-06,
+      "loss": 17.52,
+      "step": 24860
+    },
+    {
+      "epoch": 0.4496870769209906,
+      "grad_norm": 41.875,
+      "learning_rate": 9.929736513337632e-06,
+      "loss": 17.0901,
+      "step": 24870
+    },
+    {
+      "epoch": 0.44986789199011845,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.929708261030973e-06,
+      "loss": 17.2132,
+      "step": 24880
+    },
+    {
+      "epoch": 0.4500487070592463,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.929680008724313e-06,
+      "loss": 17.1365,
+      "step": 24890
+    },
+    {
+      "epoch": 0.4502295221283742,
+      "grad_norm": 38.59375,
+      "learning_rate": 9.929651756417652e-06,
+      "loss": 17.0028,
+      "step": 24900
+    },
+    {
+      "epoch": 0.45041033719750206,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.929623504110993e-06,
+      "loss": 17.0601,
+      "step": 24910
+    },
+    {
+      "epoch": 0.4505911522666299,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.929595251804335e-06,
+      "loss": 16.7189,
+      "step": 24920
+    },
+    {
+      "epoch": 0.45077196733575775,
+      "grad_norm": 39.0,
+      "learning_rate": 9.929566999497676e-06,
+      "loss": 16.7454,
+      "step": 24930
+    },
+    {
+      "epoch": 0.4509527824048856,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.929538747191015e-06,
+      "loss": 17.6337,
+      "step": 24940
+    },
+    {
+      "epoch": 0.4511335974740135,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.929510494884355e-06,
+      "loss": 16.8392,
+      "step": 24950
+    },
+    {
+      "epoch": 0.45131441254314136,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.929482242577696e-06,
+      "loss": 16.8768,
+      "step": 24960
+    },
+    {
+      "epoch": 0.4514952276122692,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.929453990271037e-06,
+      "loss": 17.0484,
+      "step": 24970
+    },
+    {
+      "epoch": 0.45167604268139705,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.929425737964377e-06,
+      "loss": 17.2024,
+      "step": 24980
+    },
+    {
+      "epoch": 0.4518568577505249,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.929397485657716e-06,
+      "loss": 17.5434,
+      "step": 24990
+    },
+    {
+      "epoch": 0.4520376728196528,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.929369233351059e-06,
+      "loss": 16.9544,
+      "step": 25000
+    },
+    {
+      "epoch": 0.4520376728196528,
+      "eval_loss": 2.1385836601257324,
+      "eval_runtime": 229.8952,
+      "eval_samples_per_second": 3158.217,
+      "eval_steps_per_second": 49.349,
+      "step": 25000
+    },
+    {
+      "epoch": 0.45221848788878066,
+      "grad_norm": 42.25,
+      "learning_rate": 9.9293409810444e-06,
+      "loss": 17.2312,
+      "step": 25010
+    },
+    {
+      "epoch": 0.45239930295790853,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.929312728737738e-06,
+      "loss": 16.9233,
+      "step": 25020
+    },
+    {
+      "epoch": 0.45258011802703635,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.929284476431079e-06,
+      "loss": 17.1353,
+      "step": 25030
+    },
+    {
+      "epoch": 0.4527609330961642,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.92925622412442e-06,
+      "loss": 17.2442,
+      "step": 25040
+    },
+    {
+      "epoch": 0.4529417481652921,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.92922797181776e-06,
+      "loss": 17.064,
+      "step": 25050
+    },
+    {
+      "epoch": 0.45312256323441996,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.9291997195111e-06,
+      "loss": 17.032,
+      "step": 25060
+    },
+    {
+      "epoch": 0.45330337830354783,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.92917146720444e-06,
+      "loss": 17.1714,
+      "step": 25070
+    },
+    {
+      "epoch": 0.4534841933726757,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.92914321489778e-06,
+      "loss": 17.0163,
+      "step": 25080
+    },
+    {
+      "epoch": 0.4536650084418035,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.929114962591122e-06,
+      "loss": 17.1397,
+      "step": 25090
+    },
+    {
+      "epoch": 0.4538458235109314,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.929086710284463e-06,
+      "loss": 16.8583,
+      "step": 25100
+    },
+    {
+      "epoch": 0.45402663858005926,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.929058457977802e-06,
+      "loss": 16.8965,
+      "step": 25110
+    },
+    {
+      "epoch": 0.45420745364918713,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.929030205671143e-06,
+      "loss": 17.3544,
+      "step": 25120
+    },
+    {
+      "epoch": 0.454388268718315,
+      "grad_norm": 38.84375,
+      "learning_rate": 9.929001953364483e-06,
+      "loss": 17.2578,
+      "step": 25130
+    },
+    {
+      "epoch": 0.4545690837874428,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.928973701057824e-06,
+      "loss": 17.2073,
+      "step": 25140
+    },
+    {
+      "epoch": 0.4547498988565707,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.928945448751165e-06,
+      "loss": 16.996,
+      "step": 25150
+    },
+    {
+      "epoch": 0.45493071392569856,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.928917196444503e-06,
+      "loss": 17.1514,
+      "step": 25160
+    },
+    {
+      "epoch": 0.45511152899482643,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.928888944137844e-06,
+      "loss": 17.1125,
+      "step": 25170
+    },
+    {
+      "epoch": 0.4552923440639543,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.928860691831186e-06,
+      "loss": 16.7267,
+      "step": 25180
+    },
+    {
+      "epoch": 0.4554731591330822,
+      "grad_norm": 40.75,
+      "learning_rate": 9.928832439524525e-06,
+      "loss": 17.3858,
+      "step": 25190
+    },
+    {
+      "epoch": 0.45565397420221,
+      "grad_norm": 41.125,
+      "learning_rate": 9.928804187217866e-06,
+      "loss": 17.0973,
+      "step": 25200
+    },
+    {
+      "epoch": 0.45583478927133786,
+      "grad_norm": 41.25,
+      "learning_rate": 9.928775934911207e-06,
+      "loss": 16.7575,
+      "step": 25210
+    },
+    {
+      "epoch": 0.45601560434046573,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.928747682604547e-06,
+      "loss": 16.9319,
+      "step": 25220
+    },
+    {
+      "epoch": 0.4561964194095936,
+      "grad_norm": 40.0,
+      "learning_rate": 9.928719430297888e-06,
+      "loss": 16.9408,
+      "step": 25230
+    },
+    {
+      "epoch": 0.45637723447872147,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.928691177991228e-06,
+      "loss": 17.4637,
+      "step": 25240
+    },
+    {
+      "epoch": 0.45655804954784934,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.928662925684567e-06,
+      "loss": 17.4776,
+      "step": 25250
+    },
+    {
+      "epoch": 0.45673886461697716,
+      "grad_norm": 41.125,
+      "learning_rate": 9.928634673377908e-06,
+      "loss": 16.9808,
+      "step": 25260
+    },
+    {
+      "epoch": 0.45691967968610503,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.92860642107125e-06,
+      "loss": 16.7948,
+      "step": 25270
+    },
+    {
+      "epoch": 0.4571004947552329,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.92857816876459e-06,
+      "loss": 17.1491,
+      "step": 25280
+    },
+    {
+      "epoch": 0.45728130982436077,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.92854991645793e-06,
+      "loss": 17.0664,
+      "step": 25290
+    },
+    {
+      "epoch": 0.45746212489348864,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.92852166415127e-06,
+      "loss": 17.0108,
+      "step": 25300
+    },
+    {
+      "epoch": 0.45764293996261646,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.928493411844611e-06,
+      "loss": 16.851,
+      "step": 25310
+    },
+    {
+      "epoch": 0.45782375503174433,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.928465159537952e-06,
+      "loss": 16.9911,
+      "step": 25320
+    },
+    {
+      "epoch": 0.4580045701008722,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.92843690723129e-06,
+      "loss": 17.1102,
+      "step": 25330
+    },
+    {
+      "epoch": 0.45818538517000007,
+      "grad_norm": 39.75,
+      "learning_rate": 9.928408654924631e-06,
+      "loss": 16.7065,
+      "step": 25340
+    },
+    {
+      "epoch": 0.45836620023912794,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.928380402617974e-06,
+      "loss": 17.1513,
+      "step": 25350
+    },
+    {
+      "epoch": 0.4585470153082558,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.928352150311314e-06,
+      "loss": 17.343,
+      "step": 25360
+    },
+    {
+      "epoch": 0.4587278303773836,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.928323898004653e-06,
+      "loss": 16.954,
+      "step": 25370
+    },
+    {
+      "epoch": 0.4589086454465115,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.928295645697994e-06,
+      "loss": 16.6668,
+      "step": 25380
+    },
+    {
+      "epoch": 0.45908946051563937,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.928267393391334e-06,
+      "loss": 17.0833,
+      "step": 25390
+    },
+    {
+      "epoch": 0.45927027558476724,
+      "grad_norm": 42.625,
+      "learning_rate": 9.928239141084675e-06,
+      "loss": 17.1839,
+      "step": 25400
+    },
+    {
+      "epoch": 0.4594510906538951,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.928210888778016e-06,
+      "loss": 17.6705,
+      "step": 25410
+    },
+    {
+      "epoch": 0.459631905723023,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.928182636471355e-06,
+      "loss": 17.2035,
+      "step": 25420
+    },
+    {
+      "epoch": 0.4598127207921508,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.928154384164695e-06,
+      "loss": 17.355,
+      "step": 25430
+    },
+    {
+      "epoch": 0.45999353586127867,
+      "grad_norm": 42.0,
+      "learning_rate": 9.928126131858037e-06,
+      "loss": 17.029,
+      "step": 25440
+    },
+    {
+      "epoch": 0.46017435093040654,
+      "grad_norm": 45.03125,
+      "learning_rate": 9.928097879551376e-06,
+      "loss": 16.6355,
+      "step": 25450
+    },
+    {
+      "epoch": 0.4603551659995344,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.928069627244717e-06,
+      "loss": 17.3171,
+      "step": 25460
+    },
+    {
+      "epoch": 0.4605359810686623,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.928041374938058e-06,
+      "loss": 16.8692,
+      "step": 25470
+    },
+    {
+      "epoch": 0.4607167961377901,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.928013122631398e-06,
+      "loss": 17.0069,
+      "step": 25480
+    },
+    {
+      "epoch": 0.46089761120691797,
+      "grad_norm": 43.25,
+      "learning_rate": 9.927984870324739e-06,
+      "loss": 17.3876,
+      "step": 25490
+    },
+    {
+      "epoch": 0.46107842627604584,
+      "grad_norm": 42.375,
+      "learning_rate": 9.927956618018078e-06,
+      "loss": 16.6849,
+      "step": 25500
+    },
+    {
+      "epoch": 0.4612592413451737,
+      "grad_norm": 43.125,
+      "learning_rate": 9.927928365711418e-06,
+      "loss": 17.0812,
+      "step": 25510
+    },
+    {
+      "epoch": 0.4614400564143016,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.927900113404759e-06,
+      "loss": 17.2766,
+      "step": 25520
+    },
+    {
+      "epoch": 0.46162087148342945,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.927871861098101e-06,
+      "loss": 17.084,
+      "step": 25530
+    },
+    {
+      "epoch": 0.46180168655255727,
+      "grad_norm": 38.625,
+      "learning_rate": 9.92784360879144e-06,
+      "loss": 17.2581,
+      "step": 25540
+    },
+    {
+      "epoch": 0.46198250162168514,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.927815356484781e-06,
+      "loss": 17.2412,
+      "step": 25550
+    },
+    {
+      "epoch": 0.462163316690813,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.927787104178122e-06,
+      "loss": 16.8625,
+      "step": 25560
+    },
+    {
+      "epoch": 0.4623441317599409,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.927758851871462e-06,
+      "loss": 17.4807,
+      "step": 25570
+    },
+    {
+      "epoch": 0.46252494682906875,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.927730599564803e-06,
+      "loss": 16.9297,
+      "step": 25580
+    },
+    {
+      "epoch": 0.4627057618981966,
+      "grad_norm": 42.5,
+      "learning_rate": 9.927702347258142e-06,
+      "loss": 17.1685,
+      "step": 25590
+    },
+    {
+      "epoch": 0.46288657696732444,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.927674094951482e-06,
+      "loss": 17.0162,
+      "step": 25600
+    },
+    {
+      "epoch": 0.4630673920364523,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.927645842644823e-06,
+      "loss": 17.2933,
+      "step": 25610
+    },
+    {
+      "epoch": 0.4632482071055802,
+      "grad_norm": 44.71875,
+      "learning_rate": 9.927617590338164e-06,
+      "loss": 16.9681,
+      "step": 25620
+    },
+    {
+      "epoch": 0.46342902217470805,
+      "grad_norm": 39.5625,
+      "learning_rate": 9.927589338031504e-06,
+      "loss": 17.1248,
+      "step": 25630
+    },
+    {
+      "epoch": 0.4636098372438359,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.927561085724845e-06,
+      "loss": 17.0732,
+      "step": 25640
+    },
+    {
+      "epoch": 0.46379065231296374,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.927532833418185e-06,
+      "loss": 17.0435,
+      "step": 25650
+    },
+    {
+      "epoch": 0.4639714673820916,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.927504581111526e-06,
+      "loss": 17.0115,
+      "step": 25660
+    },
+    {
+      "epoch": 0.4641522824512195,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.927476328804867e-06,
+      "loss": 16.9102,
+      "step": 25670
+    },
+    {
+      "epoch": 0.46433309752034735,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.927448076498206e-06,
+      "loss": 16.9313,
+      "step": 25680
+    },
+    {
+      "epoch": 0.4645139125894752,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.927419824191546e-06,
+      "loss": 17.2224,
+      "step": 25690
+    },
+    {
+      "epoch": 0.4646947276586031,
+      "grad_norm": 43.75,
+      "learning_rate": 9.927391571884889e-06,
+      "loss": 17.1843,
+      "step": 25700
+    },
+    {
+      "epoch": 0.4648755427277309,
+      "grad_norm": 39.625,
+      "learning_rate": 9.927363319578228e-06,
+      "loss": 16.6983,
+      "step": 25710
+    },
+    {
+      "epoch": 0.4650563577968588,
+      "grad_norm": 43.5,
+      "learning_rate": 9.927335067271568e-06,
+      "loss": 16.8113,
+      "step": 25720
+    },
+    {
+      "epoch": 0.46523717286598665,
+      "grad_norm": 42.0,
+      "learning_rate": 9.927306814964909e-06,
+      "loss": 16.901,
+      "step": 25730
+    },
+    {
+      "epoch": 0.4654179879351145,
+      "grad_norm": 39.875,
+      "learning_rate": 9.92727856265825e-06,
+      "loss": 17.2378,
+      "step": 25740
+    },
+    {
+      "epoch": 0.4655988030042424,
+      "grad_norm": 40.125,
+      "learning_rate": 9.92725031035159e-06,
+      "loss": 17.2241,
+      "step": 25750
+    },
+    {
+      "epoch": 0.46577961807337026,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.927222058044929e-06,
+      "loss": 16.9758,
+      "step": 25760
+    },
+    {
+      "epoch": 0.4659604331424981,
+      "grad_norm": 38.53125,
+      "learning_rate": 9.92719380573827e-06,
+      "loss": 16.8353,
+      "step": 25770
+    },
+    {
+      "epoch": 0.46614124821162595,
+      "grad_norm": 40.625,
+      "learning_rate": 9.92716555343161e-06,
+      "loss": 16.9377,
+      "step": 25780
+    },
+    {
+      "epoch": 0.4663220632807538,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.927137301124952e-06,
+      "loss": 17.0842,
+      "step": 25790
+    },
+    {
+      "epoch": 0.4665028783498817,
+      "grad_norm": 41.25,
+      "learning_rate": 9.927109048818291e-06,
+      "loss": 17.0646,
+      "step": 25800
+    },
+    {
+      "epoch": 0.46668369341900956,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.927080796511632e-06,
+      "loss": 17.0465,
+      "step": 25810
+    },
+    {
+      "epoch": 0.4668645084881374,
+      "grad_norm": 38.1875,
+      "learning_rate": 9.927052544204973e-06,
+      "loss": 16.8757,
+      "step": 25820
+    },
+    {
+      "epoch": 0.46704532355726525,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.927024291898313e-06,
+      "loss": 17.617,
+      "step": 25830
+    },
+    {
+      "epoch": 0.4672261386263931,
+      "grad_norm": 39.65625,
+      "learning_rate": 9.926996039591654e-06,
+      "loss": 16.8398,
+      "step": 25840
+    },
+    {
+      "epoch": 0.467406953695521,
+      "grad_norm": 42.875,
+      "learning_rate": 9.926967787284993e-06,
+      "loss": 16.779,
+      "step": 25850
+    },
+    {
+      "epoch": 0.46758776876464886,
+      "grad_norm": 41.75,
+      "learning_rate": 9.926939534978333e-06,
+      "loss": 16.9525,
+      "step": 25860
+    },
+    {
+      "epoch": 0.46776858383377673,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.926911282671674e-06,
+      "loss": 16.375,
+      "step": 25870
+    },
+    {
+      "epoch": 0.46794939890290455,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.926883030365015e-06,
+      "loss": 17.3035,
+      "step": 25880
+    },
+    {
+      "epoch": 0.4681302139720324,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.926854778058355e-06,
+      "loss": 17.2107,
+      "step": 25890
+    },
+    {
+      "epoch": 0.4683110290411603,
+      "grad_norm": 45.1875,
+      "learning_rate": 9.926826525751696e-06,
+      "loss": 16.8853,
+      "step": 25900
+    },
+    {
+      "epoch": 0.46849184411028816,
+      "grad_norm": 41.125,
+      "learning_rate": 9.926798273445037e-06,
+      "loss": 16.858,
+      "step": 25910
+    },
+    {
+      "epoch": 0.46867265917941603,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.926770021138377e-06,
+      "loss": 17.2101,
+      "step": 25920
+    },
+    {
+      "epoch": 0.4688534742485439,
+      "grad_norm": 37.15625,
+      "learning_rate": 9.926741768831716e-06,
+      "loss": 16.781,
+      "step": 25930
+    },
+    {
+      "epoch": 0.4690342893176717,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.926713516525057e-06,
+      "loss": 16.9944,
+      "step": 25940
+    },
+    {
+      "epoch": 0.4692151043867996,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.926685264218397e-06,
+      "loss": 17.245,
+      "step": 25950
+    },
+    {
+      "epoch": 0.46939591945592746,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.926657011911738e-06,
+      "loss": 16.9312,
+      "step": 25960
+    },
+    {
+      "epoch": 0.46957673452505533,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.926628759605079e-06,
+      "loss": 17.4238,
+      "step": 25970
+    },
+    {
+      "epoch": 0.4697575495941832,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.92660050729842e-06,
+      "loss": 16.9905,
+      "step": 25980
+    },
+    {
+      "epoch": 0.469938364663311,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.92657225499176e-06,
+      "loss": 16.793,
+      "step": 25990
+    },
+    {
+      "epoch": 0.4701191797324389,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.9265440026851e-06,
+      "loss": 17.3917,
+      "step": 26000
+    },
+    {
+      "epoch": 0.47029999480156676,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.926515750378441e-06,
+      "loss": 17.1517,
+      "step": 26010
+    },
+    {
+      "epoch": 0.4704808098706946,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.92648749807178e-06,
+      "loss": 16.935,
+      "step": 26020
+    },
+    {
+      "epoch": 0.4706616249398225,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.92645924576512e-06,
+      "loss": 16.5918,
+      "step": 26030
+    },
+    {
+      "epoch": 0.47084244000895037,
+      "grad_norm": 40.25,
+      "learning_rate": 9.926430993458461e-06,
+      "loss": 17.144,
+      "step": 26040
+    },
+    {
+      "epoch": 0.4710232550780782,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.926402741151802e-06,
+      "loss": 16.9929,
+      "step": 26050
+    },
+    {
+      "epoch": 0.47120407014720606,
+      "grad_norm": 41.75,
+      "learning_rate": 9.926374488845143e-06,
+      "loss": 16.8202,
+      "step": 26060
+    },
+    {
+      "epoch": 0.4713848852163339,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.926346236538483e-06,
+      "loss": 16.9627,
+      "step": 26070
+    },
+    {
+      "epoch": 0.4715657002854618,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.926317984231824e-06,
+      "loss": 16.9011,
+      "step": 26080
+    },
+    {
+      "epoch": 0.47174651535458967,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.926289731925164e-06,
+      "loss": 17.0805,
+      "step": 26090
+    },
+    {
+      "epoch": 0.47192733042371754,
+      "grad_norm": 43.375,
+      "learning_rate": 9.926261479618505e-06,
+      "loss": 17.2723,
+      "step": 26100
+    },
+    {
+      "epoch": 0.47210814549284535,
+      "grad_norm": 41.875,
+      "learning_rate": 9.926233227311844e-06,
+      "loss": 17.1467,
+      "step": 26110
+    },
+    {
+      "epoch": 0.4722889605619732,
+      "grad_norm": 43.25,
+      "learning_rate": 9.926204975005185e-06,
+      "loss": 16.8147,
+      "step": 26120
+    },
+    {
+      "epoch": 0.4724697756311011,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.926176722698525e-06,
+      "loss": 16.6152,
+      "step": 26130
+    },
+    {
+      "epoch": 0.47265059070022897,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.926148470391866e-06,
+      "loss": 16.9996,
+      "step": 26140
+    },
+    {
+      "epoch": 0.47283140576935684,
+      "grad_norm": 41.875,
+      "learning_rate": 9.926120218085206e-06,
+      "loss": 16.8885,
+      "step": 26150
+    },
+    {
+      "epoch": 0.47301222083848465,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.926091965778547e-06,
+      "loss": 16.686,
+      "step": 26160
+    },
+    {
+      "epoch": 0.4731930359076125,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.926063713471888e-06,
+      "loss": 17.0671,
+      "step": 26170
+    },
+    {
+      "epoch": 0.4733738509767404,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.926035461165228e-06,
+      "loss": 16.7281,
+      "step": 26180
+    },
+    {
+      "epoch": 0.47355466604586827,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.926007208858567e-06,
+      "loss": 17.1318,
+      "step": 26190
+    },
+    {
+      "epoch": 0.47373548111499614,
+      "grad_norm": 40.375,
+      "learning_rate": 9.925978956551908e-06,
+      "loss": 17.2749,
+      "step": 26200
+    },
+    {
+      "epoch": 0.473916296184124,
+      "grad_norm": 44.0,
+      "learning_rate": 9.925950704245248e-06,
+      "loss": 17.1022,
+      "step": 26210
+    },
+    {
+      "epoch": 0.4740971112532518,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.925922451938589e-06,
+      "loss": 17.0693,
+      "step": 26220
+    },
+    {
+      "epoch": 0.4742779263223797,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.92589419963193e-06,
+      "loss": 16.8296,
+      "step": 26230
+    },
+    {
+      "epoch": 0.47445874139150757,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.92586594732527e-06,
+      "loss": 17.0046,
+      "step": 26240
+    },
+    {
+      "epoch": 0.47463955646063544,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.925837695018611e-06,
+      "loss": 17.1826,
+      "step": 26250
+    },
+    {
+      "epoch": 0.4748203715297633,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.925809442711952e-06,
+      "loss": 17.0491,
+      "step": 26260
+    },
+    {
+      "epoch": 0.4750011865988912,
+      "grad_norm": 40.875,
+      "learning_rate": 9.925781190405292e-06,
+      "loss": 16.8475,
+      "step": 26270
+    },
+    {
+      "epoch": 0.475182001668019,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.925752938098631e-06,
+      "loss": 17.4165,
+      "step": 26280
+    },
+    {
+      "epoch": 0.47536281673714686,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.925724685791972e-06,
+      "loss": 17.1112,
+      "step": 26290
+    },
+    {
+      "epoch": 0.47554363180627474,
+      "grad_norm": 40.0,
+      "learning_rate": 9.925696433485312e-06,
+      "loss": 17.0229,
+      "step": 26300
+    },
+    {
+      "epoch": 0.4757244468754026,
+      "grad_norm": 38.0625,
+      "learning_rate": 9.925668181178653e-06,
+      "loss": 17.1572,
+      "step": 26310
+    },
+    {
+      "epoch": 0.4759052619445305,
+      "grad_norm": 39.28125,
+      "learning_rate": 9.925639928871994e-06,
+      "loss": 17.0653,
+      "step": 26320
+    },
+    {
+      "epoch": 0.4760860770136583,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.925611676565334e-06,
+      "loss": 16.9044,
+      "step": 26330
+    },
+    {
+      "epoch": 0.47626689208278616,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.925583424258675e-06,
+      "loss": 16.9558,
+      "step": 26340
+    },
+    {
+      "epoch": 0.47644770715191403,
+      "grad_norm": 40.625,
+      "learning_rate": 9.925555171952015e-06,
+      "loss": 17.1931,
+      "step": 26350
+    },
+    {
+      "epoch": 0.4766285222210419,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.925526919645354e-06,
+      "loss": 16.6998,
+      "step": 26360
+    },
+    {
+      "epoch": 0.4768093372901698,
+      "grad_norm": 41.0,
+      "learning_rate": 9.925498667338695e-06,
+      "loss": 16.9934,
+      "step": 26370
+    },
+    {
+      "epoch": 0.47699015235929765,
+      "grad_norm": 43.125,
+      "learning_rate": 9.925470415032036e-06,
+      "loss": 17.419,
+      "step": 26380
+    },
+    {
+      "epoch": 0.47717096742842546,
+      "grad_norm": 41.875,
+      "learning_rate": 9.925442162725376e-06,
+      "loss": 17.275,
+      "step": 26390
+    },
+    {
+      "epoch": 0.47735178249755333,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.925413910418717e-06,
+      "loss": 17.2304,
+      "step": 26400
+    },
+    {
+      "epoch": 0.4775325975666812,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.925385658112058e-06,
+      "loss": 16.9184,
+      "step": 26410
+    },
+    {
+      "epoch": 0.4777134126358091,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.925357405805398e-06,
+      "loss": 16.8243,
+      "step": 26420
+    },
+    {
+      "epoch": 0.47789422770493695,
+      "grad_norm": 42.375,
+      "learning_rate": 9.925329153498739e-06,
+      "loss": 17.2043,
+      "step": 26430
+    },
+    {
+      "epoch": 0.4780750427740648,
+      "grad_norm": 42.0,
+      "learning_rate": 9.92530090119208e-06,
+      "loss": 16.9562,
+      "step": 26440
+    },
+    {
+      "epoch": 0.47825585784319263,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.925272648885418e-06,
+      "loss": 16.9889,
+      "step": 26450
+    },
+    {
+      "epoch": 0.4784366729123205,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.925244396578759e-06,
+      "loss": 17.1232,
+      "step": 26460
+    },
+    {
+      "epoch": 0.4786174879814484,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.9252161442721e-06,
+      "loss": 17.097,
+      "step": 26470
+    },
+    {
+      "epoch": 0.47879830305057625,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.92518789196544e-06,
+      "loss": 16.9876,
+      "step": 26480
+    },
+    {
+      "epoch": 0.4789791181197041,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.92515963965878e-06,
+      "loss": 17.3516,
+      "step": 26490
+    },
+    {
+      "epoch": 0.47915993318883193,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.925131387352121e-06,
+      "loss": 17.1503,
+      "step": 26500
+    },
+    {
+      "epoch": 0.4793407482579598,
+      "grad_norm": 42.375,
+      "learning_rate": 9.925103135045462e-06,
+      "loss": 17.4379,
+      "step": 26510
+    },
+    {
+      "epoch": 0.4795215633270877,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.925074882738803e-06,
+      "loss": 16.7705,
+      "step": 26520
+    },
+    {
+      "epoch": 0.47970237839621555,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.925046630432142e-06,
+      "loss": 16.8241,
+      "step": 26530
+    },
+    {
+      "epoch": 0.4798831934653434,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.925018378125482e-06,
+      "loss": 17.3142,
+      "step": 26540
+    },
+    {
+      "epoch": 0.4800640085344713,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.924990125818823e-06,
+      "loss": 16.7404,
+      "step": 26550
+    },
+    {
+      "epoch": 0.4802448236035991,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.924961873512163e-06,
+      "loss": 17.6105,
+      "step": 26560
+    },
+    {
+      "epoch": 0.480425638672727,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.924933621205504e-06,
+      "loss": 16.9162,
+      "step": 26570
+    },
+    {
+      "epoch": 0.48060645374185484,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.924905368898845e-06,
+      "loss": 16.7582,
+      "step": 26580
+    },
+    {
+      "epoch": 0.4807872688109827,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.924877116592185e-06,
+      "loss": 17.1185,
+      "step": 26590
+    },
+    {
+      "epoch": 0.4809680838801106,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.924848864285526e-06,
+      "loss": 17.1527,
+      "step": 26600
+    },
+    {
+      "epoch": 0.48114889894923846,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.924820611978867e-06,
+      "loss": 17.2085,
+      "step": 26610
+    },
+    {
+      "epoch": 0.4813297140183663,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.924792359672206e-06,
+      "loss": 17.3011,
+      "step": 26620
+    },
+    {
+      "epoch": 0.48151052908749414,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.924764107365546e-06,
+      "loss": 17.2475,
+      "step": 26630
+    },
+    {
+      "epoch": 0.481691344156622,
+      "grad_norm": 41.125,
+      "learning_rate": 9.924735855058887e-06,
+      "loss": 17.0385,
+      "step": 26640
+    },
+    {
+      "epoch": 0.4818721592257499,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.924707602752227e-06,
+      "loss": 16.8667,
+      "step": 26650
+    },
+    {
+      "epoch": 0.48205297429487776,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.924679350445568e-06,
+      "loss": 17.118,
+      "step": 26660
+    },
+    {
+      "epoch": 0.48223378936400557,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.924651098138909e-06,
+      "loss": 17.058,
+      "step": 26670
+    },
+    {
+      "epoch": 0.48241460443313344,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.92462284583225e-06,
+      "loss": 17.0949,
+      "step": 26680
+    },
+    {
+      "epoch": 0.4825954195022613,
+      "grad_norm": 41.5,
+      "learning_rate": 9.92459459352559e-06,
+      "loss": 17.0175,
+      "step": 26690
+    },
+    {
+      "epoch": 0.4827762345713892,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.92456634121893e-06,
+      "loss": 16.6859,
+      "step": 26700
+    },
+    {
+      "epoch": 0.48295704964051706,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.92453808891227e-06,
+      "loss": 16.7084,
+      "step": 26710
+    },
+    {
+      "epoch": 0.4831378647096449,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.92450983660561e-06,
+      "loss": 16.9311,
+      "step": 26720
+    },
+    {
+      "epoch": 0.48331867977877274,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.92448158429895e-06,
+      "loss": 17.2466,
+      "step": 26730
+    },
+    {
+      "epoch": 0.4834994948479006,
+      "grad_norm": 38.46875,
+      "learning_rate": 9.924453331992291e-06,
+      "loss": 16.9546,
+      "step": 26740
+    },
+    {
+      "epoch": 0.4836803099170285,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.924425079685632e-06,
+      "loss": 17.0234,
+      "step": 26750
+    },
+    {
+      "epoch": 0.48386112498615635,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.924396827378973e-06,
+      "loss": 17.4127,
+      "step": 26760
+    },
+    {
+      "epoch": 0.4840419400552842,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.924368575072313e-06,
+      "loss": 17.1631,
+      "step": 26770
+    },
+    {
+      "epoch": 0.48422275512441204,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.924340322765654e-06,
+      "loss": 16.9457,
+      "step": 26780
+    },
+    {
+      "epoch": 0.4844035701935399,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.924312070458993e-06,
+      "loss": 17.0132,
+      "step": 26790
+    },
+    {
+      "epoch": 0.4845843852626678,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.924283818152333e-06,
+      "loss": 17.3709,
+      "step": 26800
+    },
+    {
+      "epoch": 0.48476520033179565,
+      "grad_norm": 41.25,
+      "learning_rate": 9.924255565845674e-06,
+      "loss": 17.4508,
+      "step": 26810
+    },
+    {
+      "epoch": 0.4849460154009235,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.924227313539015e-06,
+      "loss": 17.0839,
+      "step": 26820
+    },
+    {
+      "epoch": 0.4851268304700514,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.924199061232355e-06,
+      "loss": 17.0203,
+      "step": 26830
+    },
+    {
+      "epoch": 0.4853076455391792,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.924170808925696e-06,
+      "loss": 17.0355,
+      "step": 26840
+    },
+    {
+      "epoch": 0.4854884606083071,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.924142556619036e-06,
+      "loss": 17.1141,
+      "step": 26850
+    },
+    {
+      "epoch": 0.48566927567743495,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.924114304312377e-06,
+      "loss": 17.0769,
+      "step": 26860
+    },
+    {
+      "epoch": 0.4858500907465628,
+      "grad_norm": 38.59375,
+      "learning_rate": 9.924086052005718e-06,
+      "loss": 16.6634,
+      "step": 26870
+    },
+    {
+      "epoch": 0.4860309058156907,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.924057799699057e-06,
+      "loss": 16.7654,
+      "step": 26880
+    },
+    {
+      "epoch": 0.48621172088481857,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.924029547392397e-06,
+      "loss": 17.1133,
+      "step": 26890
+    },
+    {
+      "epoch": 0.4863925359539464,
+      "grad_norm": 44.125,
+      "learning_rate": 9.924001295085738e-06,
+      "loss": 17.0373,
+      "step": 26900
+    },
+    {
+      "epoch": 0.48657335102307425,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.923973042779078e-06,
+      "loss": 17.2041,
+      "step": 26910
+    },
+    {
+      "epoch": 0.4867541660922021,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.923944790472419e-06,
+      "loss": 16.9118,
+      "step": 26920
+    },
+    {
+      "epoch": 0.48693498116133,
+      "grad_norm": 43.125,
+      "learning_rate": 9.92391653816576e-06,
+      "loss": 17.4285,
+      "step": 26930
+    },
+    {
+      "epoch": 0.48711579623045786,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.9238882858591e-06,
+      "loss": 17.1475,
+      "step": 26940
+    },
+    {
+      "epoch": 0.4872966112995857,
+      "grad_norm": 40.625,
+      "learning_rate": 9.923860033552441e-06,
+      "loss": 17.7247,
+      "step": 26950
+    },
+    {
+      "epoch": 0.48747742636871355,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.92383178124578e-06,
+      "loss": 17.3309,
+      "step": 26960
+    },
+    {
+      "epoch": 0.4876582414378414,
+      "grad_norm": 46.21875,
+      "learning_rate": 9.92380352893912e-06,
+      "loss": 17.1113,
+      "step": 26970
+    },
+    {
+      "epoch": 0.4878390565069693,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.923775276632461e-06,
+      "loss": 17.2221,
+      "step": 26980
+    },
+    {
+      "epoch": 0.48801987157609716,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.923747024325802e-06,
+      "loss": 17.3459,
+      "step": 26990
+    },
+    {
+      "epoch": 0.48820068664522503,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.923718772019142e-06,
+      "loss": 17.1826,
+      "step": 27000
+    },
+    {
+      "epoch": 0.48838150171435285,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.923690519712483e-06,
+      "loss": 16.9989,
+      "step": 27010
+    },
+    {
+      "epoch": 0.4885623167834807,
+      "grad_norm": 44.6875,
+      "learning_rate": 9.923662267405824e-06,
+      "loss": 17.1884,
+      "step": 27020
+    },
+    {
+      "epoch": 0.4887431318526086,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.923634015099164e-06,
+      "loss": 17.0027,
+      "step": 27030
+    },
+    {
+      "epoch": 0.48892394692173646,
+      "grad_norm": 43.625,
+      "learning_rate": 9.923605762792505e-06,
+      "loss": 16.6061,
+      "step": 27040
+    },
+    {
+      "epoch": 0.48910476199086433,
+      "grad_norm": 42.875,
+      "learning_rate": 9.923577510485844e-06,
+      "loss": 17.0401,
+      "step": 27050
+    },
+    {
+      "epoch": 0.4892855770599922,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.923549258179184e-06,
+      "loss": 17.1379,
+      "step": 27060
+    },
+    {
+      "epoch": 0.48946639212912,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.923521005872525e-06,
+      "loss": 16.6934,
+      "step": 27070
+    },
+    {
+      "epoch": 0.4896472071982479,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.923492753565866e-06,
+      "loss": 16.9134,
+      "step": 27080
+    },
+    {
+      "epoch": 0.48982802226737576,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.923464501259206e-06,
+      "loss": 16.6742,
+      "step": 27090
+    },
+    {
+      "epoch": 0.49000883733650363,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.923436248952547e-06,
+      "loss": 17.0472,
+      "step": 27100
+    },
+    {
+      "epoch": 0.4901896524056315,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.923407996645888e-06,
+      "loss": 17.0729,
+      "step": 27110
+    },
+    {
+      "epoch": 0.4903704674747593,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.923379744339228e-06,
+      "loss": 17.1721,
+      "step": 27120
+    },
+    {
+      "epoch": 0.4905512825438872,
+      "grad_norm": 41.5,
+      "learning_rate": 9.923351492032569e-06,
+      "loss": 16.6988,
+      "step": 27130
+    },
+    {
+      "epoch": 0.49073209761301506,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.923323239725908e-06,
+      "loss": 16.7262,
+      "step": 27140
+    },
+    {
+      "epoch": 0.49091291268214293,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.923294987419248e-06,
+      "loss": 17.3932,
+      "step": 27150
+    },
+    {
+      "epoch": 0.4910937277512708,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.923266735112589e-06,
+      "loss": 17.2486,
+      "step": 27160
+    },
+    {
+      "epoch": 0.4912745428203987,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.92323848280593e-06,
+      "loss": 16.8497,
+      "step": 27170
+    },
+    {
+      "epoch": 0.4914553578895265,
+      "grad_norm": 40.625,
+      "learning_rate": 9.92321023049927e-06,
+      "loss": 16.9637,
+      "step": 27180
+    },
+    {
+      "epoch": 0.49163617295865436,
+      "grad_norm": 40.125,
+      "learning_rate": 9.92318197819261e-06,
+      "loss": 16.8488,
+      "step": 27190
+    },
+    {
+      "epoch": 0.49181698802778223,
+      "grad_norm": 42.0,
+      "learning_rate": 9.923153725885951e-06,
+      "loss": 17.0431,
+      "step": 27200
+    },
+    {
+      "epoch": 0.4919978030969101,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.923125473579292e-06,
+      "loss": 16.9398,
+      "step": 27210
+    },
+    {
+      "epoch": 0.492178618166038,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.923097221272631e-06,
+      "loss": 17.0839,
+      "step": 27220
+    },
+    {
+      "epoch": 0.49235943323516584,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.923068968965972e-06,
+      "loss": 17.1533,
+      "step": 27230
+    },
+    {
+      "epoch": 0.49254024830429366,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.923040716659312e-06,
+      "loss": 17.4095,
+      "step": 27240
+    },
+    {
+      "epoch": 0.49272106337342153,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.923012464352653e-06,
+      "loss": 16.9864,
+      "step": 27250
+    },
+    {
+      "epoch": 0.4929018784425494,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.922984212045993e-06,
+      "loss": 16.8673,
+      "step": 27260
+    },
+    {
+      "epoch": 0.4930826935116773,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.922955959739332e-06,
+      "loss": 17.4023,
+      "step": 27270
+    },
+    {
+      "epoch": 0.49326350858080514,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.922927707432675e-06,
+      "loss": 17.256,
+      "step": 27280
+    },
+    {
+      "epoch": 0.49344432364993296,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.922899455126015e-06,
+      "loss": 16.9397,
+      "step": 27290
+    },
+    {
+      "epoch": 0.49362513871906083,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.922871202819356e-06,
+      "loss": 16.7331,
+      "step": 27300
+    },
+    {
+      "epoch": 0.4938059537881887,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.922842950512695e-06,
+      "loss": 16.9543,
+      "step": 27310
+    },
+    {
+      "epoch": 0.49398676885731657,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.922814698206036e-06,
+      "loss": 17.0799,
+      "step": 27320
+    },
+    {
+      "epoch": 0.49416758392644444,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.922786445899376e-06,
+      "loss": 17.436,
+      "step": 27330
+    },
+    {
+      "epoch": 0.4943483989955723,
+      "grad_norm": 41.75,
+      "learning_rate": 9.922758193592717e-06,
+      "loss": 17.0082,
+      "step": 27340
+    },
+    {
+      "epoch": 0.49452921406470013,
+      "grad_norm": 45.0,
+      "learning_rate": 9.922729941286057e-06,
+      "loss": 16.635,
+      "step": 27350
+    },
+    {
+      "epoch": 0.494710029133828,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.922701688979396e-06,
+      "loss": 16.8168,
+      "step": 27360
+    },
+    {
+      "epoch": 0.49489084420295587,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.922673436672739e-06,
+      "loss": 17.2225,
+      "step": 27370
+    },
+    {
+      "epoch": 0.49507165927208374,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.92264518436608e-06,
+      "loss": 16.5935,
+      "step": 27380
+    },
+    {
+      "epoch": 0.4952524743412116,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.922616932059418e-06,
+      "loss": 16.63,
+      "step": 27390
+    },
+    {
+      "epoch": 0.4954332894103395,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.922588679752759e-06,
+      "loss": 16.9653,
+      "step": 27400
+    },
+    {
+      "epoch": 0.4956141044794673,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.9225604274461e-06,
+      "loss": 17.6395,
+      "step": 27410
+    },
+    {
+      "epoch": 0.49579491954859517,
+      "grad_norm": 40.875,
+      "learning_rate": 9.92253217513944e-06,
+      "loss": 17.131,
+      "step": 27420
+    },
+    {
+      "epoch": 0.49597573461772304,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.92250392283278e-06,
+      "loss": 17.4824,
+      "step": 27430
+    },
+    {
+      "epoch": 0.4961565496868509,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.922475670526121e-06,
+      "loss": 17.3884,
+      "step": 27440
+    },
+    {
+      "epoch": 0.4963373647559788,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.922447418219462e-06,
+      "loss": 17.2447,
+      "step": 27450
+    },
+    {
+      "epoch": 0.4965181798251066,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.922419165912803e-06,
+      "loss": 16.9101,
+      "step": 27460
+    },
+    {
+      "epoch": 0.49669899489423447,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.922390913606143e-06,
+      "loss": 17.1153,
+      "step": 27470
+    },
+    {
+      "epoch": 0.49687980996336234,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.922362661299482e-06,
+      "loss": 17.0791,
+      "step": 27480
+    },
+    {
+      "epoch": 0.4970606250324902,
+      "grad_norm": 39.0,
+      "learning_rate": 9.922334408992823e-06,
+      "loss": 17.3504,
+      "step": 27490
+    },
+    {
+      "epoch": 0.4972414401016181,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.922306156686163e-06,
+      "loss": 17.0333,
+      "step": 27500
+    },
+    {
+      "epoch": 0.49742225517074595,
+      "grad_norm": 42.625,
+      "learning_rate": 9.922277904379504e-06,
+      "loss": 16.9336,
+      "step": 27510
+    },
+    {
+      "epoch": 0.49760307023987377,
+      "grad_norm": 39.875,
+      "learning_rate": 9.922249652072845e-06,
+      "loss": 16.8369,
+      "step": 27520
+    },
+    {
+      "epoch": 0.49778388530900164,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.922221399766184e-06,
+      "loss": 17.2567,
+      "step": 27530
+    },
+    {
+      "epoch": 0.4979647003781295,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.922193147459526e-06,
+      "loss": 16.7613,
+      "step": 27540
+    },
+    {
+      "epoch": 0.4981455154472574,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.922164895152866e-06,
+      "loss": 17.1967,
+      "step": 27550
+    },
+    {
+      "epoch": 0.49832633051638525,
+      "grad_norm": 41.625,
+      "learning_rate": 9.922136642846207e-06,
+      "loss": 17.1377,
+      "step": 27560
+    },
+    {
+      "epoch": 0.4985071455855131,
+      "grad_norm": 41.25,
+      "learning_rate": 9.922108390539546e-06,
+      "loss": 16.5317,
+      "step": 27570
+    },
+    {
+      "epoch": 0.49868796065464094,
+      "grad_norm": 39.75,
+      "learning_rate": 9.922080138232887e-06,
+      "loss": 17.3832,
+      "step": 27580
+    },
+    {
+      "epoch": 0.4988687757237688,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.922051885926227e-06,
+      "loss": 16.9755,
+      "step": 27590
+    },
+    {
+      "epoch": 0.4990495907928967,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.922023633619568e-06,
+      "loss": 16.8368,
+      "step": 27600
+    },
+    {
+      "epoch": 0.49923040586202455,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.921995381312908e-06,
+      "loss": 16.3951,
+      "step": 27610
+    },
+    {
+      "epoch": 0.4994112209311524,
+      "grad_norm": 43.375,
+      "learning_rate": 9.921967129006247e-06,
+      "loss": 17.0727,
+      "step": 27620
+    },
+    {
+      "epoch": 0.49959203600028024,
+      "grad_norm": 39.375,
+      "learning_rate": 9.92193887669959e-06,
+      "loss": 16.7404,
+      "step": 27630
+    },
+    {
+      "epoch": 0.4997728510694081,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.92191062439293e-06,
+      "loss": 16.76,
+      "step": 27640
+    },
+    {
+      "epoch": 0.499953666138536,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.92188237208627e-06,
+      "loss": 17.0706,
+      "step": 27650
+    },
+    {
+      "epoch": 0.5001344812076638,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.92185411977961e-06,
+      "loss": 16.7611,
+      "step": 27660
+    },
+    {
+      "epoch": 0.5003152962767917,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.92182586747295e-06,
+      "loss": 17.0832,
+      "step": 27670
+    },
+    {
+      "epoch": 0.5004961113459195,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.921797615166291e-06,
+      "loss": 16.9609,
+      "step": 27680
+    },
+    {
+      "epoch": 0.5006769264150475,
+      "grad_norm": 40.0,
+      "learning_rate": 9.921769362859632e-06,
+      "loss": 17.1072,
+      "step": 27690
+    },
+    {
+      "epoch": 0.5008577414841753,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.92174111055297e-06,
+      "loss": 16.9236,
+      "step": 27700
+    },
+    {
+      "epoch": 0.5010385565533032,
+      "grad_norm": 40.75,
+      "learning_rate": 9.921712858246311e-06,
+      "loss": 16.9665,
+      "step": 27710
+    },
+    {
+      "epoch": 0.501219371622431,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.921684605939654e-06,
+      "loss": 16.732,
+      "step": 27720
+    },
+    {
+      "epoch": 0.5014001866915588,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.921656353632994e-06,
+      "loss": 16.7196,
+      "step": 27730
+    },
+    {
+      "epoch": 0.5015810017606868,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.921628101326333e-06,
+      "loss": 17.0108,
+      "step": 27740
+    },
+    {
+      "epoch": 0.5017618168298146,
+      "grad_norm": 40.25,
+      "learning_rate": 9.921599849019674e-06,
+      "loss": 16.9423,
+      "step": 27750
+    },
+    {
+      "epoch": 0.5019426318989425,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.921571596713014e-06,
+      "loss": 17.2849,
+      "step": 27760
+    },
+    {
+      "epoch": 0.5021234469680703,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.921543344406355e-06,
+      "loss": 17.3861,
+      "step": 27770
+    },
+    {
+      "epoch": 0.5023042620371981,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.921515092099696e-06,
+      "loss": 16.6662,
+      "step": 27780
+    },
+    {
+      "epoch": 0.5024850771063261,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.921486839793035e-06,
+      "loss": 16.7922,
+      "step": 27790
+    },
+    {
+      "epoch": 0.5026658921754539,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.921458587486377e-06,
+      "loss": 16.8959,
+      "step": 27800
+    },
+    {
+      "epoch": 0.5028467072445818,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.921430335179718e-06,
+      "loss": 17.2021,
+      "step": 27810
+    },
+    {
+      "epoch": 0.5030275223137096,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.921402082873056e-06,
+      "loss": 16.8186,
+      "step": 27820
+    },
+    {
+      "epoch": 0.5032083373828374,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.921373830566397e-06,
+      "loss": 16.8509,
+      "step": 27830
+    },
+    {
+      "epoch": 0.5033891524519654,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.921345578259738e-06,
+      "loss": 16.794,
+      "step": 27840
+    },
+    {
+      "epoch": 0.5035699675210932,
+      "grad_norm": 40.5,
+      "learning_rate": 9.921317325953078e-06,
+      "loss": 17.2108,
+      "step": 27850
+    },
+    {
+      "epoch": 0.5037507825902211,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.921289073646419e-06,
+      "loss": 16.9348,
+      "step": 27860
+    },
+    {
+      "epoch": 0.5039315976593489,
+      "grad_norm": 41.875,
+      "learning_rate": 9.92126082133976e-06,
+      "loss": 17.0763,
+      "step": 27870
+    },
+    {
+      "epoch": 0.5041124127284768,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.921232569033099e-06,
+      "loss": 16.7722,
+      "step": 27880
+    },
+    {
+      "epoch": 0.5042932277976047,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.92120431672644e-06,
+      "loss": 17.0283,
+      "step": 27890
+    },
+    {
+      "epoch": 0.5044740428667325,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.921176064419781e-06,
+      "loss": 17.3908,
+      "step": 27900
+    },
+    {
+      "epoch": 0.5046548579358604,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.92114781211312e-06,
+      "loss": 17.0255,
+      "step": 27910
+    },
+    {
+      "epoch": 0.5048356730049882,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.921119559806461e-06,
+      "loss": 16.8385,
+      "step": 27920
+    },
+    {
+      "epoch": 0.5050164880741161,
+      "grad_norm": 40.25,
+      "learning_rate": 9.921091307499802e-06,
+      "loss": 17.0742,
+      "step": 27930
+    },
+    {
+      "epoch": 0.505197303143244,
+      "grad_norm": 45.1875,
+      "learning_rate": 9.921063055193142e-06,
+      "loss": 17.5109,
+      "step": 27940
+    },
+    {
+      "epoch": 0.5053781182123718,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.921034802886483e-06,
+      "loss": 16.946,
+      "step": 27950
+    },
+    {
+      "epoch": 0.5055589332814997,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.921006550579822e-06,
+      "loss": 16.8611,
+      "step": 27960
+    },
+    {
+      "epoch": 0.5057397483506275,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.920978298273162e-06,
+      "loss": 16.8588,
+      "step": 27970
+    },
+    {
+      "epoch": 0.5059205634197554,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.920950045966505e-06,
+      "loss": 17.1637,
+      "step": 27980
+    },
+    {
+      "epoch": 0.5061013784888833,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.920921793659845e-06,
+      "loss": 16.828,
+      "step": 27990
+    },
+    {
+      "epoch": 0.5062821935580111,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.920893541353184e-06,
+      "loss": 17.0082,
+      "step": 28000
+    },
+    {
+      "epoch": 0.506463008627139,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.920865289046525e-06,
+      "loss": 17.4609,
+      "step": 28010
+    },
+    {
+      "epoch": 0.5066438236962668,
+      "grad_norm": 44.125,
+      "learning_rate": 9.920837036739866e-06,
+      "loss": 17.0481,
+      "step": 28020
+    },
+    {
+      "epoch": 0.5068246387653947,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.920808784433206e-06,
+      "loss": 17.2568,
+      "step": 28030
+    },
+    {
+      "epoch": 0.5070054538345226,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.920780532126547e-06,
+      "loss": 17.1309,
+      "step": 28040
+    },
+    {
+      "epoch": 0.5071862689036505,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.920752279819886e-06,
+      "loss": 16.5908,
+      "step": 28050
+    },
+    {
+      "epoch": 0.5073670839727783,
+      "grad_norm": 42.375,
+      "learning_rate": 9.920724027513226e-06,
+      "loss": 16.8779,
+      "step": 28060
+    },
+    {
+      "epoch": 0.5075478990419061,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.920695775206569e-06,
+      "loss": 17.1186,
+      "step": 28070
+    },
+    {
+      "epoch": 0.507728714111034,
+      "grad_norm": 40.0,
+      "learning_rate": 9.920667522899908e-06,
+      "loss": 16.7213,
+      "step": 28080
+    },
+    {
+      "epoch": 0.5079095291801619,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.920639270593248e-06,
+      "loss": 17.0304,
+      "step": 28090
+    },
+    {
+      "epoch": 0.5080903442492898,
+      "grad_norm": 44.5,
+      "learning_rate": 9.920611018286589e-06,
+      "loss": 17.2263,
+      "step": 28100
+    },
+    {
+      "epoch": 0.5082711593184176,
+      "grad_norm": 39.625,
+      "learning_rate": 9.92058276597993e-06,
+      "loss": 17.4506,
+      "step": 28110
+    },
+    {
+      "epoch": 0.5084519743875454,
+      "grad_norm": 46.53125,
+      "learning_rate": 9.92055451367327e-06,
+      "loss": 17.0361,
+      "step": 28120
+    },
+    {
+      "epoch": 0.5086327894566733,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.920526261366609e-06,
+      "loss": 17.1712,
+      "step": 28130
+    },
+    {
+      "epoch": 0.5088136045258012,
+      "grad_norm": 44.75,
+      "learning_rate": 9.92049800905995e-06,
+      "loss": 17.1423,
+      "step": 28140
+    },
+    {
+      "epoch": 0.5089944195949291,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.92046975675329e-06,
+      "loss": 16.952,
+      "step": 28150
+    },
+    {
+      "epoch": 0.5091752346640569,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.920441504446633e-06,
+      "loss": 17.1141,
+      "step": 28160
+    },
+    {
+      "epoch": 0.5093560497331847,
+      "grad_norm": 37.75,
+      "learning_rate": 9.920413252139971e-06,
+      "loss": 17.0306,
+      "step": 28170
+    },
+    {
+      "epoch": 0.5095368648023126,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.920384999833312e-06,
+      "loss": 17.3226,
+      "step": 28180
+    },
+    {
+      "epoch": 0.5097176798714405,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.920356747526653e-06,
+      "loss": 17.2329,
+      "step": 28190
+    },
+    {
+      "epoch": 0.5098984949405684,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.920328495219993e-06,
+      "loss": 17.0066,
+      "step": 28200
+    },
+    {
+      "epoch": 0.5100793100096962,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.920300242913334e-06,
+      "loss": 17.5045,
+      "step": 28210
+    },
+    {
+      "epoch": 0.5102601250788241,
+      "grad_norm": 41.375,
+      "learning_rate": 9.920271990606673e-06,
+      "loss": 16.7508,
+      "step": 28220
+    },
+    {
+      "epoch": 0.5104409401479519,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.920243738300014e-06,
+      "loss": 17.258,
+      "step": 28230
+    },
+    {
+      "epoch": 0.5106217552170798,
+      "grad_norm": 42.625,
+      "learning_rate": 9.920215485993356e-06,
+      "loss": 17.1619,
+      "step": 28240
+    },
+    {
+      "epoch": 0.5108025702862077,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.920187233686695e-06,
+      "loss": 16.9134,
+      "step": 28250
+    },
+    {
+      "epoch": 0.5109833853553355,
+      "grad_norm": 41.75,
+      "learning_rate": 9.920158981380035e-06,
+      "loss": 17.2607,
+      "step": 28260
+    },
+    {
+      "epoch": 0.5111642004244634,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.920130729073376e-06,
+      "loss": 16.8008,
+      "step": 28270
+    },
+    {
+      "epoch": 0.5113450154935912,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.920102476766717e-06,
+      "loss": 17.2442,
+      "step": 28280
+    },
+    {
+      "epoch": 0.511525830562719,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.920074224460057e-06,
+      "loss": 16.9537,
+      "step": 28290
+    },
+    {
+      "epoch": 0.511706645631847,
+      "grad_norm": 41.75,
+      "learning_rate": 9.920045972153398e-06,
+      "loss": 16.9558,
+      "step": 28300
+    },
+    {
+      "epoch": 0.5118874607009748,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.920017719846737e-06,
+      "loss": 16.932,
+      "step": 28310
+    },
+    {
+      "epoch": 0.5120682757701027,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.919989467540077e-06,
+      "loss": 17.1713,
+      "step": 28320
+    },
+    {
+      "epoch": 0.5122490908392305,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.91996121523342e-06,
+      "loss": 17.1424,
+      "step": 28330
+    },
+    {
+      "epoch": 0.5124299059083584,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.919932962926759e-06,
+      "loss": 16.8539,
+      "step": 28340
+    },
+    {
+      "epoch": 0.5126107209774863,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.9199047106201e-06,
+      "loss": 17.0396,
+      "step": 28350
+    },
+    {
+      "epoch": 0.5127915360466141,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.91987645831344e-06,
+      "loss": 16.9008,
+      "step": 28360
+    },
+    {
+      "epoch": 0.512972351115742,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.91984820600678e-06,
+      "loss": 16.977,
+      "step": 28370
+    },
+    {
+      "epoch": 0.5131531661848698,
+      "grad_norm": 40.5,
+      "learning_rate": 9.919819953700121e-06,
+      "loss": 16.8402,
+      "step": 28380
+    },
+    {
+      "epoch": 0.5133339812539978,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.91979170139346e-06,
+      "loss": 17.1193,
+      "step": 28390
+    },
+    {
+      "epoch": 0.5135147963231256,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.9197634490868e-06,
+      "loss": 17.002,
+      "step": 28400
+    },
+    {
+      "epoch": 0.5136956113922534,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.919735196780141e-06,
+      "loss": 17.2712,
+      "step": 28410
+    },
+    {
+      "epoch": 0.5138764264613813,
+      "grad_norm": 41.75,
+      "learning_rate": 9.919706944473484e-06,
+      "loss": 17.0316,
+      "step": 28420
+    },
+    {
+      "epoch": 0.5140572415305091,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.919678692166823e-06,
+      "loss": 17.0029,
+      "step": 28430
+    },
+    {
+      "epoch": 0.5142380565996371,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.919650439860163e-06,
+      "loss": 16.7383,
+      "step": 28440
+    },
+    {
+      "epoch": 0.5144188716687649,
+      "grad_norm": 42.25,
+      "learning_rate": 9.919622187553504e-06,
+      "loss": 16.9989,
+      "step": 28450
+    },
+    {
+      "epoch": 0.5145996867378927,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.919593935246844e-06,
+      "loss": 17.0851,
+      "step": 28460
+    },
+    {
+      "epoch": 0.5147805018070206,
+      "grad_norm": 40.75,
+      "learning_rate": 9.919565682940185e-06,
+      "loss": 16.9262,
+      "step": 28470
+    },
+    {
+      "epoch": 0.5149613168761484,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.919537430633524e-06,
+      "loss": 16.9286,
+      "step": 28480
+    },
+    {
+      "epoch": 0.5151421319452764,
+      "grad_norm": 40.75,
+      "learning_rate": 9.919509178326865e-06,
+      "loss": 16.9974,
+      "step": 28490
+    },
+    {
+      "epoch": 0.5153229470144042,
+      "grad_norm": 43.125,
+      "learning_rate": 9.919480926020205e-06,
+      "loss": 17.1487,
+      "step": 28500
+    },
+    {
+      "epoch": 0.515503762083532,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.919452673713546e-06,
+      "loss": 16.7718,
+      "step": 28510
+    },
+    {
+      "epoch": 0.5156845771526599,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.919424421406886e-06,
+      "loss": 16.9833,
+      "step": 28520
+    },
+    {
+      "epoch": 0.5158653922217877,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.919396169100227e-06,
+      "loss": 17.0211,
+      "step": 28530
+    },
+    {
+      "epoch": 0.5160462072909157,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.919367916793568e-06,
+      "loss": 17.1867,
+      "step": 28540
+    },
+    {
+      "epoch": 0.5162270223600435,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.919339664486908e-06,
+      "loss": 16.7948,
+      "step": 28550
+    },
+    {
+      "epoch": 0.5164078374291714,
+      "grad_norm": 41.75,
+      "learning_rate": 9.919311412180247e-06,
+      "loss": 17.1321,
+      "step": 28560
+    },
+    {
+      "epoch": 0.5165886524982992,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.919283159873588e-06,
+      "loss": 17.0192,
+      "step": 28570
+    },
+    {
+      "epoch": 0.516769467567427,
+      "grad_norm": 42.25,
+      "learning_rate": 9.919254907566929e-06,
+      "loss": 16.7417,
+      "step": 28580
+    },
+    {
+      "epoch": 0.516950282636555,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.91922665526027e-06,
+      "loss": 17.0249,
+      "step": 28590
+    },
+    {
+      "epoch": 0.5171310977056828,
+      "grad_norm": 38.75,
+      "learning_rate": 9.91919840295361e-06,
+      "loss": 16.8818,
+      "step": 28600
+    },
+    {
+      "epoch": 0.5173119127748107,
+      "grad_norm": 43.875,
+      "learning_rate": 9.91917015064695e-06,
+      "loss": 17.1732,
+      "step": 28610
+    },
+    {
+      "epoch": 0.5174927278439385,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.919141898340291e-06,
+      "loss": 17.4947,
+      "step": 28620
+    },
+    {
+      "epoch": 0.5176735429130663,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.919113646033632e-06,
+      "loss": 17.4457,
+      "step": 28630
+    },
+    {
+      "epoch": 0.5178543579821943,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.919085393726972e-06,
+      "loss": 16.9908,
+      "step": 28640
+    },
+    {
+      "epoch": 0.5180351730513221,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.919057141420311e-06,
+      "loss": 16.6074,
+      "step": 28650
+    },
+    {
+      "epoch": 0.51821598812045,
+      "grad_norm": 42.625,
+      "learning_rate": 9.919028889113652e-06,
+      "loss": 16.7887,
+      "step": 28660
+    },
+    {
+      "epoch": 0.5183968031895778,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.919000636806992e-06,
+      "loss": 17.326,
+      "step": 28670
+    },
+    {
+      "epoch": 0.5185776182587056,
+      "grad_norm": 40.25,
+      "learning_rate": 9.918972384500333e-06,
+      "loss": 16.7892,
+      "step": 28680
+    },
+    {
+      "epoch": 0.5187584333278336,
+      "grad_norm": 40.25,
+      "learning_rate": 9.918944132193674e-06,
+      "loss": 17.2271,
+      "step": 28690
+    },
+    {
+      "epoch": 0.5189392483969614,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.918915879887014e-06,
+      "loss": 17.1251,
+      "step": 28700
+    },
+    {
+      "epoch": 0.5191200634660893,
+      "grad_norm": 41.625,
+      "learning_rate": 9.918887627580355e-06,
+      "loss": 16.8376,
+      "step": 28710
+    },
+    {
+      "epoch": 0.5193008785352171,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.918859375273696e-06,
+      "loss": 17.0451,
+      "step": 28720
+    },
+    {
+      "epoch": 0.519481693604345,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.918831122967034e-06,
+      "loss": 17.2237,
+      "step": 28730
+    },
+    {
+      "epoch": 0.5196625086734729,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.918802870660375e-06,
+      "loss": 17.1952,
+      "step": 28740
+    },
+    {
+      "epoch": 0.5198433237426007,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.918774618353716e-06,
+      "loss": 16.9982,
+      "step": 28750
+    },
+    {
+      "epoch": 0.5200241388117286,
+      "grad_norm": 40.125,
+      "learning_rate": 9.918746366047056e-06,
+      "loss": 17.1002,
+      "step": 28760
+    },
+    {
+      "epoch": 0.5202049538808564,
+      "grad_norm": 43.25,
+      "learning_rate": 9.918718113740397e-06,
+      "loss": 16.8707,
+      "step": 28770
+    },
+    {
+      "epoch": 0.5203857689499843,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.918689861433738e-06,
+      "loss": 17.2035,
+      "step": 28780
+    },
+    {
+      "epoch": 0.5205665840191122,
+      "grad_norm": 45.375,
+      "learning_rate": 9.918661609127078e-06,
+      "loss": 16.9605,
+      "step": 28790
+    },
+    {
+      "epoch": 0.52074739908824,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.918633356820419e-06,
+      "loss": 17.3515,
+      "step": 28800
+    },
+    {
+      "epoch": 0.5209282141573679,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.91860510451376e-06,
+      "loss": 16.8148,
+      "step": 28810
+    },
+    {
+      "epoch": 0.5211090292264957,
+      "grad_norm": 42.75,
+      "learning_rate": 9.918576852207098e-06,
+      "loss": 16.5191,
+      "step": 28820
+    },
+    {
+      "epoch": 0.5212898442956236,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.918548599900439e-06,
+      "loss": 17.1934,
+      "step": 28830
+    },
+    {
+      "epoch": 0.5214706593647515,
+      "grad_norm": 41.5,
+      "learning_rate": 9.91852034759378e-06,
+      "loss": 17.5222,
+      "step": 28840
+    },
+    {
+      "epoch": 0.5216514744338793,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.91849209528712e-06,
+      "loss": 17.0393,
+      "step": 28850
+    },
+    {
+      "epoch": 0.5218322895030072,
+      "grad_norm": 44.875,
+      "learning_rate": 9.918463842980461e-06,
+      "loss": 17.2174,
+      "step": 28860
+    },
+    {
+      "epoch": 0.522013104572135,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.918435590673801e-06,
+      "loss": 17.1666,
+      "step": 28870
+    },
+    {
+      "epoch": 0.5221939196412629,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.918407338367142e-06,
+      "loss": 16.7025,
+      "step": 28880
+    },
+    {
+      "epoch": 0.5223747347103908,
+      "grad_norm": 40.5,
+      "learning_rate": 9.918379086060483e-06,
+      "loss": 16.7877,
+      "step": 28890
+    },
+    {
+      "epoch": 0.5225555497795187,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.918350833753823e-06,
+      "loss": 17.1532,
+      "step": 28900
+    },
+    {
+      "epoch": 0.5227363648486465,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.918322581447162e-06,
+      "loss": 17.0564,
+      "step": 28910
+    },
+    {
+      "epoch": 0.5229171799177743,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.918294329140503e-06,
+      "loss": 16.9364,
+      "step": 28920
+    },
+    {
+      "epoch": 0.5230979949869022,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.918266076833844e-06,
+      "loss": 17.3431,
+      "step": 28930
+    },
+    {
+      "epoch": 0.52327881005603,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.918237824527184e-06,
+      "loss": 16.791,
+      "step": 28940
+    },
+    {
+      "epoch": 0.523459625125158,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.918209572220525e-06,
+      "loss": 17.3021,
+      "step": 28950
+    },
+    {
+      "epoch": 0.5236404401942858,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.918181319913865e-06,
+      "loss": 16.802,
+      "step": 28960
+    },
+    {
+      "epoch": 0.5238212552634136,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.918153067607206e-06,
+      "loss": 17.1349,
+      "step": 28970
+    },
+    {
+      "epoch": 0.5240020703325415,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.918124815300547e-06,
+      "loss": 16.9321,
+      "step": 28980
+    },
+    {
+      "epoch": 0.5241828854016694,
+      "grad_norm": 41.25,
+      "learning_rate": 9.918096562993886e-06,
+      "loss": 17.3189,
+      "step": 28990
+    },
+    {
+      "epoch": 0.5243637004707973,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.918068310687226e-06,
+      "loss": 16.9273,
+      "step": 29000
+    },
+    {
+      "epoch": 0.5245445155399251,
+      "grad_norm": 40.875,
+      "learning_rate": 9.918040058380567e-06,
+      "loss": 17.2656,
+      "step": 29010
+    },
+    {
+      "epoch": 0.5247253306090529,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.918011806073907e-06,
+      "loss": 17.2891,
+      "step": 29020
+    },
+    {
+      "epoch": 0.5249061456781808,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.917983553767248e-06,
+      "loss": 16.7634,
+      "step": 29030
+    },
+    {
+      "epoch": 0.5250869607473087,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.917955301460589e-06,
+      "loss": 16.9445,
+      "step": 29040
+    },
+    {
+      "epoch": 0.5252677758164366,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.91792704915393e-06,
+      "loss": 17.0999,
+      "step": 29050
+    },
+    {
+      "epoch": 0.5254485908855644,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.91789879684727e-06,
+      "loss": 17.1245,
+      "step": 29060
+    },
+    {
+      "epoch": 0.5256294059546923,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.91787054454061e-06,
+      "loss": 16.812,
+      "step": 29070
+    },
+    {
+      "epoch": 0.5258102210238201,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.91784229223395e-06,
+      "loss": 17.0611,
+      "step": 29080
+    },
+    {
+      "epoch": 0.525991036092948,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.91781403992729e-06,
+      "loss": 17.081,
+      "step": 29090
+    },
+    {
+      "epoch": 0.5261718511620759,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.91778578762063e-06,
+      "loss": 16.5785,
+      "step": 29100
+    },
+    {
+      "epoch": 0.5263526662312037,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.917757535313971e-06,
+      "loss": 16.8719,
+      "step": 29110
+    },
+    {
+      "epoch": 0.5265334813003316,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.917729283007312e-06,
+      "loss": 16.9369,
+      "step": 29120
+    },
+    {
+      "epoch": 0.5267142963694594,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.917701030700653e-06,
+      "loss": 17.2821,
+      "step": 29130
+    },
+    {
+      "epoch": 0.5268951114385872,
+      "grad_norm": 38.15625,
+      "learning_rate": 9.917672778393993e-06,
+      "loss": 16.6713,
+      "step": 29140
+    },
+    {
+      "epoch": 0.5270759265077152,
+      "grad_norm": 41.75,
+      "learning_rate": 9.917644526087334e-06,
+      "loss": 16.9598,
+      "step": 29150
+    },
+    {
+      "epoch": 0.527256741576843,
+      "grad_norm": 38.96875,
+      "learning_rate": 9.917616273780673e-06,
+      "loss": 17.0933,
+      "step": 29160
+    },
+    {
+      "epoch": 0.5274375566459709,
+      "grad_norm": 38.3125,
+      "learning_rate": 9.917588021474013e-06,
+      "loss": 17.3192,
+      "step": 29170
+    },
+    {
+      "epoch": 0.5276183717150987,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.917559769167354e-06,
+      "loss": 17.0815,
+      "step": 29180
+    },
+    {
+      "epoch": 0.5277991867842265,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.917531516860695e-06,
+      "loss": 17.0196,
+      "step": 29190
+    },
+    {
+      "epoch": 0.5279800018533545,
+      "grad_norm": 39.875,
+      "learning_rate": 9.917503264554035e-06,
+      "loss": 16.7482,
+      "step": 29200
+    },
+    {
+      "epoch": 0.5281608169224823,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.917475012247376e-06,
+      "loss": 16.6207,
+      "step": 29210
+    },
+    {
+      "epoch": 0.5283416319916102,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.917446759940716e-06,
+      "loss": 17.0133,
+      "step": 29220
+    },
+    {
+      "epoch": 0.528522447060738,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.917418507634057e-06,
+      "loss": 17.2084,
+      "step": 29230
+    },
+    {
+      "epoch": 0.528703262129866,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.917390255327398e-06,
+      "loss": 17.0293,
+      "step": 29240
+    },
+    {
+      "epoch": 0.5288840771989938,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.917362003020737e-06,
+      "loss": 16.935,
+      "step": 29250
+    },
+    {
+      "epoch": 0.5290648922681216,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.917333750714077e-06,
+      "loss": 16.9455,
+      "step": 29260
+    },
+    {
+      "epoch": 0.5292457073372495,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.917305498407418e-06,
+      "loss": 17.1935,
+      "step": 29270
+    },
+    {
+      "epoch": 0.5294265224063773,
+      "grad_norm": 40.625,
+      "learning_rate": 9.917277246100759e-06,
+      "loss": 17.0945,
+      "step": 29280
+    },
+    {
+      "epoch": 0.5296073374755053,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.9172489937941e-06,
+      "loss": 17.1142,
+      "step": 29290
+    },
+    {
+      "epoch": 0.5297881525446331,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.91722074148744e-06,
+      "loss": 17.0181,
+      "step": 29300
+    },
+    {
+      "epoch": 0.5299689676137609,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.91719248918078e-06,
+      "loss": 17.0982,
+      "step": 29310
+    },
+    {
+      "epoch": 0.5301497826828888,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.917164236874121e-06,
+      "loss": 17.1721,
+      "step": 29320
+    },
+    {
+      "epoch": 0.5303305977520166,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.917135984567462e-06,
+      "loss": 17.2262,
+      "step": 29330
+    },
+    {
+      "epoch": 0.5305114128211446,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.9171077322608e-06,
+      "loss": 17.0821,
+      "step": 29340
+    },
+    {
+      "epoch": 0.5306922278902724,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.917079479954141e-06,
+      "loss": 17.1578,
+      "step": 29350
+    },
+    {
+      "epoch": 0.5308730429594002,
+      "grad_norm": 42.25,
+      "learning_rate": 9.917051227647482e-06,
+      "loss": 17.4814,
+      "step": 29360
+    },
+    {
+      "epoch": 0.5310538580285281,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.917022975340822e-06,
+      "loss": 17.022,
+      "step": 29370
+    },
+    {
+      "epoch": 0.5312346730976559,
+      "grad_norm": 41.5,
+      "learning_rate": 9.916994723034163e-06,
+      "loss": 17.2217,
+      "step": 29380
+    },
+    {
+      "epoch": 0.5314154881667839,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.916966470727504e-06,
+      "loss": 17.1552,
+      "step": 29390
+    },
+    {
+      "epoch": 0.5315963032359117,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.916938218420844e-06,
+      "loss": 16.6741,
+      "step": 29400
+    },
+    {
+      "epoch": 0.5317771183050395,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.916909966114185e-06,
+      "loss": 17.2433,
+      "step": 29410
+    },
+    {
+      "epoch": 0.5319579333741674,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.916881713807524e-06,
+      "loss": 16.9512,
+      "step": 29420
+    },
+    {
+      "epoch": 0.5321387484432952,
+      "grad_norm": 45.71875,
+      "learning_rate": 9.916853461500864e-06,
+      "loss": 17.1226,
+      "step": 29430
+    },
+    {
+      "epoch": 0.5323195635124232,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.916825209194205e-06,
+      "loss": 16.9795,
+      "step": 29440
+    },
+    {
+      "epoch": 0.532500378581551,
+      "grad_norm": 41.625,
+      "learning_rate": 9.916796956887546e-06,
+      "loss": 16.7698,
+      "step": 29450
+    },
+    {
+      "epoch": 0.5326811936506789,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.916768704580886e-06,
+      "loss": 17.3518,
+      "step": 29460
+    },
+    {
+      "epoch": 0.5328620087198067,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.916740452274227e-06,
+      "loss": 16.6971,
+      "step": 29470
+    },
+    {
+      "epoch": 0.5330428237889345,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.916712199967568e-06,
+      "loss": 16.7716,
+      "step": 29480
+    },
+    {
+      "epoch": 0.5332236388580625,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.916683947660908e-06,
+      "loss": 17.3883,
+      "step": 29490
+    },
+    {
+      "epoch": 0.5334044539271903,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.916655695354249e-06,
+      "loss": 17.0945,
+      "step": 29500
+    },
+    {
+      "epoch": 0.5335852689963182,
+      "grad_norm": 42.0,
+      "learning_rate": 9.916627443047588e-06,
+      "loss": 16.6286,
+      "step": 29510
+    },
+    {
+      "epoch": 0.533766084065446,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.916599190740928e-06,
+      "loss": 16.7957,
+      "step": 29520
+    },
+    {
+      "epoch": 0.5339468991345738,
+      "grad_norm": 37.21875,
+      "learning_rate": 9.916570938434269e-06,
+      "loss": 16.9727,
+      "step": 29530
+    },
+    {
+      "epoch": 0.5341277142037018,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.91654268612761e-06,
+      "loss": 17.2793,
+      "step": 29540
+    },
+    {
+      "epoch": 0.5343085292728296,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.91651443382095e-06,
+      "loss": 16.8023,
+      "step": 29550
+    },
+    {
+      "epoch": 0.5344893443419575,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.916486181514291e-06,
+      "loss": 16.812,
+      "step": 29560
+    },
+    {
+      "epoch": 0.5346701594110853,
+      "grad_norm": 41.75,
+      "learning_rate": 9.916457929207632e-06,
+      "loss": 17.0138,
+      "step": 29570
+    },
+    {
+      "epoch": 0.5348509744802131,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.916429676900972e-06,
+      "loss": 17.0572,
+      "step": 29580
+    },
+    {
+      "epoch": 0.535031789549341,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.916401424594311e-06,
+      "loss": 16.8175,
+      "step": 29590
+    },
+    {
+      "epoch": 0.5352126046184689,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.916373172287652e-06,
+      "loss": 17.1891,
+      "step": 29600
+    },
+    {
+      "epoch": 0.5353934196875968,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.916344919980992e-06,
+      "loss": 17.1858,
+      "step": 29610
+    },
+    {
+      "epoch": 0.5355742347567246,
+      "grad_norm": 42.875,
+      "learning_rate": 9.916316667674333e-06,
+      "loss": 17.156,
+      "step": 29620
+    },
+    {
+      "epoch": 0.5357550498258525,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.916288415367674e-06,
+      "loss": 16.9915,
+      "step": 29630
+    },
+    {
+      "epoch": 0.5359358648949804,
+      "grad_norm": 41.25,
+      "learning_rate": 9.916260163061014e-06,
+      "loss": 17.008,
+      "step": 29640
+    },
+    {
+      "epoch": 0.5361166799641082,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.916231910754355e-06,
+      "loss": 17.0597,
+      "step": 29650
+    },
+    {
+      "epoch": 0.5362974950332361,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.916203658447695e-06,
+      "loss": 17.3012,
+      "step": 29660
+    },
+    {
+      "epoch": 0.5364783101023639,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.916175406141036e-06,
+      "loss": 16.7412,
+      "step": 29670
+    },
+    {
+      "epoch": 0.5366591251714918,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.916147153834375e-06,
+      "loss": 16.666,
+      "step": 29680
+    },
+    {
+      "epoch": 0.5368399402406197,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.916118901527716e-06,
+      "loss": 17.0808,
+      "step": 29690
+    },
+    {
+      "epoch": 0.5370207553097475,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.916090649221056e-06,
+      "loss": 16.8892,
+      "step": 29700
+    },
+    {
+      "epoch": 0.5372015703788754,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.916062396914397e-06,
+      "loss": 16.88,
+      "step": 29710
+    },
+    {
+      "epoch": 0.5373823854480032,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.916034144607737e-06,
+      "loss": 16.8642,
+      "step": 29720
+    },
+    {
+      "epoch": 0.5375632005171311,
+      "grad_norm": 43.59375,
+      "learning_rate": 9.916005892301078e-06,
+      "loss": 17.3012,
+      "step": 29730
+    },
+    {
+      "epoch": 0.537744015586259,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.915977639994419e-06,
+      "loss": 17.0513,
+      "step": 29740
+    },
+    {
+      "epoch": 0.5379248306553868,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.91594938768776e-06,
+      "loss": 17.2723,
+      "step": 29750
+    },
+    {
+      "epoch": 0.5381056457245147,
+      "grad_norm": 41.625,
+      "learning_rate": 9.9159211353811e-06,
+      "loss": 17.2688,
+      "step": 29760
+    },
+    {
+      "epoch": 0.5382864607936425,
+      "grad_norm": 38.59375,
+      "learning_rate": 9.915892883074439e-06,
+      "loss": 17.1386,
+      "step": 29770
+    },
+    {
+      "epoch": 0.5384672758627704,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.91586463076778e-06,
+      "loss": 16.8733,
+      "step": 29780
+    },
+    {
+      "epoch": 0.5386480909318982,
+      "grad_norm": 40.375,
+      "learning_rate": 9.91583637846112e-06,
+      "loss": 17.0199,
+      "step": 29790
+    },
+    {
+      "epoch": 0.5388289060010262,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.91580812615446e-06,
+      "loss": 17.2014,
+      "step": 29800
+    },
+    {
+      "epoch": 0.539009721070154,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.915779873847801e-06,
+      "loss": 16.9838,
+      "step": 29810
+    },
+    {
+      "epoch": 0.5391905361392818,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.915751621541142e-06,
+      "loss": 17.3729,
+      "step": 29820
+    },
+    {
+      "epoch": 0.5393713512084097,
+      "grad_norm": 41.875,
+      "learning_rate": 9.915723369234483e-06,
+      "loss": 17.075,
+      "step": 29830
+    },
+    {
+      "epoch": 0.5395521662775375,
+      "grad_norm": 38.125,
+      "learning_rate": 9.915695116927823e-06,
+      "loss": 17.0875,
+      "step": 29840
+    },
+    {
+      "epoch": 0.5397329813466655,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.915666864621162e-06,
+      "loss": 16.6237,
+      "step": 29850
+    },
+    {
+      "epoch": 0.5399137964157933,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.915638612314503e-06,
+      "loss": 17.0236,
+      "step": 29860
+    },
+    {
+      "epoch": 0.5400946114849211,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.915610360007843e-06,
+      "loss": 17.3734,
+      "step": 29870
+    },
+    {
+      "epoch": 0.540275426554049,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.915582107701184e-06,
+      "loss": 16.5518,
+      "step": 29880
+    },
+    {
+      "epoch": 0.5404562416231768,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.915553855394525e-06,
+      "loss": 16.4954,
+      "step": 29890
+    },
+    {
+      "epoch": 0.5406370566923048,
+      "grad_norm": 40.5,
+      "learning_rate": 9.915525603087864e-06,
+      "loss": 16.9942,
+      "step": 29900
+    },
+    {
+      "epoch": 0.5408178717614326,
+      "grad_norm": 40.25,
+      "learning_rate": 9.915497350781206e-06,
+      "loss": 16.9093,
+      "step": 29910
+    },
+    {
+      "epoch": 0.5409986868305604,
+      "grad_norm": 40.875,
+      "learning_rate": 9.915469098474547e-06,
+      "loss": 17.0349,
+      "step": 29920
+    },
+    {
+      "epoch": 0.5411795018996883,
+      "grad_norm": 41.625,
+      "learning_rate": 9.915440846167887e-06,
+      "loss": 16.9415,
+      "step": 29930
+    },
+    {
+      "epoch": 0.5413603169688161,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.915412593861226e-06,
+      "loss": 16.9677,
+      "step": 29940
+    },
+    {
+      "epoch": 0.5415411320379441,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.915384341554567e-06,
+      "loss": 17.1662,
+      "step": 29950
+    },
+    {
+      "epoch": 0.5417219471070719,
+      "grad_norm": 46.5,
+      "learning_rate": 9.915356089247907e-06,
+      "loss": 17.0126,
+      "step": 29960
+    },
+    {
+      "epoch": 0.5419027621761998,
+      "grad_norm": 37.96875,
+      "learning_rate": 9.915327836941248e-06,
+      "loss": 17.1567,
+      "step": 29970
+    },
+    {
+      "epoch": 0.5420835772453276,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.915299584634589e-06,
+      "loss": 17.0694,
+      "step": 29980
+    },
+    {
+      "epoch": 0.5422643923144554,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.91527133232793e-06,
+      "loss": 16.7261,
+      "step": 29990
+    },
+    {
+      "epoch": 0.5424452073835834,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.91524308002127e-06,
+      "loss": 16.8142,
+      "step": 30000
+    },
+    {
+      "epoch": 0.5424452073835834,
+      "eval_loss": 2.1271157264709473,
+      "eval_runtime": 229.7564,
+      "eval_samples_per_second": 3160.125,
+      "eval_steps_per_second": 49.378,
+      "step": 30000
+    },
+    {
+      "epoch": 0.5426260224527112,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.91521482771461e-06,
+      "loss": 16.946,
+      "step": 30010
+    },
+    {
+      "epoch": 0.5428068375218391,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.91518657540795e-06,
+      "loss": 17.0405,
+      "step": 30020
+    },
+    {
+      "epoch": 0.5429876525909669,
+      "grad_norm": 40.875,
+      "learning_rate": 9.91515832310129e-06,
+      "loss": 16.5716,
+      "step": 30030
+    },
+    {
+      "epoch": 0.5431684676600947,
+      "grad_norm": 40.25,
+      "learning_rate": 9.91513007079463e-06,
+      "loss": 16.8354,
+      "step": 30040
+    },
+    {
+      "epoch": 0.5433492827292227,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.915101818487971e-06,
+      "loss": 17.1805,
+      "step": 30050
+    },
+    {
+      "epoch": 0.5435300977983505,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.915073566181312e-06,
+      "loss": 17.2318,
+      "step": 30060
+    },
+    {
+      "epoch": 0.5437109128674784,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.915045313874652e-06,
+      "loss": 16.813,
+      "step": 30070
+    },
+    {
+      "epoch": 0.5438917279366062,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.915017061567993e-06,
+      "loss": 16.9205,
+      "step": 30080
+    },
+    {
+      "epoch": 0.544072543005734,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.914988809261334e-06,
+      "loss": 16.9416,
+      "step": 30090
+    },
+    {
+      "epoch": 0.544253358074862,
+      "grad_norm": 42.75,
+      "learning_rate": 9.914960556954674e-06,
+      "loss": 17.3216,
+      "step": 30100
+    },
+    {
+      "epoch": 0.5444341731439898,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.914932304648013e-06,
+      "loss": 17.1828,
+      "step": 30110
+    },
+    {
+      "epoch": 0.5446149882131177,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.914904052341354e-06,
+      "loss": 16.7072,
+      "step": 30120
+    },
+    {
+      "epoch": 0.5447958032822455,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.914875800034694e-06,
+      "loss": 17.0456,
+      "step": 30130
+    },
+    {
+      "epoch": 0.5449766183513735,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.914847547728035e-06,
+      "loss": 16.8821,
+      "step": 30140
+    },
+    {
+      "epoch": 0.5451574334205013,
+      "grad_norm": 41.5,
+      "learning_rate": 9.914819295421376e-06,
+      "loss": 16.8177,
+      "step": 30150
+    },
+    {
+      "epoch": 0.5453382484896291,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.914791043114715e-06,
+      "loss": 16.8663,
+      "step": 30160
+    },
+    {
+      "epoch": 0.545519063558757,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.914762790808057e-06,
+      "loss": 17.1785,
+      "step": 30170
+    },
+    {
+      "epoch": 0.5456998786278848,
+      "grad_norm": 42.25,
+      "learning_rate": 9.914734538501398e-06,
+      "loss": 16.9884,
+      "step": 30180
+    },
+    {
+      "epoch": 0.5458806936970128,
+      "grad_norm": 42.75,
+      "learning_rate": 9.914706286194738e-06,
+      "loss": 16.6989,
+      "step": 30190
+    },
+    {
+      "epoch": 0.5460615087661406,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.914678033888077e-06,
+      "loss": 17.1729,
+      "step": 30200
+    },
+    {
+      "epoch": 0.5462423238352684,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.914649781581418e-06,
+      "loss": 17.0265,
+      "step": 30210
+    },
+    {
+      "epoch": 0.5464231389043963,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.914621529274758e-06,
+      "loss": 17.4852,
+      "step": 30220
+    },
+    {
+      "epoch": 0.5466039539735241,
+      "grad_norm": 41.625,
+      "learning_rate": 9.914593276968099e-06,
+      "loss": 17.1923,
+      "step": 30230
+    },
+    {
+      "epoch": 0.546784769042652,
+      "grad_norm": 41.375,
+      "learning_rate": 9.91456502466144e-06,
+      "loss": 17.2628,
+      "step": 30240
+    },
+    {
+      "epoch": 0.5469655841117799,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.914536772354779e-06,
+      "loss": 17.0136,
+      "step": 30250
+    },
+    {
+      "epoch": 0.5471463991809077,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.914508520048121e-06,
+      "loss": 16.633,
+      "step": 30260
+    },
+    {
+      "epoch": 0.5473272142500356,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.914480267741462e-06,
+      "loss": 16.9213,
+      "step": 30270
+    },
+    {
+      "epoch": 0.5475080293191634,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.9144520154348e-06,
+      "loss": 16.8474,
+      "step": 30280
+    },
+    {
+      "epoch": 0.5476888443882914,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.914423763128141e-06,
+      "loss": 16.8602,
+      "step": 30290
+    },
+    {
+      "epoch": 0.5478696594574192,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.914395510821482e-06,
+      "loss": 17.1963,
+      "step": 30300
+    },
+    {
+      "epoch": 0.5480504745265471,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.914367258514822e-06,
+      "loss": 17.1725,
+      "step": 30310
+    },
+    {
+      "epoch": 0.5482312895956749,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.914339006208163e-06,
+      "loss": 17.3516,
+      "step": 30320
+    },
+    {
+      "epoch": 0.5484121046648027,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.914310753901502e-06,
+      "loss": 17.3431,
+      "step": 30330
+    },
+    {
+      "epoch": 0.5485929197339307,
+      "grad_norm": 38.40625,
+      "learning_rate": 9.914282501594844e-06,
+      "loss": 16.9414,
+      "step": 30340
+    },
+    {
+      "epoch": 0.5487737348030585,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.914254249288185e-06,
+      "loss": 16.793,
+      "step": 30350
+    },
+    {
+      "epoch": 0.5489545498721864,
+      "grad_norm": 43.75,
+      "learning_rate": 9.914225996981525e-06,
+      "loss": 16.7673,
+      "step": 30360
+    },
+    {
+      "epoch": 0.5491353649413142,
+      "grad_norm": 41.625,
+      "learning_rate": 9.914197744674864e-06,
+      "loss": 17.0738,
+      "step": 30370
+    },
+    {
+      "epoch": 0.549316180010442,
+      "grad_norm": 39.75,
+      "learning_rate": 9.914169492368205e-06,
+      "loss": 17.0317,
+      "step": 30380
+    },
+    {
+      "epoch": 0.54949699507957,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.914141240061546e-06,
+      "loss": 17.0928,
+      "step": 30390
+    },
+    {
+      "epoch": 0.5496778101486978,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.914112987754886e-06,
+      "loss": 17.2458,
+      "step": 30400
+    },
+    {
+      "epoch": 0.5498586252178257,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.914084735448227e-06,
+      "loss": 17.1515,
+      "step": 30410
+    },
+    {
+      "epoch": 0.5500394402869535,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.914056483141566e-06,
+      "loss": 17.6208,
+      "step": 30420
+    },
+    {
+      "epoch": 0.5502202553560813,
+      "grad_norm": 42.0,
+      "learning_rate": 9.914028230834908e-06,
+      "loss": 17.0189,
+      "step": 30430
+    },
+    {
+      "epoch": 0.5504010704252092,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.913999978528249e-06,
+      "loss": 16.9412,
+      "step": 30440
+    },
+    {
+      "epoch": 0.5505818854943371,
+      "grad_norm": 44.5,
+      "learning_rate": 9.913971726221588e-06,
+      "loss": 16.8629,
+      "step": 30450
+    },
+    {
+      "epoch": 0.550762700563465,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.913943473914928e-06,
+      "loss": 17.4288,
+      "step": 30460
+    },
+    {
+      "epoch": 0.5509435156325928,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.913915221608269e-06,
+      "loss": 16.7881,
+      "step": 30470
+    },
+    {
+      "epoch": 0.5511243307017207,
+      "grad_norm": 44.125,
+      "learning_rate": 9.91388696930161e-06,
+      "loss": 17.1656,
+      "step": 30480
+    },
+    {
+      "epoch": 0.5513051457708485,
+      "grad_norm": 40.125,
+      "learning_rate": 9.91385871699495e-06,
+      "loss": 16.8119,
+      "step": 30490
+    },
+    {
+      "epoch": 0.5514859608399764,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.91383046468829e-06,
+      "loss": 16.8665,
+      "step": 30500
+    },
+    {
+      "epoch": 0.5516667759091043,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.91380221238163e-06,
+      "loss": 17.2542,
+      "step": 30510
+    },
+    {
+      "epoch": 0.5518475909782321,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.913773960074972e-06,
+      "loss": 16.7592,
+      "step": 30520
+    },
+    {
+      "epoch": 0.55202840604736,
+      "grad_norm": 42.625,
+      "learning_rate": 9.913745707768313e-06,
+      "loss": 16.7518,
+      "step": 30530
+    },
+    {
+      "epoch": 0.5522092211164878,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.913717455461652e-06,
+      "loss": 16.7956,
+      "step": 30540
+    },
+    {
+      "epoch": 0.5523900361856157,
+      "grad_norm": 38.28125,
+      "learning_rate": 9.913689203154992e-06,
+      "loss": 17.1653,
+      "step": 30550
+    },
+    {
+      "epoch": 0.5525708512547436,
+      "grad_norm": 42.75,
+      "learning_rate": 9.913660950848333e-06,
+      "loss": 17.4375,
+      "step": 30560
+    },
+    {
+      "epoch": 0.5527516663238714,
+      "grad_norm": 41.0,
+      "learning_rate": 9.913632698541673e-06,
+      "loss": 16.7577,
+      "step": 30570
+    },
+    {
+      "epoch": 0.5529324813929993,
+      "grad_norm": 44.0,
+      "learning_rate": 9.913604446235014e-06,
+      "loss": 16.8749,
+      "step": 30580
+    },
+    {
+      "epoch": 0.5531132964621271,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.913576193928353e-06,
+      "loss": 17.3017,
+      "step": 30590
+    },
+    {
+      "epoch": 0.553294111531255,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.913547941621694e-06,
+      "loss": 16.8087,
+      "step": 30600
+    },
+    {
+      "epoch": 0.5534749266003829,
+      "grad_norm": 45.84375,
+      "learning_rate": 9.913519689315036e-06,
+      "loss": 16.8531,
+      "step": 30610
+    },
+    {
+      "epoch": 0.5536557416695107,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.913491437008377e-06,
+      "loss": 16.864,
+      "step": 30620
+    },
+    {
+      "epoch": 0.5538365567386386,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.913463184701715e-06,
+      "loss": 16.9866,
+      "step": 30630
+    },
+    {
+      "epoch": 0.5540173718077664,
+      "grad_norm": 42.25,
+      "learning_rate": 9.913434932395056e-06,
+      "loss": 17.0031,
+      "step": 30640
+    },
+    {
+      "epoch": 0.5541981868768944,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.913406680088397e-06,
+      "loss": 16.7895,
+      "step": 30650
+    },
+    {
+      "epoch": 0.5543790019460222,
+      "grad_norm": 43.5,
+      "learning_rate": 9.913378427781737e-06,
+      "loss": 17.0026,
+      "step": 30660
+    },
+    {
+      "epoch": 0.55455981701515,
+      "grad_norm": 40.25,
+      "learning_rate": 9.913350175475078e-06,
+      "loss": 17.2128,
+      "step": 30670
+    },
+    {
+      "epoch": 0.5547406320842779,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.913321923168417e-06,
+      "loss": 17.0078,
+      "step": 30680
+    },
+    {
+      "epoch": 0.5549214471534057,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.91329367086176e-06,
+      "loss": 17.1819,
+      "step": 30690
+    },
+    {
+      "epoch": 0.5551022622225337,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.9132654185551e-06,
+      "loss": 16.6908,
+      "step": 30700
+    },
+    {
+      "epoch": 0.5552830772916615,
+      "grad_norm": 43.5,
+      "learning_rate": 9.913237166248439e-06,
+      "loss": 17.0772,
+      "step": 30710
+    },
+    {
+      "epoch": 0.5554638923607893,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.91320891394178e-06,
+      "loss": 17.0964,
+      "step": 30720
+    },
+    {
+      "epoch": 0.5556447074299172,
+      "grad_norm": 43.875,
+      "learning_rate": 9.91318066163512e-06,
+      "loss": 16.7243,
+      "step": 30730
+    },
+    {
+      "epoch": 0.555825522499045,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.91315240932846e-06,
+      "loss": 17.2681,
+      "step": 30740
+    },
+    {
+      "epoch": 0.556006337568173,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.913124157021801e-06,
+      "loss": 17.031,
+      "step": 30750
+    },
+    {
+      "epoch": 0.5561871526373008,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.91309590471514e-06,
+      "loss": 16.7159,
+      "step": 30760
+    },
+    {
+      "epoch": 0.5563679677064286,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.91306765240848e-06,
+      "loss": 16.9925,
+      "step": 30770
+    },
+    {
+      "epoch": 0.5565487827755565,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.913039400101823e-06,
+      "loss": 16.9813,
+      "step": 30780
+    },
+    {
+      "epoch": 0.5567295978446843,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.913011147795164e-06,
+      "loss": 16.9735,
+      "step": 30790
+    },
+    {
+      "epoch": 0.5569104129138123,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.912982895488503e-06,
+      "loss": 17.1974,
+      "step": 30800
+    },
+    {
+      "epoch": 0.5570912279829401,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.912954643181843e-06,
+      "loss": 17.0436,
+      "step": 30810
+    },
+    {
+      "epoch": 0.557272043052068,
+      "grad_norm": 43.375,
+      "learning_rate": 9.912926390875184e-06,
+      "loss": 17.1123,
+      "step": 30820
+    },
+    {
+      "epoch": 0.5574528581211958,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.912898138568525e-06,
+      "loss": 16.9163,
+      "step": 30830
+    },
+    {
+      "epoch": 0.5576336731903236,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.912869886261865e-06,
+      "loss": 17.1751,
+      "step": 30840
+    },
+    {
+      "epoch": 0.5578144882594516,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.912841633955204e-06,
+      "loss": 16.9831,
+      "step": 30850
+    },
+    {
+      "epoch": 0.5579953033285794,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.912813381648545e-06,
+      "loss": 16.887,
+      "step": 30860
+    },
+    {
+      "epoch": 0.5581761183977073,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.912785129341887e-06,
+      "loss": 16.9941,
+      "step": 30870
+    },
+    {
+      "epoch": 0.5583569334668351,
+      "grad_norm": 44.25,
+      "learning_rate": 9.912756877035226e-06,
+      "loss": 16.9951,
+      "step": 30880
+    },
+    {
+      "epoch": 0.5585377485359629,
+      "grad_norm": 43.5,
+      "learning_rate": 9.912728624728567e-06,
+      "loss": 17.0885,
+      "step": 30890
+    },
+    {
+      "epoch": 0.5587185636050909,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.912700372421907e-06,
+      "loss": 16.9381,
+      "step": 30900
+    },
+    {
+      "epoch": 0.5588993786742187,
+      "grad_norm": 45.90625,
+      "learning_rate": 9.912672120115248e-06,
+      "loss": 16.8079,
+      "step": 30910
+    },
+    {
+      "epoch": 0.5590801937433466,
+      "grad_norm": 45.1875,
+      "learning_rate": 9.912643867808588e-06,
+      "loss": 17.1373,
+      "step": 30920
+    },
+    {
+      "epoch": 0.5592610088124744,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.912615615501929e-06,
+      "loss": 17.3978,
+      "step": 30930
+    },
+    {
+      "epoch": 0.5594418238816022,
+      "grad_norm": 40.0,
+      "learning_rate": 9.912587363195268e-06,
+      "loss": 17.0463,
+      "step": 30940
+    },
+    {
+      "epoch": 0.5596226389507302,
+      "grad_norm": 40.25,
+      "learning_rate": 9.912559110888609e-06,
+      "loss": 17.0721,
+      "step": 30950
+    },
+    {
+      "epoch": 0.559803454019858,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.912530858581951e-06,
+      "loss": 17.0719,
+      "step": 30960
+    },
+    {
+      "epoch": 0.5599842690889859,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.91250260627529e-06,
+      "loss": 17.2505,
+      "step": 30970
+    },
+    {
+      "epoch": 0.5601650841581137,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.91247435396863e-06,
+      "loss": 17.0649,
+      "step": 30980
+    },
+    {
+      "epoch": 0.5603458992272417,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.912446101661971e-06,
+      "loss": 17.0292,
+      "step": 30990
+    },
+    {
+      "epoch": 0.5605267142963695,
+      "grad_norm": 45.75,
+      "learning_rate": 9.912417849355312e-06,
+      "loss": 16.998,
+      "step": 31000
+    },
+    {
+      "epoch": 0.5607075293654973,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.912389597048652e-06,
+      "loss": 16.8837,
+      "step": 31010
+    },
+    {
+      "epoch": 0.5608883444346252,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.912361344741991e-06,
+      "loss": 17.1694,
+      "step": 31020
+    },
+    {
+      "epoch": 0.561069159503753,
+      "grad_norm": 40.5,
+      "learning_rate": 9.912333092435332e-06,
+      "loss": 16.8809,
+      "step": 31030
+    },
+    {
+      "epoch": 0.561249974572881,
+      "grad_norm": 42.125,
+      "learning_rate": 9.912304840128674e-06,
+      "loss": 17.1144,
+      "step": 31040
+    },
+    {
+      "epoch": 0.5614307896420088,
+      "grad_norm": 40.25,
+      "learning_rate": 9.912276587822015e-06,
+      "loss": 17.2917,
+      "step": 31050
+    },
+    {
+      "epoch": 0.5616116047111366,
+      "grad_norm": 44.6875,
+      "learning_rate": 9.912248335515354e-06,
+      "loss": 16.9495,
+      "step": 31060
+    },
+    {
+      "epoch": 0.5617924197802645,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.912220083208694e-06,
+      "loss": 17.3728,
+      "step": 31070
+    },
+    {
+      "epoch": 0.5619732348493923,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.912191830902035e-06,
+      "loss": 16.7284,
+      "step": 31080
+    },
+    {
+      "epoch": 0.5621540499185202,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.912163578595376e-06,
+      "loss": 16.6271,
+      "step": 31090
+    },
+    {
+      "epoch": 0.5623348649876481,
+      "grad_norm": 40.375,
+      "learning_rate": 9.912135326288716e-06,
+      "loss": 16.8732,
+      "step": 31100
+    },
+    {
+      "epoch": 0.5625156800567759,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.912107073982055e-06,
+      "loss": 16.6208,
+      "step": 31110
+    },
+    {
+      "epoch": 0.5626964951259038,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.912078821675396e-06,
+      "loss": 17.2431,
+      "step": 31120
+    },
+    {
+      "epoch": 0.5628773101950316,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.912050569368738e-06,
+      "loss": 16.7951,
+      "step": 31130
+    },
+    {
+      "epoch": 0.5630581252641595,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.912022317062077e-06,
+      "loss": 17.2329,
+      "step": 31140
+    },
+    {
+      "epoch": 0.5632389403332874,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.911994064755418e-06,
+      "loss": 16.9604,
+      "step": 31150
+    },
+    {
+      "epoch": 0.5634197554024153,
+      "grad_norm": 43.0,
+      "learning_rate": 9.911965812448758e-06,
+      "loss": 16.9863,
+      "step": 31160
+    },
+    {
+      "epoch": 0.5636005704715431,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.911937560142099e-06,
+      "loss": 16.8509,
+      "step": 31170
+    },
+    {
+      "epoch": 0.5637813855406709,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.91190930783544e-06,
+      "loss": 16.956,
+      "step": 31180
+    },
+    {
+      "epoch": 0.5639622006097988,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.911881055528778e-06,
+      "loss": 16.8709,
+      "step": 31190
+    },
+    {
+      "epoch": 0.5641430156789267,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.911852803222119e-06,
+      "loss": 16.7424,
+      "step": 31200
+    },
+    {
+      "epoch": 0.5643238307480546,
+      "grad_norm": 43.0,
+      "learning_rate": 9.91182455091546e-06,
+      "loss": 16.9176,
+      "step": 31210
+    },
+    {
+      "epoch": 0.5645046458171824,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.911796298608802e-06,
+      "loss": 17.2313,
+      "step": 31220
+    },
+    {
+      "epoch": 0.5646854608863102,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.911768046302141e-06,
+      "loss": 17.1907,
+      "step": 31230
+    },
+    {
+      "epoch": 0.5648662759554381,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.911739793995482e-06,
+      "loss": 16.7637,
+      "step": 31240
+    },
+    {
+      "epoch": 0.565047091024566,
+      "grad_norm": 41.0,
+      "learning_rate": 9.911711541688822e-06,
+      "loss": 17.0071,
+      "step": 31250
+    },
+    {
+      "epoch": 0.5652279060936939,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.911683289382163e-06,
+      "loss": 17.097,
+      "step": 31260
+    },
+    {
+      "epoch": 0.5654087211628217,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.911655037075503e-06,
+      "loss": 17.3105,
+      "step": 31270
+    },
+    {
+      "epoch": 0.5655895362319495,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.911626784768842e-06,
+      "loss": 17.0906,
+      "step": 31280
+    },
+    {
+      "epoch": 0.5657703513010774,
+      "grad_norm": 39.96875,
+      "learning_rate": 9.911598532462183e-06,
+      "loss": 17.2088,
+      "step": 31290
+    },
+    {
+      "epoch": 0.5659511663702053,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.911570280155524e-06,
+      "loss": 16.8181,
+      "step": 31300
+    },
+    {
+      "epoch": 0.5661319814393332,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.911542027848864e-06,
+      "loss": 17.0384,
+      "step": 31310
+    },
+    {
+      "epoch": 0.566312796508461,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.911513775542205e-06,
+      "loss": 16.741,
+      "step": 31320
+    },
+    {
+      "epoch": 0.5664936115775889,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.911485523235545e-06,
+      "loss": 16.7638,
+      "step": 31330
+    },
+    {
+      "epoch": 0.5666744266467167,
+      "grad_norm": 43.75,
+      "learning_rate": 9.911457270928886e-06,
+      "loss": 16.7134,
+      "step": 31340
+    },
+    {
+      "epoch": 0.5668552417158446,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.911429018622227e-06,
+      "loss": 16.9839,
+      "step": 31350
+    },
+    {
+      "epoch": 0.5670360567849725,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.911400766315566e-06,
+      "loss": 16.8958,
+      "step": 31360
+    },
+    {
+      "epoch": 0.5672168718541003,
+      "grad_norm": 41.875,
+      "learning_rate": 9.911372514008906e-06,
+      "loss": 16.3046,
+      "step": 31370
+    },
+    {
+      "epoch": 0.5673976869232282,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.911344261702247e-06,
+      "loss": 17.342,
+      "step": 31380
+    },
+    {
+      "epoch": 0.567578501992356,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.91131600939559e-06,
+      "loss": 17.1161,
+      "step": 31390
+    },
+    {
+      "epoch": 0.5677593170614839,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.911287757088928e-06,
+      "loss": 16.7149,
+      "step": 31400
+    },
+    {
+      "epoch": 0.5679401321306118,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.911259504782269e-06,
+      "loss": 16.841,
+      "step": 31410
+    },
+    {
+      "epoch": 0.5681209471997396,
+      "grad_norm": 40.625,
+      "learning_rate": 9.91123125247561e-06,
+      "loss": 16.8366,
+      "step": 31420
+    },
+    {
+      "epoch": 0.5683017622688675,
+      "grad_norm": 40.125,
+      "learning_rate": 9.91120300016895e-06,
+      "loss": 17.2759,
+      "step": 31430
+    },
+    {
+      "epoch": 0.5684825773379953,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.91117474786229e-06,
+      "loss": 16.9346,
+      "step": 31440
+    },
+    {
+      "epoch": 0.5686633924071232,
+      "grad_norm": 40.5,
+      "learning_rate": 9.91114649555563e-06,
+      "loss": 16.5734,
+      "step": 31450
+    },
+    {
+      "epoch": 0.5688442074762511,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.91111824324897e-06,
+      "loss": 16.8365,
+      "step": 31460
+    },
+    {
+      "epoch": 0.5690250225453789,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.91108999094231e-06,
+      "loss": 17.1106,
+      "step": 31470
+    },
+    {
+      "epoch": 0.5692058376145068,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.911061738635651e-06,
+      "loss": 16.7902,
+      "step": 31480
+    },
+    {
+      "epoch": 0.5693866526836346,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.911033486328992e-06,
+      "loss": 16.4987,
+      "step": 31490
+    },
+    {
+      "epoch": 0.5695674677527626,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.911005234022333e-06,
+      "loss": 16.7677,
+      "step": 31500
+    },
+    {
+      "epoch": 0.5697482828218904,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.910976981715673e-06,
+      "loss": 16.971,
+      "step": 31510
+    },
+    {
+      "epoch": 0.5699290978910182,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.910948729409014e-06,
+      "loss": 16.8891,
+      "step": 31520
+    },
+    {
+      "epoch": 0.5701099129601461,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.910920477102355e-06,
+      "loss": 16.5515,
+      "step": 31530
+    },
+    {
+      "epoch": 0.5702907280292739,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.910892224795693e-06,
+      "loss": 17.0996,
+      "step": 31540
+    },
+    {
+      "epoch": 0.5704715430984019,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.910863972489034e-06,
+      "loss": 17.3,
+      "step": 31550
+    },
+    {
+      "epoch": 0.5706523581675297,
+      "grad_norm": 41.875,
+      "learning_rate": 9.910835720182375e-06,
+      "loss": 16.7294,
+      "step": 31560
+    },
+    {
+      "epoch": 0.5708331732366575,
+      "grad_norm": 41.25,
+      "learning_rate": 9.910807467875715e-06,
+      "loss": 17.0507,
+      "step": 31570
+    },
+    {
+      "epoch": 0.5710139883057854,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.910779215569056e-06,
+      "loss": 16.7798,
+      "step": 31580
+    },
+    {
+      "epoch": 0.5711948033749132,
+      "grad_norm": 40.875,
+      "learning_rate": 9.910750963262397e-06,
+      "loss": 16.729,
+      "step": 31590
+    },
+    {
+      "epoch": 0.5713756184440412,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.910722710955737e-06,
+      "loss": 16.7486,
+      "step": 31600
+    },
+    {
+      "epoch": 0.571556433513169,
+      "grad_norm": 45.0,
+      "learning_rate": 9.910694458649078e-06,
+      "loss": 17.0809,
+      "step": 31610
+    },
+    {
+      "epoch": 0.5717372485822968,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.910666206342417e-06,
+      "loss": 17.0275,
+      "step": 31620
+    },
+    {
+      "epoch": 0.5719180636514247,
+      "grad_norm": 40.5,
+      "learning_rate": 9.910637954035757e-06,
+      "loss": 16.9974,
+      "step": 31630
+    },
+    {
+      "epoch": 0.5720988787205525,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.910609701729098e-06,
+      "loss": 16.9911,
+      "step": 31640
+    },
+    {
+      "epoch": 0.5722796937896805,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.910581449422439e-06,
+      "loss": 16.7812,
+      "step": 31650
+    },
+    {
+      "epoch": 0.5724605088588083,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.91055319711578e-06,
+      "loss": 16.8301,
+      "step": 31660
+    },
+    {
+      "epoch": 0.5726413239279362,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.91052494480912e-06,
+      "loss": 16.7667,
+      "step": 31670
+    },
+    {
+      "epoch": 0.572822138997064,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.91049669250246e-06,
+      "loss": 16.5191,
+      "step": 31680
+    },
+    {
+      "epoch": 0.5730029540661918,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.910468440195801e-06,
+      "loss": 17.1354,
+      "step": 31690
+    },
+    {
+      "epoch": 0.5731837691353198,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.910440187889142e-06,
+      "loss": 16.8287,
+      "step": 31700
+    },
+    {
+      "epoch": 0.5733645842044476,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.91041193558248e-06,
+      "loss": 17.183,
+      "step": 31710
+    },
+    {
+      "epoch": 0.5735453992735755,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.910383683275821e-06,
+      "loss": 17.053,
+      "step": 31720
+    },
+    {
+      "epoch": 0.5737262143427033,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.910355430969162e-06,
+      "loss": 17.1543,
+      "step": 31730
+    },
+    {
+      "epoch": 0.5739070294118311,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.910327178662503e-06,
+      "loss": 16.8542,
+      "step": 31740
+    },
+    {
+      "epoch": 0.5740878444809591,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.910298926355843e-06,
+      "loss": 16.7303,
+      "step": 31750
+    },
+    {
+      "epoch": 0.5742686595500869,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.910270674049184e-06,
+      "loss": 16.62,
+      "step": 31760
+    },
+    {
+      "epoch": 0.5744494746192148,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.910242421742524e-06,
+      "loss": 17.0608,
+      "step": 31770
+    },
+    {
+      "epoch": 0.5746302896883426,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.910214169435865e-06,
+      "loss": 16.7006,
+      "step": 31780
+    },
+    {
+      "epoch": 0.5748111047574704,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.910185917129204e-06,
+      "loss": 17.0885,
+      "step": 31790
+    },
+    {
+      "epoch": 0.5749919198265984,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.910157664822545e-06,
+      "loss": 16.7022,
+      "step": 31800
+    },
+    {
+      "epoch": 0.5751727348957262,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.910129412515885e-06,
+      "loss": 16.9512,
+      "step": 31810
+    },
+    {
+      "epoch": 0.5753535499648541,
+      "grad_norm": 42.375,
+      "learning_rate": 9.910101160209226e-06,
+      "loss": 16.9256,
+      "step": 31820
+    },
+    {
+      "epoch": 0.5755343650339819,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.910072907902566e-06,
+      "loss": 16.6284,
+      "step": 31830
+    },
+    {
+      "epoch": 0.5757151801031098,
+      "grad_norm": 41.25,
+      "learning_rate": 9.910044655595907e-06,
+      "loss": 16.9051,
+      "step": 31840
+    },
+    {
+      "epoch": 0.5758959951722377,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.910016403289248e-06,
+      "loss": 16.7478,
+      "step": 31850
+    },
+    {
+      "epoch": 0.5760768102413655,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.909988150982588e-06,
+      "loss": 16.8593,
+      "step": 31860
+    },
+    {
+      "epoch": 0.5762576253104934,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.909959898675929e-06,
+      "loss": 16.7016,
+      "step": 31870
+    },
+    {
+      "epoch": 0.5764384403796212,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.909931646369268e-06,
+      "loss": 17.125,
+      "step": 31880
+    },
+    {
+      "epoch": 0.5766192554487491,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.909903394062608e-06,
+      "loss": 16.9871,
+      "step": 31890
+    },
+    {
+      "epoch": 0.576800070517877,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.909875141755949e-06,
+      "loss": 17.0037,
+      "step": 31900
+    },
+    {
+      "epoch": 0.5769808855870048,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.90984688944929e-06,
+      "loss": 17.2558,
+      "step": 31910
+    },
+    {
+      "epoch": 0.5771617006561327,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.90981863714263e-06,
+      "loss": 17.2594,
+      "step": 31920
+    },
+    {
+      "epoch": 0.5773425157252605,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.909790384835971e-06,
+      "loss": 16.8757,
+      "step": 31930
+    },
+    {
+      "epoch": 0.5775233307943884,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.909762132529312e-06,
+      "loss": 16.9364,
+      "step": 31940
+    },
+    {
+      "epoch": 0.5777041458635163,
+      "grad_norm": 41.875,
+      "learning_rate": 9.909733880222652e-06,
+      "loss": 16.6703,
+      "step": 31950
+    },
+    {
+      "epoch": 0.5778849609326441,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.909705627915993e-06,
+      "loss": 16.8876,
+      "step": 31960
+    },
+    {
+      "epoch": 0.578065776001772,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.909677375609332e-06,
+      "loss": 16.8418,
+      "step": 31970
+    },
+    {
+      "epoch": 0.5782465910708998,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.909649123302672e-06,
+      "loss": 16.9372,
+      "step": 31980
+    },
+    {
+      "epoch": 0.5784274061400277,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.909620870996013e-06,
+      "loss": 16.7588,
+      "step": 31990
+    },
+    {
+      "epoch": 0.5786082212091556,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.909592618689354e-06,
+      "loss": 17.0603,
+      "step": 32000
+    },
+    {
+      "epoch": 0.5787890362782835,
+      "grad_norm": 45.75,
+      "learning_rate": 9.909564366382694e-06,
+      "loss": 17.18,
+      "step": 32010
+    },
+    {
+      "epoch": 0.5789698513474113,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.909536114076035e-06,
+      "loss": 16.7754,
+      "step": 32020
+    },
+    {
+      "epoch": 0.5791506664165391,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.909507861769375e-06,
+      "loss": 17.1534,
+      "step": 32030
+    },
+    {
+      "epoch": 0.579331481485667,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.909479609462716e-06,
+      "loss": 17.2115,
+      "step": 32040
+    },
+    {
+      "epoch": 0.5795122965547949,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.909451357156055e-06,
+      "loss": 17.0768,
+      "step": 32050
+    },
+    {
+      "epoch": 0.5796931116239228,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.909423104849396e-06,
+      "loss": 17.2024,
+      "step": 32060
+    },
+    {
+      "epoch": 0.5798739266930506,
+      "grad_norm": 45.84375,
+      "learning_rate": 9.909394852542736e-06,
+      "loss": 16.6366,
+      "step": 32070
+    },
+    {
+      "epoch": 0.5800547417621784,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.909366600236077e-06,
+      "loss": 17.1224,
+      "step": 32080
+    },
+    {
+      "epoch": 0.5802355568313063,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.909338347929418e-06,
+      "loss": 16.9536,
+      "step": 32090
+    },
+    {
+      "epoch": 0.5804163719004342,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.909310095622758e-06,
+      "loss": 16.8179,
+      "step": 32100
+    },
+    {
+      "epoch": 0.5805971869695621,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.909281843316099e-06,
+      "loss": 17.0561,
+      "step": 32110
+    },
+    {
+      "epoch": 0.5807780020386899,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.90925359100944e-06,
+      "loss": 16.8479,
+      "step": 32120
+    },
+    {
+      "epoch": 0.5809588171078177,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.90922533870278e-06,
+      "loss": 17.2642,
+      "step": 32130
+    },
+    {
+      "epoch": 0.5811396321769456,
+      "grad_norm": 41.5,
+      "learning_rate": 9.909197086396119e-06,
+      "loss": 17.1699,
+      "step": 32140
+    },
+    {
+      "epoch": 0.5813204472460735,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.90916883408946e-06,
+      "loss": 16.8568,
+      "step": 32150
+    },
+    {
+      "epoch": 0.5815012623152014,
+      "grad_norm": 39.0,
+      "learning_rate": 9.9091405817828e-06,
+      "loss": 16.9422,
+      "step": 32160
+    },
+    {
+      "epoch": 0.5816820773843292,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.90911232947614e-06,
+      "loss": 16.7037,
+      "step": 32170
+    },
+    {
+      "epoch": 0.5818628924534571,
+      "grad_norm": 42.125,
+      "learning_rate": 9.909084077169481e-06,
+      "loss": 17.1012,
+      "step": 32180
+    },
+    {
+      "epoch": 0.5820437075225849,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.909055824862822e-06,
+      "loss": 16.9235,
+      "step": 32190
+    },
+    {
+      "epoch": 0.5822245225917128,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.909027572556163e-06,
+      "loss": 17.0028,
+      "step": 32200
+    },
+    {
+      "epoch": 0.5824053376608407,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.908999320249503e-06,
+      "loss": 17.3045,
+      "step": 32210
+    },
+    {
+      "epoch": 0.5825861527299685,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.908971067942842e-06,
+      "loss": 16.6335,
+      "step": 32220
+    },
+    {
+      "epoch": 0.5827669677990964,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.908942815636183e-06,
+      "loss": 17.434,
+      "step": 32230
+    },
+    {
+      "epoch": 0.5829477828682242,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.908914563329523e-06,
+      "loss": 16.849,
+      "step": 32240
+    },
+    {
+      "epoch": 0.5831285979373521,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.908886311022864e-06,
+      "loss": 17.0407,
+      "step": 32250
+    },
+    {
+      "epoch": 0.58330941300648,
+      "grad_norm": 43.0,
+      "learning_rate": 9.908858058716205e-06,
+      "loss": 16.7928,
+      "step": 32260
+    },
+    {
+      "epoch": 0.5834902280756078,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.908829806409545e-06,
+      "loss": 16.5376,
+      "step": 32270
+    },
+    {
+      "epoch": 0.5836710431447357,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.908801554102886e-06,
+      "loss": 16.9865,
+      "step": 32280
+    },
+    {
+      "epoch": 0.5838518582138635,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.908773301796227e-06,
+      "loss": 16.6478,
+      "step": 32290
+    },
+    {
+      "epoch": 0.5840326732829914,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.908745049489567e-06,
+      "loss": 17.0553,
+      "step": 32300
+    },
+    {
+      "epoch": 0.5842134883521193,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.908716797182906e-06,
+      "loss": 16.9444,
+      "step": 32310
+    },
+    {
+      "epoch": 0.5843943034212471,
+      "grad_norm": 45.0,
+      "learning_rate": 9.908688544876247e-06,
+      "loss": 17.1495,
+      "step": 32320
+    },
+    {
+      "epoch": 0.584575118490375,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.908660292569587e-06,
+      "loss": 17.0832,
+      "step": 32330
+    },
+    {
+      "epoch": 0.5847559335595028,
+      "grad_norm": 38.375,
+      "learning_rate": 9.908632040262928e-06,
+      "loss": 16.2737,
+      "step": 32340
+    },
+    {
+      "epoch": 0.5849367486286308,
+      "grad_norm": 44.0,
+      "learning_rate": 9.908603787956269e-06,
+      "loss": 16.8861,
+      "step": 32350
+    },
+    {
+      "epoch": 0.5851175636977586,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.90857553564961e-06,
+      "loss": 16.8836,
+      "step": 32360
+    },
+    {
+      "epoch": 0.5852983787668864,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.90854728334295e-06,
+      "loss": 17.1306,
+      "step": 32370
+    },
+    {
+      "epoch": 0.5854791938360143,
+      "grad_norm": 43.375,
+      "learning_rate": 9.90851903103629e-06,
+      "loss": 17.097,
+      "step": 32380
+    },
+    {
+      "epoch": 0.5856600089051421,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.908490778729631e-06,
+      "loss": 16.9224,
+      "step": 32390
+    },
+    {
+      "epoch": 0.5858408239742701,
+      "grad_norm": 45.375,
+      "learning_rate": 9.90846252642297e-06,
+      "loss": 16.8028,
+      "step": 32400
+    },
+    {
+      "epoch": 0.5860216390433979,
+      "grad_norm": 39.15625,
+      "learning_rate": 9.90843427411631e-06,
+      "loss": 16.8687,
+      "step": 32410
+    },
+    {
+      "epoch": 0.5862024541125257,
+      "grad_norm": 39.375,
+      "learning_rate": 9.908406021809651e-06,
+      "loss": 16.8831,
+      "step": 32420
+    },
+    {
+      "epoch": 0.5863832691816536,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.908377769502992e-06,
+      "loss": 16.6693,
+      "step": 32430
+    },
+    {
+      "epoch": 0.5865640842507814,
+      "grad_norm": 41.125,
+      "learning_rate": 9.908349517196333e-06,
+      "loss": 16.9407,
+      "step": 32440
+    },
+    {
+      "epoch": 0.5867448993199094,
+      "grad_norm": 46.59375,
+      "learning_rate": 9.908321264889673e-06,
+      "loss": 17.4105,
+      "step": 32450
+    },
+    {
+      "epoch": 0.5869257143890372,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.908293012583014e-06,
+      "loss": 16.7872,
+      "step": 32460
+    },
+    {
+      "epoch": 0.587106529458165,
+      "grad_norm": 40.0,
+      "learning_rate": 9.908264760276354e-06,
+      "loss": 17.1423,
+      "step": 32470
+    },
+    {
+      "epoch": 0.5872873445272929,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.908236507969693e-06,
+      "loss": 17.2344,
+      "step": 32480
+    },
+    {
+      "epoch": 0.5874681595964207,
+      "grad_norm": 40.5,
+      "learning_rate": 9.908208255663034e-06,
+      "loss": 17.0029,
+      "step": 32490
+    },
+    {
+      "epoch": 0.5876489746655487,
+      "grad_norm": 46.3125,
+      "learning_rate": 9.908180003356375e-06,
+      "loss": 17.0556,
+      "step": 32500
+    },
+    {
+      "epoch": 0.5878297897346765,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.908151751049715e-06,
+      "loss": 16.9346,
+      "step": 32510
+    },
+    {
+      "epoch": 0.5880106048038044,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.908123498743056e-06,
+      "loss": 16.9969,
+      "step": 32520
+    },
+    {
+      "epoch": 0.5881914198729322,
+      "grad_norm": 45.1875,
+      "learning_rate": 9.908095246436396e-06,
+      "loss": 17.1584,
+      "step": 32530
+    },
+    {
+      "epoch": 0.58837223494206,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.908066994129737e-06,
+      "loss": 16.9199,
+      "step": 32540
+    },
+    {
+      "epoch": 0.588553050011188,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.908038741823078e-06,
+      "loss": 16.8951,
+      "step": 32550
+    },
+    {
+      "epoch": 0.5887338650803158,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.908010489516418e-06,
+      "loss": 17.4404,
+      "step": 32560
+    },
+    {
+      "epoch": 0.5889146801494437,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.907982237209757e-06,
+      "loss": 17.1576,
+      "step": 32570
+    },
+    {
+      "epoch": 0.5890954952185715,
+      "grad_norm": 42.625,
+      "learning_rate": 9.907953984903098e-06,
+      "loss": 16.656,
+      "step": 32580
+    },
+    {
+      "epoch": 0.5892763102876993,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.907925732596438e-06,
+      "loss": 16.8173,
+      "step": 32590
+    },
+    {
+      "epoch": 0.5894571253568273,
+      "grad_norm": 41.375,
+      "learning_rate": 9.907897480289779e-06,
+      "loss": 16.9153,
+      "step": 32600
+    },
+    {
+      "epoch": 0.5896379404259551,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.90786922798312e-06,
+      "loss": 17.1193,
+      "step": 32610
+    },
+    {
+      "epoch": 0.589818755495083,
+      "grad_norm": 43.59375,
+      "learning_rate": 9.90784097567646e-06,
+      "loss": 17.2083,
+      "step": 32620
+    },
+    {
+      "epoch": 0.5899995705642108,
+      "grad_norm": 44.75,
+      "learning_rate": 9.907812723369801e-06,
+      "loss": 17.1789,
+      "step": 32630
+    },
+    {
+      "epoch": 0.5901803856333386,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.907784471063142e-06,
+      "loss": 16.5824,
+      "step": 32640
+    },
+    {
+      "epoch": 0.5903612007024666,
+      "grad_norm": 39.375,
+      "learning_rate": 9.90775621875648e-06,
+      "loss": 16.6448,
+      "step": 32650
+    },
+    {
+      "epoch": 0.5905420157715944,
+      "grad_norm": 44.125,
+      "learning_rate": 9.907727966449821e-06,
+      "loss": 17.0394,
+      "step": 32660
+    },
+    {
+      "epoch": 0.5907228308407223,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.907699714143162e-06,
+      "loss": 16.9122,
+      "step": 32670
+    },
+    {
+      "epoch": 0.5909036459098501,
+      "grad_norm": 44.6875,
+      "learning_rate": 9.907671461836502e-06,
+      "loss": 16.9744,
+      "step": 32680
+    },
+    {
+      "epoch": 0.591084460978978,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.907643209529843e-06,
+      "loss": 16.7347,
+      "step": 32690
+    },
+    {
+      "epoch": 0.5912652760481059,
+      "grad_norm": 44.84375,
+      "learning_rate": 9.907614957223184e-06,
+      "loss": 17.1523,
+      "step": 32700
+    },
+    {
+      "epoch": 0.5914460911172337,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.907586704916524e-06,
+      "loss": 16.64,
+      "step": 32710
+    },
+    {
+      "epoch": 0.5916269061863616,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.907558452609865e-06,
+      "loss": 16.8095,
+      "step": 32720
+    },
+    {
+      "epoch": 0.5918077212554894,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.907530200303205e-06,
+      "loss": 17.1435,
+      "step": 32730
+    },
+    {
+      "epoch": 0.5919885363246173,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.907501947996544e-06,
+      "loss": 17.3664,
+      "step": 32740
+    },
+    {
+      "epoch": 0.5921693513937452,
+      "grad_norm": 42.25,
+      "learning_rate": 9.907473695689885e-06,
+      "loss": 17.0529,
+      "step": 32750
+    },
+    {
+      "epoch": 0.592350166462873,
+      "grad_norm": 42.625,
+      "learning_rate": 9.907445443383226e-06,
+      "loss": 16.9959,
+      "step": 32760
+    },
+    {
+      "epoch": 0.5925309815320009,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.907417191076566e-06,
+      "loss": 16.3658,
+      "step": 32770
+    },
+    {
+      "epoch": 0.5927117966011287,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.907388938769907e-06,
+      "loss": 16.7799,
+      "step": 32780
+    },
+    {
+      "epoch": 0.5928926116702566,
+      "grad_norm": 40.875,
+      "learning_rate": 9.907360686463246e-06,
+      "loss": 17.0065,
+      "step": 32790
+    },
+    {
+      "epoch": 0.5930734267393845,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.907332434156588e-06,
+      "loss": 16.8945,
+      "step": 32800
+    },
+    {
+      "epoch": 0.5932542418085123,
+      "grad_norm": 41.875,
+      "learning_rate": 9.907304181849929e-06,
+      "loss": 16.9943,
+      "step": 32810
+    },
+    {
+      "epoch": 0.5934350568776402,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.90727592954327e-06,
+      "loss": 16.6761,
+      "step": 32820
+    },
+    {
+      "epoch": 0.593615871946768,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.907247677236608e-06,
+      "loss": 16.9836,
+      "step": 32830
+    },
+    {
+      "epoch": 0.5937966870158959,
+      "grad_norm": 42.0,
+      "learning_rate": 9.907219424929949e-06,
+      "loss": 17.4699,
+      "step": 32840
+    },
+    {
+      "epoch": 0.5939775020850238,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.90719117262329e-06,
+      "loss": 17.0958,
+      "step": 32850
+    },
+    {
+      "epoch": 0.5941583171541516,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.90716292031663e-06,
+      "loss": 17.0023,
+      "step": 32860
+    },
+    {
+      "epoch": 0.5943391322232795,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.90713466800997e-06,
+      "loss": 16.894,
+      "step": 32870
+    },
+    {
+      "epoch": 0.5945199472924073,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.907106415703311e-06,
+      "loss": 17.3038,
+      "step": 32880
+    },
+    {
+      "epoch": 0.5947007623615352,
+      "grad_norm": 41.5,
+      "learning_rate": 9.907078163396652e-06,
+      "loss": 16.8994,
+      "step": 32890
+    },
+    {
+      "epoch": 0.594881577430663,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.907049911089993e-06,
+      "loss": 17.6324,
+      "step": 32900
+    },
+    {
+      "epoch": 0.595062392499791,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.907021658783332e-06,
+      "loss": 16.7333,
+      "step": 32910
+    },
+    {
+      "epoch": 0.5952432075689188,
+      "grad_norm": 43.0,
+      "learning_rate": 9.906993406476672e-06,
+      "loss": 16.869,
+      "step": 32920
+    },
+    {
+      "epoch": 0.5954240226380466,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.906965154170013e-06,
+      "loss": 16.7851,
+      "step": 32930
+    },
+    {
+      "epoch": 0.5956048377071745,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.906936901863353e-06,
+      "loss": 16.8765,
+      "step": 32940
+    },
+    {
+      "epoch": 0.5957856527763024,
+      "grad_norm": 38.3125,
+      "learning_rate": 9.906908649556694e-06,
+      "loss": 17.119,
+      "step": 32950
+    },
+    {
+      "epoch": 0.5959664678454303,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.906880397250033e-06,
+      "loss": 16.8493,
+      "step": 32960
+    },
+    {
+      "epoch": 0.5961472829145581,
+      "grad_norm": 42.625,
+      "learning_rate": 9.906852144943375e-06,
+      "loss": 17.293,
+      "step": 32970
+    },
+    {
+      "epoch": 0.5963280979836859,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.906823892636716e-06,
+      "loss": 16.4654,
+      "step": 32980
+    },
+    {
+      "epoch": 0.5965089130528138,
+      "grad_norm": 46.65625,
+      "learning_rate": 9.906795640330057e-06,
+      "loss": 16.9846,
+      "step": 32990
+    },
+    {
+      "epoch": 0.5966897281219417,
+      "grad_norm": 38.6875,
+      "learning_rate": 9.906767388023396e-06,
+      "loss": 17.0765,
+      "step": 33000
+    },
+    {
+      "epoch": 0.5968705431910696,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.906739135716736e-06,
+      "loss": 17.217,
+      "step": 33010
+    },
+    {
+      "epoch": 0.5970513582601974,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.906710883410077e-06,
+      "loss": 16.6237,
+      "step": 33020
+    },
+    {
+      "epoch": 0.5972321733293252,
+      "grad_norm": 44.125,
+      "learning_rate": 9.906682631103417e-06,
+      "loss": 16.7351,
+      "step": 33030
+    },
+    {
+      "epoch": 0.5974129883984531,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.906654378796758e-06,
+      "loss": 16.7977,
+      "step": 33040
+    },
+    {
+      "epoch": 0.597593803467581,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.906626126490097e-06,
+      "loss": 17.443,
+      "step": 33050
+    },
+    {
+      "epoch": 0.5977746185367089,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.90659787418344e-06,
+      "loss": 16.9387,
+      "step": 33060
+    },
+    {
+      "epoch": 0.5979554336058367,
+      "grad_norm": 38.90625,
+      "learning_rate": 9.90656962187678e-06,
+      "loss": 16.9259,
+      "step": 33070
+    },
+    {
+      "epoch": 0.5981362486749646,
+      "grad_norm": 39.4375,
+      "learning_rate": 9.906541369570119e-06,
+      "loss": 16.7404,
+      "step": 33080
+    },
+    {
+      "epoch": 0.5983170637440924,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.90651311726346e-06,
+      "loss": 16.6768,
+      "step": 33090
+    },
+    {
+      "epoch": 0.5984978788132203,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.9064848649568e-06,
+      "loss": 16.6058,
+      "step": 33100
+    },
+    {
+      "epoch": 0.5986786938823482,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.90645661265014e-06,
+      "loss": 16.9653,
+      "step": 33110
+    },
+    {
+      "epoch": 0.598859508951476,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.906428360343481e-06,
+      "loss": 16.6989,
+      "step": 33120
+    },
+    {
+      "epoch": 0.5990403240206039,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.906400108036822e-06,
+      "loss": 16.8293,
+      "step": 33130
+    },
+    {
+      "epoch": 0.5992211390897317,
+      "grad_norm": 42.25,
+      "learning_rate": 9.90637185573016e-06,
+      "loss": 16.9864,
+      "step": 33140
+    },
+    {
+      "epoch": 0.5994019541588596,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.906343603423503e-06,
+      "loss": 17.4234,
+      "step": 33150
+    },
+    {
+      "epoch": 0.5995827692279875,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.906315351116844e-06,
+      "loss": 17.622,
+      "step": 33160
+    },
+    {
+      "epoch": 0.5997635842971153,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.906287098810183e-06,
+      "loss": 17.2536,
+      "step": 33170
+    },
+    {
+      "epoch": 0.5999443993662432,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.906258846503523e-06,
+      "loss": 17.4132,
+      "step": 33180
+    },
+    {
+      "epoch": 0.600125214435371,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.906230594196864e-06,
+      "loss": 17.1735,
+      "step": 33190
+    },
+    {
+      "epoch": 0.6003060295044989,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.906202341890205e-06,
+      "loss": 16.8919,
+      "step": 33200
+    },
+    {
+      "epoch": 0.6004868445736268,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.906174089583545e-06,
+      "loss": 16.8507,
+      "step": 33210
+    },
+    {
+      "epoch": 0.6006676596427546,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.906145837276884e-06,
+      "loss": 16.8082,
+      "step": 33220
+    },
+    {
+      "epoch": 0.6008484747118825,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.906117584970226e-06,
+      "loss": 17.3458,
+      "step": 33230
+    },
+    {
+      "epoch": 0.6010292897810103,
+      "grad_norm": 41.0,
+      "learning_rate": 9.906089332663567e-06,
+      "loss": 17.2312,
+      "step": 33240
+    },
+    {
+      "epoch": 0.6012101048501383,
+      "grad_norm": 39.375,
+      "learning_rate": 9.906061080356908e-06,
+      "loss": 16.5374,
+      "step": 33250
+    },
+    {
+      "epoch": 0.6013909199192661,
+      "grad_norm": 41.375,
+      "learning_rate": 9.906032828050247e-06,
+      "loss": 16.3421,
+      "step": 33260
+    },
+    {
+      "epoch": 0.6015717349883939,
+      "grad_norm": 45.34375,
+      "learning_rate": 9.906004575743587e-06,
+      "loss": 16.4727,
+      "step": 33270
+    },
+    {
+      "epoch": 0.6017525500575218,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.905976323436928e-06,
+      "loss": 16.9591,
+      "step": 33280
+    },
+    {
+      "epoch": 0.6019333651266496,
+      "grad_norm": 41.875,
+      "learning_rate": 9.905948071130268e-06,
+      "loss": 17.0089,
+      "step": 33290
+    },
+    {
+      "epoch": 0.6021141801957776,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.905919818823609e-06,
+      "loss": 17.0707,
+      "step": 33300
+    },
+    {
+      "epoch": 0.6022949952649054,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.905891566516948e-06,
+      "loss": 17.0987,
+      "step": 33310
+    },
+    {
+      "epoch": 0.6024758103340332,
+      "grad_norm": 43.5,
+      "learning_rate": 9.90586331421029e-06,
+      "loss": 17.0295,
+      "step": 33320
+    },
+    {
+      "epoch": 0.6026566254031611,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.905835061903631e-06,
+      "loss": 16.9252,
+      "step": 33330
+    },
+    {
+      "epoch": 0.6028374404722889,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.90580680959697e-06,
+      "loss": 16.6911,
+      "step": 33340
+    },
+    {
+      "epoch": 0.6030182555414169,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.90577855729031e-06,
+      "loss": 17.1652,
+      "step": 33350
+    },
+    {
+      "epoch": 0.6031990706105447,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.905750304983651e-06,
+      "loss": 16.8341,
+      "step": 33360
+    },
+    {
+      "epoch": 0.6033798856796725,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.905722052676992e-06,
+      "loss": 16.6755,
+      "step": 33370
+    },
+    {
+      "epoch": 0.6035607007488004,
+      "grad_norm": 44.25,
+      "learning_rate": 9.905693800370332e-06,
+      "loss": 16.725,
+      "step": 33380
+    },
+    {
+      "epoch": 0.6037415158179282,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.905665548063671e-06,
+      "loss": 16.9334,
+      "step": 33390
+    },
+    {
+      "epoch": 0.6039223308870562,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.905637295757012e-06,
+      "loss": 16.5482,
+      "step": 33400
+    },
+    {
+      "epoch": 0.604103145956184,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.905609043450354e-06,
+      "loss": 16.8465,
+      "step": 33410
+    },
+    {
+      "epoch": 0.6042839610253119,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.905580791143695e-06,
+      "loss": 16.6368,
+      "step": 33420
+    },
+    {
+      "epoch": 0.6044647760944397,
+      "grad_norm": 41.75,
+      "learning_rate": 9.905552538837034e-06,
+      "loss": 16.757,
+      "step": 33430
+    },
+    {
+      "epoch": 0.6046455911635675,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.905524286530374e-06,
+      "loss": 17.0386,
+      "step": 33440
+    },
+    {
+      "epoch": 0.6048264062326955,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.905496034223715e-06,
+      "loss": 16.8365,
+      "step": 33450
+    },
+    {
+      "epoch": 0.6050072213018233,
+      "grad_norm": 40.75,
+      "learning_rate": 9.905467781917056e-06,
+      "loss": 17.0499,
+      "step": 33460
+    },
+    {
+      "epoch": 0.6051880363709512,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.905439529610396e-06,
+      "loss": 17.1388,
+      "step": 33470
+    },
+    {
+      "epoch": 0.605368851440079,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.905411277303735e-06,
+      "loss": 17.1348,
+      "step": 33480
+    },
+    {
+      "epoch": 0.6055496665092068,
+      "grad_norm": 39.34375,
+      "learning_rate": 9.905383024997076e-06,
+      "loss": 16.5983,
+      "step": 33490
+    },
+    {
+      "epoch": 0.6057304815783348,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.905354772690418e-06,
+      "loss": 17.5614,
+      "step": 33500
+    },
+    {
+      "epoch": 0.6059112966474626,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.905326520383757e-06,
+      "loss": 16.7848,
+      "step": 33510
+    },
+    {
+      "epoch": 0.6060921117165905,
+      "grad_norm": 40.0,
+      "learning_rate": 9.905298268077098e-06,
+      "loss": 16.9091,
+      "step": 33520
+    },
+    {
+      "epoch": 0.6062729267857183,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.905270015770438e-06,
+      "loss": 16.7644,
+      "step": 33530
+    },
+    {
+      "epoch": 0.6064537418548461,
+      "grad_norm": 43.25,
+      "learning_rate": 9.905241763463779e-06,
+      "loss": 16.7903,
+      "step": 33540
+    },
+    {
+      "epoch": 0.606634556923974,
+      "grad_norm": 44.875,
+      "learning_rate": 9.90521351115712e-06,
+      "loss": 16.7235,
+      "step": 33550
+    },
+    {
+      "epoch": 0.6068153719931019,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.905185258850458e-06,
+      "loss": 16.9318,
+      "step": 33560
+    },
+    {
+      "epoch": 0.6069961870622298,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.905157006543799e-06,
+      "loss": 16.8678,
+      "step": 33570
+    },
+    {
+      "epoch": 0.6071770021313576,
+      "grad_norm": 40.375,
+      "learning_rate": 9.905128754237141e-06,
+      "loss": 16.9585,
+      "step": 33580
+    },
+    {
+      "epoch": 0.6073578172004855,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.905100501930482e-06,
+      "loss": 16.8712,
+      "step": 33590
+    },
+    {
+      "epoch": 0.6075386322696134,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.905072249623821e-06,
+      "loss": 16.741,
+      "step": 33600
+    },
+    {
+      "epoch": 0.6077194473387412,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.905043997317162e-06,
+      "loss": 16.962,
+      "step": 33610
+    },
+    {
+      "epoch": 0.6079002624078691,
+      "grad_norm": 41.25,
+      "learning_rate": 9.905015745010502e-06,
+      "loss": 16.7619,
+      "step": 33620
+    },
+    {
+      "epoch": 0.6080810774769969,
+      "grad_norm": 42.375,
+      "learning_rate": 9.904987492703843e-06,
+      "loss": 16.7792,
+      "step": 33630
+    },
+    {
+      "epoch": 0.6082618925461248,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.904959240397183e-06,
+      "loss": 17.2013,
+      "step": 33640
+    },
+    {
+      "epoch": 0.6084427076152527,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.904930988090522e-06,
+      "loss": 17.1005,
+      "step": 33650
+    },
+    {
+      "epoch": 0.6086235226843805,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.904902735783863e-06,
+      "loss": 17.0336,
+      "step": 33660
+    },
+    {
+      "epoch": 0.6088043377535084,
+      "grad_norm": 41.75,
+      "learning_rate": 9.904874483477205e-06,
+      "loss": 16.9484,
+      "step": 33670
+    },
+    {
+      "epoch": 0.6089851528226362,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.904846231170544e-06,
+      "loss": 16.7472,
+      "step": 33680
+    },
+    {
+      "epoch": 0.6091659678917641,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.904817978863885e-06,
+      "loss": 16.9834,
+      "step": 33690
+    },
+    {
+      "epoch": 0.609346782960892,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.904789726557226e-06,
+      "loss": 16.5895,
+      "step": 33700
+    },
+    {
+      "epoch": 0.6095275980300198,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.904761474250566e-06,
+      "loss": 16.9442,
+      "step": 33710
+    },
+    {
+      "epoch": 0.6097084130991477,
+      "grad_norm": 43.75,
+      "learning_rate": 9.904733221943907e-06,
+      "loss": 17.1539,
+      "step": 33720
+    },
+    {
+      "epoch": 0.6098892281682755,
+      "grad_norm": 39.625,
+      "learning_rate": 9.904704969637247e-06,
+      "loss": 17.1007,
+      "step": 33730
+    },
+    {
+      "epoch": 0.6100700432374034,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.904676717330586e-06,
+      "loss": 16.9644,
+      "step": 33740
+    },
+    {
+      "epoch": 0.6102508583065313,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.904648465023927e-06,
+      "loss": 16.9215,
+      "step": 33750
+    },
+    {
+      "epoch": 0.6104316733756592,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.90462021271727e-06,
+      "loss": 17.2342,
+      "step": 33760
+    },
+    {
+      "epoch": 0.610612488444787,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.904591960410608e-06,
+      "loss": 17.427,
+      "step": 33770
+    },
+    {
+      "epoch": 0.6107933035139148,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.904563708103949e-06,
+      "loss": 17.1716,
+      "step": 33780
+    },
+    {
+      "epoch": 0.6109741185830427,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.90453545579729e-06,
+      "loss": 17.059,
+      "step": 33790
+    },
+    {
+      "epoch": 0.6111549336521706,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.90450720349063e-06,
+      "loss": 17.2268,
+      "step": 33800
+    },
+    {
+      "epoch": 0.6113357487212985,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.90447895118397e-06,
+      "loss": 16.9237,
+      "step": 33810
+    },
+    {
+      "epoch": 0.6115165637904263,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.90445069887731e-06,
+      "loss": 17.0366,
+      "step": 33820
+    },
+    {
+      "epoch": 0.6116973788595541,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.90442244657065e-06,
+      "loss": 16.9747,
+      "step": 33830
+    },
+    {
+      "epoch": 0.611878193928682,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.90439419426399e-06,
+      "loss": 16.7878,
+      "step": 33840
+    },
+    {
+      "epoch": 0.6120590089978099,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.904365941957333e-06,
+      "loss": 16.8532,
+      "step": 33850
+    },
+    {
+      "epoch": 0.6122398240669378,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.904337689650672e-06,
+      "loss": 16.7214,
+      "step": 33860
+    },
+    {
+      "epoch": 0.6124206391360656,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.904309437344013e-06,
+      "loss": 16.9797,
+      "step": 33870
+    },
+    {
+      "epoch": 0.6126014542051934,
+      "grad_norm": 41.75,
+      "learning_rate": 9.904281185037353e-06,
+      "loss": 16.8734,
+      "step": 33880
+    },
+    {
+      "epoch": 0.6127822692743213,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.904252932730694e-06,
+      "loss": 17.0882,
+      "step": 33890
+    },
+    {
+      "epoch": 0.6129630843434491,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.904224680424035e-06,
+      "loss": 16.4828,
+      "step": 33900
+    },
+    {
+      "epoch": 0.6131438994125771,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.904196428117374e-06,
+      "loss": 16.8991,
+      "step": 33910
+    },
+    {
+      "epoch": 0.6133247144817049,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.904168175810714e-06,
+      "loss": 16.8868,
+      "step": 33920
+    },
+    {
+      "epoch": 0.6135055295508328,
+      "grad_norm": 40.625,
+      "learning_rate": 9.904139923504056e-06,
+      "loss": 16.8543,
+      "step": 33930
+    },
+    {
+      "epoch": 0.6136863446199606,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.904111671197395e-06,
+      "loss": 17.1237,
+      "step": 33940
+    },
+    {
+      "epoch": 0.6138671596890884,
+      "grad_norm": 41.25,
+      "learning_rate": 9.904083418890736e-06,
+      "loss": 17.0901,
+      "step": 33950
+    },
+    {
+      "epoch": 0.6140479747582164,
+      "grad_norm": 43.125,
+      "learning_rate": 9.904055166584077e-06,
+      "loss": 17.1514,
+      "step": 33960
+    },
+    {
+      "epoch": 0.6142287898273442,
+      "grad_norm": 43.875,
+      "learning_rate": 9.904026914277417e-06,
+      "loss": 17.3739,
+      "step": 33970
+    },
+    {
+      "epoch": 0.6144096048964721,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.903998661970758e-06,
+      "loss": 17.0267,
+      "step": 33980
+    },
+    {
+      "epoch": 0.6145904199655999,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.903970409664097e-06,
+      "loss": 16.9841,
+      "step": 33990
+    },
+    {
+      "epoch": 0.6147712350347277,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.903942157357437e-06,
+      "loss": 16.8561,
+      "step": 34000
+    },
+    {
+      "epoch": 0.6149520501038557,
+      "grad_norm": 39.78125,
+      "learning_rate": 9.903913905050778e-06,
+      "loss": 17.2727,
+      "step": 34010
+    },
+    {
+      "epoch": 0.6151328651729835,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.90388565274412e-06,
+      "loss": 16.6296,
+      "step": 34020
+    },
+    {
+      "epoch": 0.6153136802421114,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.90385740043746e-06,
+      "loss": 16.8987,
+      "step": 34030
+    },
+    {
+      "epoch": 0.6154944953112392,
+      "grad_norm": 40.25,
+      "learning_rate": 9.9038291481308e-06,
+      "loss": 17.0696,
+      "step": 34040
+    },
+    {
+      "epoch": 0.615675310380367,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.90380089582414e-06,
+      "loss": 17.1479,
+      "step": 34050
+    },
+    {
+      "epoch": 0.615856125449495,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.903772643517481e-06,
+      "loss": 17.0279,
+      "step": 34060
+    },
+    {
+      "epoch": 0.6160369405186228,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.903744391210822e-06,
+      "loss": 17.0861,
+      "step": 34070
+    },
+    {
+      "epoch": 0.6162177555877507,
+      "grad_norm": 41.25,
+      "learning_rate": 9.90371613890416e-06,
+      "loss": 17.1457,
+      "step": 34080
+    },
+    {
+      "epoch": 0.6163985706568785,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.903687886597501e-06,
+      "loss": 16.8054,
+      "step": 34090
+    },
+    {
+      "epoch": 0.6165793857260065,
+      "grad_norm": 39.71875,
+      "learning_rate": 9.903659634290842e-06,
+      "loss": 16.8803,
+      "step": 34100
+    },
+    {
+      "epoch": 0.6167602007951343,
+      "grad_norm": 40.0,
+      "learning_rate": 9.903631381984183e-06,
+      "loss": 17.2851,
+      "step": 34110
+    },
+    {
+      "epoch": 0.6169410158642621,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.903603129677523e-06,
+      "loss": 17.0829,
+      "step": 34120
+    },
+    {
+      "epoch": 0.61712183093339,
+      "grad_norm": 43.625,
+      "learning_rate": 9.903574877370864e-06,
+      "loss": 16.758,
+      "step": 34130
+    },
+    {
+      "epoch": 0.6173026460025178,
+      "grad_norm": 39.09375,
+      "learning_rate": 9.903546625064204e-06,
+      "loss": 17.0564,
+      "step": 34140
+    },
+    {
+      "epoch": 0.6174834610716458,
+      "grad_norm": 41.625,
+      "learning_rate": 9.903518372757545e-06,
+      "loss": 16.8258,
+      "step": 34150
+    },
+    {
+      "epoch": 0.6176642761407736,
+      "grad_norm": 44.84375,
+      "learning_rate": 9.903490120450886e-06,
+      "loss": 17.0703,
+      "step": 34160
+    },
+    {
+      "epoch": 0.6178450912099014,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.903461868144225e-06,
+      "loss": 16.8211,
+      "step": 34170
+    },
+    {
+      "epoch": 0.6180259062790293,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.903433615837565e-06,
+      "loss": 16.6875,
+      "step": 34180
+    },
+    {
+      "epoch": 0.6182067213481571,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.903405363530906e-06,
+      "loss": 16.5914,
+      "step": 34190
+    },
+    {
+      "epoch": 0.618387536417285,
+      "grad_norm": 39.875,
+      "learning_rate": 9.903377111224246e-06,
+      "loss": 16.8584,
+      "step": 34200
+    },
+    {
+      "epoch": 0.6185683514864129,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.903348858917587e-06,
+      "loss": 16.7128,
+      "step": 34210
+    },
+    {
+      "epoch": 0.6187491665555407,
+      "grad_norm": 41.875,
+      "learning_rate": 9.903320606610928e-06,
+      "loss": 16.6644,
+      "step": 34220
+    },
+    {
+      "epoch": 0.6189299816246686,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.903292354304268e-06,
+      "loss": 16.4386,
+      "step": 34230
+    },
+    {
+      "epoch": 0.6191107966937964,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.903264101997609e-06,
+      "loss": 16.9011,
+      "step": 34240
+    },
+    {
+      "epoch": 0.6192916117629244,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.903235849690948e-06,
+      "loss": 16.8479,
+      "step": 34250
+    },
+    {
+      "epoch": 0.6194724268320522,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.903207597384289e-06,
+      "loss": 17.3777,
+      "step": 34260
+    },
+    {
+      "epoch": 0.6196532419011801,
+      "grad_norm": 45.84375,
+      "learning_rate": 9.903179345077629e-06,
+      "loss": 17.1973,
+      "step": 34270
+    },
+    {
+      "epoch": 0.6198340569703079,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.903151092770971e-06,
+      "loss": 17.3628,
+      "step": 34280
+    },
+    {
+      "epoch": 0.6200148720394357,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.90312284046431e-06,
+      "loss": 17.0504,
+      "step": 34290
+    },
+    {
+      "epoch": 0.6201956871085637,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.903094588157651e-06,
+      "loss": 17.6827,
+      "step": 34300
+    },
+    {
+      "epoch": 0.6203765021776915,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.903066335850992e-06,
+      "loss": 16.3788,
+      "step": 34310
+    },
+    {
+      "epoch": 0.6205573172468194,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.903038083544332e-06,
+      "loss": 16.5896,
+      "step": 34320
+    },
+    {
+      "epoch": 0.6207381323159472,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.903009831237673e-06,
+      "loss": 16.5937,
+      "step": 34330
+    },
+    {
+      "epoch": 0.620918947385075,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.902981578931012e-06,
+      "loss": 16.8482,
+      "step": 34340
+    },
+    {
+      "epoch": 0.621099762454203,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.902953326624352e-06,
+      "loss": 16.9854,
+      "step": 34350
+    },
+    {
+      "epoch": 0.6212805775233308,
+      "grad_norm": 44.625,
+      "learning_rate": 9.902925074317693e-06,
+      "loss": 16.7246,
+      "step": 34360
+    },
+    {
+      "epoch": 0.6214613925924587,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.902896822011034e-06,
+      "loss": 16.9704,
+      "step": 34370
+    },
+    {
+      "epoch": 0.6216422076615865,
+      "grad_norm": 38.78125,
+      "learning_rate": 9.902868569704374e-06,
+      "loss": 16.8478,
+      "step": 34380
+    },
+    {
+      "epoch": 0.6218230227307143,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.902840317397715e-06,
+      "loss": 16.8959,
+      "step": 34390
+    },
+    {
+      "epoch": 0.6220038377998423,
+      "grad_norm": 42.25,
+      "learning_rate": 9.902812065091056e-06,
+      "loss": 16.4676,
+      "step": 34400
+    },
+    {
+      "epoch": 0.6221846528689701,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.902783812784396e-06,
+      "loss": 16.6884,
+      "step": 34410
+    },
+    {
+      "epoch": 0.622365467938098,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.902755560477735e-06,
+      "loss": 17.0956,
+      "step": 34420
+    },
+    {
+      "epoch": 0.6225462830072258,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.902727308171076e-06,
+      "loss": 16.7984,
+      "step": 34430
+    },
+    {
+      "epoch": 0.6227270980763537,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.902699055864416e-06,
+      "loss": 16.6023,
+      "step": 34440
+    },
+    {
+      "epoch": 0.6229079131454816,
+      "grad_norm": 44.53125,
+      "learning_rate": 9.902670803557757e-06,
+      "loss": 16.9605,
+      "step": 34450
+    },
+    {
+      "epoch": 0.6230887282146094,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.902642551251098e-06,
+      "loss": 17.1605,
+      "step": 34460
+    },
+    {
+      "epoch": 0.6232695432837373,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.902614298944438e-06,
+      "loss": 16.8769,
+      "step": 34470
+    },
+    {
+      "epoch": 0.6234503583528651,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.902586046637779e-06,
+      "loss": 17.3661,
+      "step": 34480
+    },
+    {
+      "epoch": 0.623631173421993,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.90255779433112e-06,
+      "loss": 16.8391,
+      "step": 34490
+    },
+    {
+      "epoch": 0.6238119884911209,
+      "grad_norm": 41.625,
+      "learning_rate": 9.90252954202446e-06,
+      "loss": 17.2878,
+      "step": 34500
+    },
+    {
+      "epoch": 0.6239928035602487,
+      "grad_norm": 39.0625,
+      "learning_rate": 9.902501289717799e-06,
+      "loss": 17.1672,
+      "step": 34510
+    },
+    {
+      "epoch": 0.6241736186293766,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.90247303741114e-06,
+      "loss": 17.1609,
+      "step": 34520
+    },
+    {
+      "epoch": 0.6243544336985044,
+      "grad_norm": 42.125,
+      "learning_rate": 9.90244478510448e-06,
+      "loss": 17.036,
+      "step": 34530
+    },
+    {
+      "epoch": 0.6245352487676323,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.902416532797821e-06,
+      "loss": 16.9945,
+      "step": 34540
+    },
+    {
+      "epoch": 0.6247160638367601,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.902388280491161e-06,
+      "loss": 16.446,
+      "step": 34550
+    },
+    {
+      "epoch": 0.624896878905888,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.902360028184502e-06,
+      "loss": 16.5309,
+      "step": 34560
+    },
+    {
+      "epoch": 0.6250776939750159,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.902331775877843e-06,
+      "loss": 17.0712,
+      "step": 34570
+    },
+    {
+      "epoch": 0.6252585090441437,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.902303523571183e-06,
+      "loss": 16.878,
+      "step": 34580
+    },
+    {
+      "epoch": 0.6254393241132716,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.902275271264524e-06,
+      "loss": 16.8346,
+      "step": 34590
+    },
+    {
+      "epoch": 0.6256201391823994,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.902247018957863e-06,
+      "loss": 16.9437,
+      "step": 34600
+    },
+    {
+      "epoch": 0.6258009542515274,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.902218766651204e-06,
+      "loss": 17.3732,
+      "step": 34610
+    },
+    {
+      "epoch": 0.6259817693206552,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.902190514344544e-06,
+      "loss": 17.4373,
+      "step": 34620
+    },
+    {
+      "epoch": 0.626162584389783,
+      "grad_norm": 40.25,
+      "learning_rate": 9.902162262037885e-06,
+      "loss": 16.7355,
+      "step": 34630
+    },
+    {
+      "epoch": 0.6263433994589109,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.902134009731225e-06,
+      "loss": 16.743,
+      "step": 34640
+    },
+    {
+      "epoch": 0.6265242145280387,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.902105757424566e-06,
+      "loss": 16.8201,
+      "step": 34650
+    },
+    {
+      "epoch": 0.6267050295971667,
+      "grad_norm": 43.625,
+      "learning_rate": 9.902077505117907e-06,
+      "loss": 17.0255,
+      "step": 34660
+    },
+    {
+      "epoch": 0.6268858446662945,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.902049252811247e-06,
+      "loss": 16.6278,
+      "step": 34670
+    },
+    {
+      "epoch": 0.6270666597354223,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.902021000504586e-06,
+      "loss": 16.7172,
+      "step": 34680
+    },
+    {
+      "epoch": 0.6272474748045502,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.901992748197927e-06,
+      "loss": 17.1172,
+      "step": 34690
+    },
+    {
+      "epoch": 0.627428289873678,
+      "grad_norm": 38.4375,
+      "learning_rate": 9.901964495891267e-06,
+      "loss": 17.0091,
+      "step": 34700
+    },
+    {
+      "epoch": 0.627609104942806,
+      "grad_norm": 43.625,
+      "learning_rate": 9.901936243584608e-06,
+      "loss": 16.8104,
+      "step": 34710
+    },
+    {
+      "epoch": 0.6277899200119338,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.901907991277949e-06,
+      "loss": 16.8872,
+      "step": 34720
+    },
+    {
+      "epoch": 0.6279707350810616,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.90187973897129e-06,
+      "loss": 17.1016,
+      "step": 34730
+    },
+    {
+      "epoch": 0.6281515501501895,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.90185148666463e-06,
+      "loss": 17.1182,
+      "step": 34740
+    },
+    {
+      "epoch": 0.6283323652193173,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.90182323435797e-06,
+      "loss": 16.9325,
+      "step": 34750
+    },
+    {
+      "epoch": 0.6285131802884453,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.901794982051311e-06,
+      "loss": 16.7678,
+      "step": 34760
+    },
+    {
+      "epoch": 0.6286939953575731,
+      "grad_norm": 44.0,
+      "learning_rate": 9.90176672974465e-06,
+      "loss": 16.7194,
+      "step": 34770
+    },
+    {
+      "epoch": 0.628874810426701,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.90173847743799e-06,
+      "loss": 16.9852,
+      "step": 34780
+    },
+    {
+      "epoch": 0.6290556254958288,
+      "grad_norm": 40.34375,
+      "learning_rate": 9.901710225131331e-06,
+      "loss": 17.3694,
+      "step": 34790
+    },
+    {
+      "epoch": 0.6292364405649566,
+      "grad_norm": 45.625,
+      "learning_rate": 9.901681972824672e-06,
+      "loss": 16.7443,
+      "step": 34800
+    },
+    {
+      "epoch": 0.6294172556340846,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.901653720518013e-06,
+      "loss": 16.5887,
+      "step": 34810
+    },
+    {
+      "epoch": 0.6295980707032124,
+      "grad_norm": 43.125,
+      "learning_rate": 9.901625468211353e-06,
+      "loss": 16.9845,
+      "step": 34820
+    },
+    {
+      "epoch": 0.6297788857723403,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.901597215904694e-06,
+      "loss": 16.8,
+      "step": 34830
+    },
+    {
+      "epoch": 0.6299597008414681,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.901568963598034e-06,
+      "loss": 17.1302,
+      "step": 34840
+    },
+    {
+      "epoch": 0.6301405159105959,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.901540711291373e-06,
+      "loss": 16.5467,
+      "step": 34850
+    },
+    {
+      "epoch": 0.6303213309797239,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.901512458984714e-06,
+      "loss": 16.4088,
+      "step": 34860
+    },
+    {
+      "epoch": 0.6305021460488517,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.901484206678055e-06,
+      "loss": 16.5629,
+      "step": 34870
+    },
+    {
+      "epoch": 0.6306829611179796,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.901455954371395e-06,
+      "loss": 16.8379,
+      "step": 34880
+    },
+    {
+      "epoch": 0.6308637761871074,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.901427702064736e-06,
+      "loss": 16.6736,
+      "step": 34890
+    },
+    {
+      "epoch": 0.6310445912562352,
+      "grad_norm": 42.5,
+      "learning_rate": 9.901399449758076e-06,
+      "loss": 16.9239,
+      "step": 34900
+    },
+    {
+      "epoch": 0.6312254063253632,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.901371197451417e-06,
+      "loss": 17.1634,
+      "step": 34910
+    },
+    {
+      "epoch": 0.631406221394491,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.901342945144758e-06,
+      "loss": 17.1069,
+      "step": 34920
+    },
+    {
+      "epoch": 0.6315870364636189,
+      "grad_norm": 42.875,
+      "learning_rate": 9.901314692838098e-06,
+      "loss": 16.9312,
+      "step": 34930
+    },
+    {
+      "epoch": 0.6317678515327467,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.901286440531437e-06,
+      "loss": 16.988,
+      "step": 34940
+    },
+    {
+      "epoch": 0.6319486666018747,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.901258188224778e-06,
+      "loss": 16.9073,
+      "step": 34950
+    },
+    {
+      "epoch": 0.6321294816710025,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.901229935918119e-06,
+      "loss": 17.0261,
+      "step": 34960
+    },
+    {
+      "epoch": 0.6323102967401303,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.901201683611459e-06,
+      "loss": 16.8868,
+      "step": 34970
+    },
+    {
+      "epoch": 0.6324911118092582,
+      "grad_norm": 42.375,
+      "learning_rate": 9.9011734313048e-06,
+      "loss": 16.634,
+      "step": 34980
+    },
+    {
+      "epoch": 0.632671926878386,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.90114517899814e-06,
+      "loss": 16.5571,
+      "step": 34990
+    },
+    {
+      "epoch": 0.632852741947514,
+      "grad_norm": 41.625,
+      "learning_rate": 9.901116926691481e-06,
+      "loss": 16.7899,
+      "step": 35000
+    },
+    {
+      "epoch": 0.632852741947514,
+      "eval_loss": 2.1153111457824707,
+      "eval_runtime": 229.0986,
+      "eval_samples_per_second": 3169.199,
+      "eval_steps_per_second": 49.52,
+      "step": 35000
+    },
+    {
+      "epoch": 0.6330335570166418,
+      "grad_norm": 40.875,
+      "learning_rate": 9.901088674384822e-06,
+      "loss": 17.0118,
+      "step": 35010
+    },
+    {
+      "epoch": 0.6332143720857696,
+      "grad_norm": 42.5,
+      "learning_rate": 9.901060422078162e-06,
+      "loss": 17.3374,
+      "step": 35020
+    },
+    {
+      "epoch": 0.6333951871548975,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.901032169771501e-06,
+      "loss": 17.1585,
+      "step": 35030
+    },
+    {
+      "epoch": 0.6335760022240253,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.901003917464842e-06,
+      "loss": 16.9927,
+      "step": 35040
+    },
+    {
+      "epoch": 0.6337568172931533,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.900975665158182e-06,
+      "loss": 16.7819,
+      "step": 35050
+    },
+    {
+      "epoch": 0.6339376323622811,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.900947412851523e-06,
+      "loss": 16.6033,
+      "step": 35060
+    },
+    {
+      "epoch": 0.6341184474314089,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.900919160544864e-06,
+      "loss": 16.9377,
+      "step": 35070
+    },
+    {
+      "epoch": 0.6342992625005368,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.900890908238204e-06,
+      "loss": 16.9437,
+      "step": 35080
+    },
+    {
+      "epoch": 0.6344800775696646,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.900862655931545e-06,
+      "loss": 16.9132,
+      "step": 35090
+    },
+    {
+      "epoch": 0.6346608926387926,
+      "grad_norm": 39.625,
+      "learning_rate": 9.900834403624886e-06,
+      "loss": 16.9262,
+      "step": 35100
+    },
+    {
+      "epoch": 0.6348417077079204,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.900806151318224e-06,
+      "loss": 16.8212,
+      "step": 35110
+    },
+    {
+      "epoch": 0.6350225227770483,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.900777899011565e-06,
+      "loss": 16.7947,
+      "step": 35120
+    },
+    {
+      "epoch": 0.6352033378461761,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.900749646704906e-06,
+      "loss": 16.9076,
+      "step": 35130
+    },
+    {
+      "epoch": 0.6353841529153039,
+      "grad_norm": 42.25,
+      "learning_rate": 9.900721394398246e-06,
+      "loss": 16.4483,
+      "step": 35140
+    },
+    {
+      "epoch": 0.6355649679844319,
+      "grad_norm": 43.625,
+      "learning_rate": 9.900693142091587e-06,
+      "loss": 17.4467,
+      "step": 35150
+    },
+    {
+      "epoch": 0.6357457830535597,
+      "grad_norm": 39.375,
+      "learning_rate": 9.900664889784928e-06,
+      "loss": 16.6712,
+      "step": 35160
+    },
+    {
+      "epoch": 0.6359265981226876,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.900636637478268e-06,
+      "loss": 16.6812,
+      "step": 35170
+    },
+    {
+      "epoch": 0.6361074131918154,
+      "grad_norm": 43.25,
+      "learning_rate": 9.900608385171609e-06,
+      "loss": 16.7277,
+      "step": 35180
+    },
+    {
+      "epoch": 0.6362882282609432,
+      "grad_norm": 42.75,
+      "learning_rate": 9.90058013286495e-06,
+      "loss": 17.0572,
+      "step": 35190
+    },
+    {
+      "epoch": 0.6364690433300711,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.900551880558288e-06,
+      "loss": 17.105,
+      "step": 35200
+    },
+    {
+      "epoch": 0.636649858399199,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.900523628251629e-06,
+      "loss": 16.6574,
+      "step": 35210
+    },
+    {
+      "epoch": 0.6368306734683269,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.90049537594497e-06,
+      "loss": 16.8217,
+      "step": 35220
+    },
+    {
+      "epoch": 0.6370114885374547,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.90046712363831e-06,
+      "loss": 16.7569,
+      "step": 35230
+    },
+    {
+      "epoch": 0.6371923036065825,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.900438871331651e-06,
+      "loss": 16.4877,
+      "step": 35240
+    },
+    {
+      "epoch": 0.6373731186757104,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.900410619024991e-06,
+      "loss": 16.7122,
+      "step": 35250
+    },
+    {
+      "epoch": 0.6375539337448383,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.900382366718332e-06,
+      "loss": 16.9515,
+      "step": 35260
+    },
+    {
+      "epoch": 0.6377347488139662,
+      "grad_norm": 42.25,
+      "learning_rate": 9.900354114411673e-06,
+      "loss": 16.655,
+      "step": 35270
+    },
+    {
+      "epoch": 0.637915563883094,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.900325862105012e-06,
+      "loss": 16.847,
+      "step": 35280
+    },
+    {
+      "epoch": 0.6380963789522219,
+      "grad_norm": 38.375,
+      "learning_rate": 9.900297609798352e-06,
+      "loss": 16.9473,
+      "step": 35290
+    },
+    {
+      "epoch": 0.6382771940213497,
+      "grad_norm": 42.625,
+      "learning_rate": 9.900269357491693e-06,
+      "loss": 16.8541,
+      "step": 35300
+    },
+    {
+      "epoch": 0.6384580090904776,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.900241105185034e-06,
+      "loss": 16.9603,
+      "step": 35310
+    },
+    {
+      "epoch": 0.6386388241596055,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.900212852878374e-06,
+      "loss": 16.9738,
+      "step": 35320
+    },
+    {
+      "epoch": 0.6388196392287333,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.900184600571715e-06,
+      "loss": 17.0404,
+      "step": 35330
+    },
+    {
+      "epoch": 0.6390004542978612,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.900156348265055e-06,
+      "loss": 17.3818,
+      "step": 35340
+    },
+    {
+      "epoch": 0.639181269366989,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.900128095958396e-06,
+      "loss": 16.7372,
+      "step": 35350
+    },
+    {
+      "epoch": 0.6393620844361169,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.900099843651737e-06,
+      "loss": 16.9702,
+      "step": 35360
+    },
+    {
+      "epoch": 0.6395428995052448,
+      "grad_norm": 45.375,
+      "learning_rate": 9.900071591345076e-06,
+      "loss": 17.1497,
+      "step": 35370
+    },
+    {
+      "epoch": 0.6397237145743726,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.900043339038416e-06,
+      "loss": 16.9083,
+      "step": 35380
+    },
+    {
+      "epoch": 0.6399045296435005,
+      "grad_norm": 45.34375,
+      "learning_rate": 9.900015086731757e-06,
+      "loss": 16.762,
+      "step": 35390
+    },
+    {
+      "epoch": 0.6400853447126283,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.899986834425097e-06,
+      "loss": 17.2308,
+      "step": 35400
+    },
+    {
+      "epoch": 0.6402661597817562,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.899958582118438e-06,
+      "loss": 16.7875,
+      "step": 35410
+    },
+    {
+      "epoch": 0.6404469748508841,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.899930329811779e-06,
+      "loss": 16.6921,
+      "step": 35420
+    },
+    {
+      "epoch": 0.6406277899200119,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.89990207750512e-06,
+      "loss": 16.8197,
+      "step": 35430
+    },
+    {
+      "epoch": 0.6408086049891398,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.89987382519846e-06,
+      "loss": 16.9357,
+      "step": 35440
+    },
+    {
+      "epoch": 0.6409894200582676,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.8998455728918e-06,
+      "loss": 16.8973,
+      "step": 35450
+    },
+    {
+      "epoch": 0.6411702351273956,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.89981732058514e-06,
+      "loss": 16.7642,
+      "step": 35460
+    },
+    {
+      "epoch": 0.6413510501965234,
+      "grad_norm": 46.84375,
+      "learning_rate": 9.89978906827848e-06,
+      "loss": 16.9361,
+      "step": 35470
+    },
+    {
+      "epoch": 0.6415318652656512,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.89976081597182e-06,
+      "loss": 17.1652,
+      "step": 35480
+    },
+    {
+      "epoch": 0.6417126803347791,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.899732563665161e-06,
+      "loss": 16.9174,
+      "step": 35490
+    },
+    {
+      "epoch": 0.6418934954039069,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.899704311358502e-06,
+      "loss": 16.7135,
+      "step": 35500
+    },
+    {
+      "epoch": 0.6420743104730349,
+      "grad_norm": 38.71875,
+      "learning_rate": 9.899676059051843e-06,
+      "loss": 16.6854,
+      "step": 35510
+    },
+    {
+      "epoch": 0.6422551255421627,
+      "grad_norm": 42.375,
+      "learning_rate": 9.899647806745183e-06,
+      "loss": 17.3483,
+      "step": 35520
+    },
+    {
+      "epoch": 0.6424359406112905,
+      "grad_norm": 40.625,
+      "learning_rate": 9.899619554438524e-06,
+      "loss": 16.9488,
+      "step": 35530
+    },
+    {
+      "epoch": 0.6426167556804184,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.899591302131863e-06,
+      "loss": 17.0364,
+      "step": 35540
+    },
+    {
+      "epoch": 0.6427975707495462,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.899563049825203e-06,
+      "loss": 17.0453,
+      "step": 35550
+    },
+    {
+      "epoch": 0.6429783858186742,
+      "grad_norm": 42.0,
+      "learning_rate": 9.899534797518544e-06,
+      "loss": 17.2179,
+      "step": 35560
+    },
+    {
+      "epoch": 0.643159200887802,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.899506545211885e-06,
+      "loss": 16.6237,
+      "step": 35570
+    },
+    {
+      "epoch": 0.6433400159569298,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.899478292905225e-06,
+      "loss": 16.5776,
+      "step": 35580
+    },
+    {
+      "epoch": 0.6435208310260577,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.899450040598564e-06,
+      "loss": 17.0083,
+      "step": 35590
+    },
+    {
+      "epoch": 0.6437016460951855,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.899421788291906e-06,
+      "loss": 16.8802,
+      "step": 35600
+    },
+    {
+      "epoch": 0.6438824611643135,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.899393535985247e-06,
+      "loss": 16.8426,
+      "step": 35610
+    },
+    {
+      "epoch": 0.6440632762334413,
+      "grad_norm": 40.5,
+      "learning_rate": 9.899365283678588e-06,
+      "loss": 16.6048,
+      "step": 35620
+    },
+    {
+      "epoch": 0.6442440913025692,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.899337031371927e-06,
+      "loss": 16.8645,
+      "step": 35630
+    },
+    {
+      "epoch": 0.644424906371697,
+      "grad_norm": 42.5,
+      "learning_rate": 9.899308779065267e-06,
+      "loss": 17.0983,
+      "step": 35640
+    },
+    {
+      "epoch": 0.6446057214408248,
+      "grad_norm": 40.625,
+      "learning_rate": 9.899280526758608e-06,
+      "loss": 16.9962,
+      "step": 35650
+    },
+    {
+      "epoch": 0.6447865365099528,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.899252274451949e-06,
+      "loss": 16.5154,
+      "step": 35660
+    },
+    {
+      "epoch": 0.6449673515790806,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.899224022145289e-06,
+      "loss": 17.1147,
+      "step": 35670
+    },
+    {
+      "epoch": 0.6451481666482085,
+      "grad_norm": 43.75,
+      "learning_rate": 9.899195769838628e-06,
+      "loss": 17.1466,
+      "step": 35680
+    },
+    {
+      "epoch": 0.6453289817173363,
+      "grad_norm": 43.875,
+      "learning_rate": 9.89916751753197e-06,
+      "loss": 16.6516,
+      "step": 35690
+    },
+    {
+      "epoch": 0.6455097967864641,
+      "grad_norm": 44.375,
+      "learning_rate": 9.899139265225311e-06,
+      "loss": 16.9173,
+      "step": 35700
+    },
+    {
+      "epoch": 0.6456906118555921,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.89911101291865e-06,
+      "loss": 16.9888,
+      "step": 35710
+    },
+    {
+      "epoch": 0.6458714269247199,
+      "grad_norm": 40.125,
+      "learning_rate": 9.89908276061199e-06,
+      "loss": 16.8233,
+      "step": 35720
+    },
+    {
+      "epoch": 0.6460522419938478,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.899054508305331e-06,
+      "loss": 16.9166,
+      "step": 35730
+    },
+    {
+      "epoch": 0.6462330570629756,
+      "grad_norm": 38.09375,
+      "learning_rate": 9.899026255998672e-06,
+      "loss": 16.7543,
+      "step": 35740
+    },
+    {
+      "epoch": 0.6464138721321034,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.898998003692012e-06,
+      "loss": 17.1185,
+      "step": 35750
+    },
+    {
+      "epoch": 0.6465946872012314,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.898969751385353e-06,
+      "loss": 16.4558,
+      "step": 35760
+    },
+    {
+      "epoch": 0.6467755022703592,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.898941499078694e-06,
+      "loss": 16.7642,
+      "step": 35770
+    },
+    {
+      "epoch": 0.6469563173394871,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.898913246772034e-06,
+      "loss": 16.9298,
+      "step": 35780
+    },
+    {
+      "epoch": 0.6471371324086149,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.898884994465375e-06,
+      "loss": 17.1463,
+      "step": 35790
+    },
+    {
+      "epoch": 0.6473179474777428,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.898856742158714e-06,
+      "loss": 17.0144,
+      "step": 35800
+    },
+    {
+      "epoch": 0.6474987625468707,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.898828489852054e-06,
+      "loss": 16.8369,
+      "step": 35810
+    },
+    {
+      "epoch": 0.6476795776159985,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.898800237545395e-06,
+      "loss": 16.9437,
+      "step": 35820
+    },
+    {
+      "epoch": 0.6478603926851264,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.898771985238736e-06,
+      "loss": 16.7855,
+      "step": 35830
+    },
+    {
+      "epoch": 0.6480412077542542,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.898743732932076e-06,
+      "loss": 16.9263,
+      "step": 35840
+    },
+    {
+      "epoch": 0.6482220228233821,
+      "grad_norm": 45.625,
+      "learning_rate": 9.898715480625415e-06,
+      "loss": 16.818,
+      "step": 35850
+    },
+    {
+      "epoch": 0.64840283789251,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.898687228318758e-06,
+      "loss": 16.6835,
+      "step": 35860
+    },
+    {
+      "epoch": 0.6485836529616378,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.898658976012098e-06,
+      "loss": 16.8036,
+      "step": 35870
+    },
+    {
+      "epoch": 0.6487644680307657,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.898630723705439e-06,
+      "loss": 17.0221,
+      "step": 35880
+    },
+    {
+      "epoch": 0.6489452830998935,
+      "grad_norm": 44.25,
+      "learning_rate": 9.898602471398778e-06,
+      "loss": 16.639,
+      "step": 35890
+    },
+    {
+      "epoch": 0.6491260981690214,
+      "grad_norm": 47.71875,
+      "learning_rate": 9.898574219092118e-06,
+      "loss": 16.8011,
+      "step": 35900
+    },
+    {
+      "epoch": 0.6493069132381493,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.898545966785459e-06,
+      "loss": 16.8954,
+      "step": 35910
+    },
+    {
+      "epoch": 0.6494877283072771,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.8985177144788e-06,
+      "loss": 16.8521,
+      "step": 35920
+    },
+    {
+      "epoch": 0.649668543376405,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.89848946217214e-06,
+      "loss": 17.0667,
+      "step": 35930
+    },
+    {
+      "epoch": 0.6498493584455328,
+      "grad_norm": 41.75,
+      "learning_rate": 9.89846120986548e-06,
+      "loss": 17.2669,
+      "step": 35940
+    },
+    {
+      "epoch": 0.6500301735146607,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.898432957558821e-06,
+      "loss": 16.4704,
+      "step": 35950
+    },
+    {
+      "epoch": 0.6502109885837886,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.898404705252162e-06,
+      "loss": 16.8687,
+      "step": 35960
+    },
+    {
+      "epoch": 0.6503918036529165,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.898376452945501e-06,
+      "loss": 17.2093,
+      "step": 35970
+    },
+    {
+      "epoch": 0.6505726187220443,
+      "grad_norm": 43.125,
+      "learning_rate": 9.898348200638842e-06,
+      "loss": 17.1447,
+      "step": 35980
+    },
+    {
+      "epoch": 0.6507534337911721,
+      "grad_norm": 46.625,
+      "learning_rate": 9.898319948332182e-06,
+      "loss": 16.7466,
+      "step": 35990
+    },
+    {
+      "epoch": 0.6509342488603,
+      "grad_norm": 43.875,
+      "learning_rate": 9.898291696025523e-06,
+      "loss": 16.8183,
+      "step": 36000
+    },
+    {
+      "epoch": 0.6511150639294279,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.898263443718864e-06,
+      "loss": 17.1838,
+      "step": 36010
+    },
+    {
+      "epoch": 0.6512958789985558,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.898235191412202e-06,
+      "loss": 16.6245,
+      "step": 36020
+    },
+    {
+      "epoch": 0.6514766940676836,
+      "grad_norm": 42.25,
+      "learning_rate": 9.898206939105543e-06,
+      "loss": 16.658,
+      "step": 36030
+    },
+    {
+      "epoch": 0.6516575091368114,
+      "grad_norm": 45.125,
+      "learning_rate": 9.898178686798885e-06,
+      "loss": 17.396,
+      "step": 36040
+    },
+    {
+      "epoch": 0.6518383242059393,
+      "grad_norm": 41.875,
+      "learning_rate": 9.898150434492226e-06,
+      "loss": 16.9268,
+      "step": 36050
+    },
+    {
+      "epoch": 0.6520191392750672,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.898122182185565e-06,
+      "loss": 16.686,
+      "step": 36060
+    },
+    {
+      "epoch": 0.6521999543441951,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.898093929878906e-06,
+      "loss": 16.705,
+      "step": 36070
+    },
+    {
+      "epoch": 0.6523807694133229,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.898065677572246e-06,
+      "loss": 16.6551,
+      "step": 36080
+    },
+    {
+      "epoch": 0.6525615844824507,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.898037425265587e-06,
+      "loss": 16.6608,
+      "step": 36090
+    },
+    {
+      "epoch": 0.6527423995515786,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.898009172958927e-06,
+      "loss": 17.0485,
+      "step": 36100
+    },
+    {
+      "epoch": 0.6529232146207065,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.897980920652266e-06,
+      "loss": 17.0704,
+      "step": 36110
+    },
+    {
+      "epoch": 0.6531040296898344,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.897952668345609e-06,
+      "loss": 16.8169,
+      "step": 36120
+    },
+    {
+      "epoch": 0.6532848447589622,
+      "grad_norm": 42.5,
+      "learning_rate": 9.89792441603895e-06,
+      "loss": 16.8907,
+      "step": 36130
+    },
+    {
+      "epoch": 0.6534656598280901,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.897896163732288e-06,
+      "loss": 16.8032,
+      "step": 36140
+    },
+    {
+      "epoch": 0.6536464748972179,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.897867911425629e-06,
+      "loss": 16.6132,
+      "step": 36150
+    },
+    {
+      "epoch": 0.6538272899663458,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.89783965911897e-06,
+      "loss": 16.8102,
+      "step": 36160
+    },
+    {
+      "epoch": 0.6540081050354737,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.89781140681231e-06,
+      "loss": 17.1489,
+      "step": 36170
+    },
+    {
+      "epoch": 0.6541889201046015,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.89778315450565e-06,
+      "loss": 16.9509,
+      "step": 36180
+    },
+    {
+      "epoch": 0.6543697351737294,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.89775490219899e-06,
+      "loss": 17.1862,
+      "step": 36190
+    },
+    {
+      "epoch": 0.6545505502428572,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.89772664989233e-06,
+      "loss": 17.2136,
+      "step": 36200
+    },
+    {
+      "epoch": 0.6547313653119851,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.897698397585673e-06,
+      "loss": 16.7174,
+      "step": 36210
+    },
+    {
+      "epoch": 0.654912180381113,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.897670145279013e-06,
+      "loss": 16.752,
+      "step": 36220
+    },
+    {
+      "epoch": 0.6550929954502408,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.897641892972352e-06,
+      "loss": 16.7826,
+      "step": 36230
+    },
+    {
+      "epoch": 0.6552738105193687,
+      "grad_norm": 42.875,
+      "learning_rate": 9.897613640665693e-06,
+      "loss": 16.906,
+      "step": 36240
+    },
+    {
+      "epoch": 0.6554546255884965,
+      "grad_norm": 42.625,
+      "learning_rate": 9.897585388359033e-06,
+      "loss": 16.7243,
+      "step": 36250
+    },
+    {
+      "epoch": 0.6556354406576244,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.897557136052374e-06,
+      "loss": 17.1138,
+      "step": 36260
+    },
+    {
+      "epoch": 0.6558162557267523,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.897528883745715e-06,
+      "loss": 17.3506,
+      "step": 36270
+    },
+    {
+      "epoch": 0.6559970707958801,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.897500631439054e-06,
+      "loss": 17.1032,
+      "step": 36280
+    },
+    {
+      "epoch": 0.656177885865008,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.897472379132394e-06,
+      "loss": 16.862,
+      "step": 36290
+    },
+    {
+      "epoch": 0.6563587009341358,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.897444126825736e-06,
+      "loss": 17.4687,
+      "step": 36300
+    },
+    {
+      "epoch": 0.6565395160032637,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.897415874519075e-06,
+      "loss": 16.8392,
+      "step": 36310
+    },
+    {
+      "epoch": 0.6567203310723916,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.897387622212416e-06,
+      "loss": 17.142,
+      "step": 36320
+    },
+    {
+      "epoch": 0.6569011461415194,
+      "grad_norm": 42.125,
+      "learning_rate": 9.897359369905757e-06,
+      "loss": 16.773,
+      "step": 36330
+    },
+    {
+      "epoch": 0.6570819612106473,
+      "grad_norm": 41.375,
+      "learning_rate": 9.897331117599097e-06,
+      "loss": 16.9312,
+      "step": 36340
+    },
+    {
+      "epoch": 0.6572627762797751,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.897302865292438e-06,
+      "loss": 17.084,
+      "step": 36350
+    },
+    {
+      "epoch": 0.6574435913489031,
+      "grad_norm": 46.78125,
+      "learning_rate": 9.897274612985779e-06,
+      "loss": 17.0247,
+      "step": 36360
+    },
+    {
+      "epoch": 0.6576244064180309,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.897246360679117e-06,
+      "loss": 16.9223,
+      "step": 36370
+    },
+    {
+      "epoch": 0.6578052214871587,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.897218108372458e-06,
+      "loss": 16.8791,
+      "step": 36380
+    },
+    {
+      "epoch": 0.6579860365562866,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.8971898560658e-06,
+      "loss": 17.133,
+      "step": 36390
+    },
+    {
+      "epoch": 0.6581668516254144,
+      "grad_norm": 44.25,
+      "learning_rate": 9.89716160375914e-06,
+      "loss": 17.3585,
+      "step": 36400
+    },
+    {
+      "epoch": 0.6583476666945424,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.89713335145248e-06,
+      "loss": 16.954,
+      "step": 36410
+    },
+    {
+      "epoch": 0.6585284817636702,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.89710509914582e-06,
+      "loss": 16.6315,
+      "step": 36420
+    },
+    {
+      "epoch": 0.658709296832798,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.897076846839161e-06,
+      "loss": 17.1309,
+      "step": 36430
+    },
+    {
+      "epoch": 0.6588901119019259,
+      "grad_norm": 43.75,
+      "learning_rate": 9.897048594532502e-06,
+      "loss": 16.5603,
+      "step": 36440
+    },
+    {
+      "epoch": 0.6590709269710537,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.89702034222584e-06,
+      "loss": 17.2705,
+      "step": 36450
+    },
+    {
+      "epoch": 0.6592517420401817,
+      "grad_norm": 43.125,
+      "learning_rate": 9.896992089919181e-06,
+      "loss": 17.4744,
+      "step": 36460
+    },
+    {
+      "epoch": 0.6594325571093095,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.896963837612524e-06,
+      "loss": 16.9674,
+      "step": 36470
+    },
+    {
+      "epoch": 0.6596133721784373,
+      "grad_norm": 44.5,
+      "learning_rate": 9.896935585305864e-06,
+      "loss": 17.0317,
+      "step": 36480
+    },
+    {
+      "epoch": 0.6597941872475652,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.896907332999203e-06,
+      "loss": 17.1149,
+      "step": 36490
+    },
+    {
+      "epoch": 0.659975002316693,
+      "grad_norm": 43.75,
+      "learning_rate": 9.896879080692544e-06,
+      "loss": 16.7619,
+      "step": 36500
+    },
+    {
+      "epoch": 0.660155817385821,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.896850828385884e-06,
+      "loss": 16.9837,
+      "step": 36510
+    },
+    {
+      "epoch": 0.6603366324549488,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.896822576079225e-06,
+      "loss": 16.7384,
+      "step": 36520
+    },
+    {
+      "epoch": 0.6605174475240767,
+      "grad_norm": 42.375,
+      "learning_rate": 9.896794323772566e-06,
+      "loss": 16.8103,
+      "step": 36530
+    },
+    {
+      "epoch": 0.6606982625932045,
+      "grad_norm": 48.125,
+      "learning_rate": 9.896766071465905e-06,
+      "loss": 16.7246,
+      "step": 36540
+    },
+    {
+      "epoch": 0.6608790776623323,
+      "grad_norm": 41.125,
+      "learning_rate": 9.896737819159245e-06,
+      "loss": 16.5504,
+      "step": 36550
+    },
+    {
+      "epoch": 0.6610598927314603,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.896709566852588e-06,
+      "loss": 17.1036,
+      "step": 36560
+    },
+    {
+      "epoch": 0.6612407078005881,
+      "grad_norm": 42.0,
+      "learning_rate": 9.896681314545927e-06,
+      "loss": 16.7741,
+      "step": 36570
+    },
+    {
+      "epoch": 0.661421522869716,
+      "grad_norm": 47.5,
+      "learning_rate": 9.896653062239267e-06,
+      "loss": 16.8923,
+      "step": 36580
+    },
+    {
+      "epoch": 0.6616023379388438,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.896624809932608e-06,
+      "loss": 16.797,
+      "step": 36590
+    },
+    {
+      "epoch": 0.6617831530079716,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.896596557625948e-06,
+      "loss": 16.9909,
+      "step": 36600
+    },
+    {
+      "epoch": 0.6619639680770996,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.896568305319289e-06,
+      "loss": 16.6533,
+      "step": 36610
+    },
+    {
+      "epoch": 0.6621447831462274,
+      "grad_norm": 43.75,
+      "learning_rate": 9.896540053012628e-06,
+      "loss": 16.85,
+      "step": 36620
+    },
+    {
+      "epoch": 0.6623255982153553,
+      "grad_norm": 45.125,
+      "learning_rate": 9.896511800705969e-06,
+      "loss": 16.742,
+      "step": 36630
+    },
+    {
+      "epoch": 0.6625064132844831,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.89648354839931e-06,
+      "loss": 16.9515,
+      "step": 36640
+    },
+    {
+      "epoch": 0.6626872283536109,
+      "grad_norm": 42.75,
+      "learning_rate": 9.896455296092652e-06,
+      "loss": 17.0241,
+      "step": 36650
+    },
+    {
+      "epoch": 0.6628680434227389,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.89642704378599e-06,
+      "loss": 17.213,
+      "step": 36660
+    },
+    {
+      "epoch": 0.6630488584918667,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.896398791479331e-06,
+      "loss": 16.7815,
+      "step": 36670
+    },
+    {
+      "epoch": 0.6632296735609946,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.896370539172672e-06,
+      "loss": 17.3715,
+      "step": 36680
+    },
+    {
+      "epoch": 0.6634104886301224,
+      "grad_norm": 45.6875,
+      "learning_rate": 9.896342286866012e-06,
+      "loss": 17.1651,
+      "step": 36690
+    },
+    {
+      "epoch": 0.6635913036992503,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.896314034559353e-06,
+      "loss": 17.0294,
+      "step": 36700
+    },
+    {
+      "epoch": 0.6637721187683782,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.896285782252692e-06,
+      "loss": 16.8683,
+      "step": 36710
+    },
+    {
+      "epoch": 0.663952933837506,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.896257529946032e-06,
+      "loss": 17.0183,
+      "step": 36720
+    },
+    {
+      "epoch": 0.6641337489066339,
+      "grad_norm": 40.625,
+      "learning_rate": 9.896229277639373e-06,
+      "loss": 16.8197,
+      "step": 36730
+    },
+    {
+      "epoch": 0.6643145639757617,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.896201025332714e-06,
+      "loss": 16.8733,
+      "step": 36740
+    },
+    {
+      "epoch": 0.6644953790448896,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.896172773026054e-06,
+      "loss": 16.7025,
+      "step": 36750
+    },
+    {
+      "epoch": 0.6646761941140175,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.896144520719395e-06,
+      "loss": 16.9805,
+      "step": 36760
+    },
+    {
+      "epoch": 0.6648570091831453,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.896116268412736e-06,
+      "loss": 17.3065,
+      "step": 36770
+    },
+    {
+      "epoch": 0.6650378242522732,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.896088016106076e-06,
+      "loss": 17.0735,
+      "step": 36780
+    },
+    {
+      "epoch": 0.665218639321401,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.896059763799417e-06,
+      "loss": 17.0477,
+      "step": 36790
+    },
+    {
+      "epoch": 0.6653994543905289,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.896031511492756e-06,
+      "loss": 16.4885,
+      "step": 36800
+    },
+    {
+      "epoch": 0.6655802694596568,
+      "grad_norm": 42.0,
+      "learning_rate": 9.896003259186096e-06,
+      "loss": 17.0502,
+      "step": 36810
+    },
+    {
+      "epoch": 0.6657610845287846,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.895975006879439e-06,
+      "loss": 16.9869,
+      "step": 36820
+    },
+    {
+      "epoch": 0.6659418995979125,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.895946754572778e-06,
+      "loss": 16.7416,
+      "step": 36830
+    },
+    {
+      "epoch": 0.6661227146670403,
+      "grad_norm": 44.53125,
+      "learning_rate": 9.895918502266118e-06,
+      "loss": 16.9338,
+      "step": 36840
+    },
+    {
+      "epoch": 0.6663035297361682,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.895890249959459e-06,
+      "loss": 16.6768,
+      "step": 36850
+    },
+    {
+      "epoch": 0.6664843448052961,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.8958619976528e-06,
+      "loss": 16.521,
+      "step": 36860
+    },
+    {
+      "epoch": 0.666665159874424,
+      "grad_norm": 38.9375,
+      "learning_rate": 9.89583374534614e-06,
+      "loss": 16.6117,
+      "step": 36870
+    },
+    {
+      "epoch": 0.6668459749435518,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.895805493039479e-06,
+      "loss": 16.6247,
+      "step": 36880
+    },
+    {
+      "epoch": 0.6670267900126796,
+      "grad_norm": 41.25,
+      "learning_rate": 9.89577724073282e-06,
+      "loss": 16.7686,
+      "step": 36890
+    },
+    {
+      "epoch": 0.6672076050818075,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.89574898842616e-06,
+      "loss": 16.936,
+      "step": 36900
+    },
+    {
+      "epoch": 0.6673884201509354,
+      "grad_norm": 39.03125,
+      "learning_rate": 9.895720736119503e-06,
+      "loss": 17.2432,
+      "step": 36910
+    },
+    {
+      "epoch": 0.6675692352200633,
+      "grad_norm": 41.75,
+      "learning_rate": 9.895692483812842e-06,
+      "loss": 17.0423,
+      "step": 36920
+    },
+    {
+      "epoch": 0.6677500502891911,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.895664231506182e-06,
+      "loss": 16.7086,
+      "step": 36930
+    },
+    {
+      "epoch": 0.6679308653583189,
+      "grad_norm": 40.09375,
+      "learning_rate": 9.895635979199523e-06,
+      "loss": 17.0203,
+      "step": 36940
+    },
+    {
+      "epoch": 0.6681116804274468,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.895607726892863e-06,
+      "loss": 16.739,
+      "step": 36950
+    },
+    {
+      "epoch": 0.6682924954965747,
+      "grad_norm": 40.375,
+      "learning_rate": 9.895579474586204e-06,
+      "loss": 16.8941,
+      "step": 36960
+    },
+    {
+      "epoch": 0.6684733105657026,
+      "grad_norm": 44.625,
+      "learning_rate": 9.895551222279543e-06,
+      "loss": 17.0405,
+      "step": 36970
+    },
+    {
+      "epoch": 0.6686541256348304,
+      "grad_norm": 43.5,
+      "learning_rate": 9.895522969972884e-06,
+      "loss": 17.3071,
+      "step": 36980
+    },
+    {
+      "epoch": 0.6688349407039582,
+      "grad_norm": 41.5,
+      "learning_rate": 9.895494717666224e-06,
+      "loss": 16.9758,
+      "step": 36990
+    },
+    {
+      "epoch": 0.6690157557730861,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.895466465359565e-06,
+      "loss": 17.0213,
+      "step": 37000
+    },
+    {
+      "epoch": 0.669196570842214,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.895438213052905e-06,
+      "loss": 16.3765,
+      "step": 37010
+    },
+    {
+      "epoch": 0.6693773859113419,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.895409960746246e-06,
+      "loss": 16.96,
+      "step": 37020
+    },
+    {
+      "epoch": 0.6695582009804697,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.895381708439587e-06,
+      "loss": 16.0563,
+      "step": 37030
+    },
+    {
+      "epoch": 0.6697390160495976,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.895353456132927e-06,
+      "loss": 16.8258,
+      "step": 37040
+    },
+    {
+      "epoch": 0.6699198311187254,
+      "grad_norm": 41.625,
+      "learning_rate": 9.895325203826266e-06,
+      "loss": 16.9237,
+      "step": 37050
+    },
+    {
+      "epoch": 0.6701006461878533,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.895296951519607e-06,
+      "loss": 16.9691,
+      "step": 37060
+    },
+    {
+      "epoch": 0.6702814612569812,
+      "grad_norm": 45.4375,
+      "learning_rate": 9.895268699212947e-06,
+      "loss": 16.9079,
+      "step": 37070
+    },
+    {
+      "epoch": 0.670462276326109,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.895240446906288e-06,
+      "loss": 16.7044,
+      "step": 37080
+    },
+    {
+      "epoch": 0.6706430913952369,
+      "grad_norm": 42.5,
+      "learning_rate": 9.895212194599629e-06,
+      "loss": 16.7271,
+      "step": 37090
+    },
+    {
+      "epoch": 0.6708239064643647,
+      "grad_norm": 41.75,
+      "learning_rate": 9.89518394229297e-06,
+      "loss": 16.8627,
+      "step": 37100
+    },
+    {
+      "epoch": 0.6710047215334926,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.89515568998631e-06,
+      "loss": 16.9436,
+      "step": 37110
+    },
+    {
+      "epoch": 0.6711855366026205,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.89512743767965e-06,
+      "loss": 16.9911,
+      "step": 37120
+    },
+    {
+      "epoch": 0.6713663516717483,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.895099185372991e-06,
+      "loss": 16.8614,
+      "step": 37130
+    },
+    {
+      "epoch": 0.6715471667408762,
+      "grad_norm": 39.46875,
+      "learning_rate": 9.89507093306633e-06,
+      "loss": 16.749,
+      "step": 37140
+    },
+    {
+      "epoch": 0.671727981810004,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.89504268075967e-06,
+      "loss": 17.2616,
+      "step": 37150
+    },
+    {
+      "epoch": 0.6719087968791319,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.895014428453011e-06,
+      "loss": 17.2514,
+      "step": 37160
+    },
+    {
+      "epoch": 0.6720896119482598,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.894986176146352e-06,
+      "loss": 16.9123,
+      "step": 37170
+    },
+    {
+      "epoch": 0.6722704270173876,
+      "grad_norm": 45.4375,
+      "learning_rate": 9.894957923839693e-06,
+      "loss": 16.9001,
+      "step": 37180
+    },
+    {
+      "epoch": 0.6724512420865155,
+      "grad_norm": 42.75,
+      "learning_rate": 9.894929671533033e-06,
+      "loss": 16.7307,
+      "step": 37190
+    },
+    {
+      "epoch": 0.6726320571556433,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.894901419226374e-06,
+      "loss": 16.4719,
+      "step": 37200
+    },
+    {
+      "epoch": 0.6728128722247713,
+      "grad_norm": 41.375,
+      "learning_rate": 9.894873166919714e-06,
+      "loss": 16.7455,
+      "step": 37210
+    },
+    {
+      "epoch": 0.6729936872938991,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.894844914613055e-06,
+      "loss": 17.014,
+      "step": 37220
+    },
+    {
+      "epoch": 0.6731745023630269,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.894816662306394e-06,
+      "loss": 16.5502,
+      "step": 37230
+    },
+    {
+      "epoch": 0.6733553174321548,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.894788409999735e-06,
+      "loss": 16.7081,
+      "step": 37240
+    },
+    {
+      "epoch": 0.6735361325012826,
+      "grad_norm": 43.0,
+      "learning_rate": 9.894760157693075e-06,
+      "loss": 16.4107,
+      "step": 37250
+    },
+    {
+      "epoch": 0.6737169475704106,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.894731905386416e-06,
+      "loss": 16.7003,
+      "step": 37260
+    },
+    {
+      "epoch": 0.6738977626395384,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.894703653079757e-06,
+      "loss": 17.0678,
+      "step": 37270
+    },
+    {
+      "epoch": 0.6740785777086662,
+      "grad_norm": 43.75,
+      "learning_rate": 9.894675400773097e-06,
+      "loss": 17.2967,
+      "step": 37280
+    },
+    {
+      "epoch": 0.6742593927777941,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.894647148466438e-06,
+      "loss": 16.9273,
+      "step": 37290
+    },
+    {
+      "epoch": 0.6744402078469219,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.894618896159778e-06,
+      "loss": 16.7979,
+      "step": 37300
+    },
+    {
+      "epoch": 0.6746210229160499,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.894590643853117e-06,
+      "loss": 16.8169,
+      "step": 37310
+    },
+    {
+      "epoch": 0.6748018379851777,
+      "grad_norm": 39.21875,
+      "learning_rate": 9.894562391546458e-06,
+      "loss": 16.4699,
+      "step": 37320
+    },
+    {
+      "epoch": 0.6749826530543055,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.894534139239799e-06,
+      "loss": 17.2835,
+      "step": 37330
+    },
+    {
+      "epoch": 0.6751634681234334,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.89450588693314e-06,
+      "loss": 17.0986,
+      "step": 37340
+    },
+    {
+      "epoch": 0.6753442831925612,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.89447763462648e-06,
+      "loss": 16.5451,
+      "step": 37350
+    },
+    {
+      "epoch": 0.6755250982616892,
+      "grad_norm": 38.75,
+      "learning_rate": 9.89444938231982e-06,
+      "loss": 16.5853,
+      "step": 37360
+    },
+    {
+      "epoch": 0.675705913330817,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.894421130013161e-06,
+      "loss": 16.5989,
+      "step": 37370
+    },
+    {
+      "epoch": 0.6758867283999449,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.894392877706502e-06,
+      "loss": 16.3647,
+      "step": 37380
+    },
+    {
+      "epoch": 0.6760675434690727,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.894364625399842e-06,
+      "loss": 16.8922,
+      "step": 37390
+    },
+    {
+      "epoch": 0.6762483585382005,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.894336373093181e-06,
+      "loss": 16.6042,
+      "step": 37400
+    },
+    {
+      "epoch": 0.6764291736073285,
+      "grad_norm": 44.375,
+      "learning_rate": 9.894308120786522e-06,
+      "loss": 16.704,
+      "step": 37410
+    },
+    {
+      "epoch": 0.6766099886764563,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.894279868479862e-06,
+      "loss": 16.9412,
+      "step": 37420
+    },
+    {
+      "epoch": 0.6767908037455842,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.894251616173203e-06,
+      "loss": 16.9898,
+      "step": 37430
+    },
+    {
+      "epoch": 0.676971618814712,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.894223363866544e-06,
+      "loss": 17.0491,
+      "step": 37440
+    },
+    {
+      "epoch": 0.6771524338838398,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.894195111559884e-06,
+      "loss": 16.946,
+      "step": 37450
+    },
+    {
+      "epoch": 0.6773332489529678,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.894166859253225e-06,
+      "loss": 16.8964,
+      "step": 37460
+    },
+    {
+      "epoch": 0.6775140640220956,
+      "grad_norm": 42.125,
+      "learning_rate": 9.894138606946566e-06,
+      "loss": 16.7398,
+      "step": 37470
+    },
+    {
+      "epoch": 0.6776948790912235,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.894110354639905e-06,
+      "loss": 17.4216,
+      "step": 37480
+    },
+    {
+      "epoch": 0.6778756941603513,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.894082102333245e-06,
+      "loss": 16.6143,
+      "step": 37490
+    },
+    {
+      "epoch": 0.6780565092294791,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.894053850026586e-06,
+      "loss": 16.975,
+      "step": 37500
+    },
+    {
+      "epoch": 0.6782373242986071,
+      "grad_norm": 41.0,
+      "learning_rate": 9.894025597719926e-06,
+      "loss": 17.1048,
+      "step": 37510
+    },
+    {
+      "epoch": 0.6784181393677349,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.893997345413267e-06,
+      "loss": 16.8411,
+      "step": 37520
+    },
+    {
+      "epoch": 0.6785989544368628,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.893969093106608e-06,
+      "loss": 16.628,
+      "step": 37530
+    },
+    {
+      "epoch": 0.6787797695059906,
+      "grad_norm": 42.875,
+      "learning_rate": 9.893940840799948e-06,
+      "loss": 16.8008,
+      "step": 37540
+    },
+    {
+      "epoch": 0.6789605845751185,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.893912588493289e-06,
+      "loss": 17.0204,
+      "step": 37550
+    },
+    {
+      "epoch": 0.6791413996442464,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.89388433618663e-06,
+      "loss": 16.8503,
+      "step": 37560
+    },
+    {
+      "epoch": 0.6793222147133742,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.893856083879968e-06,
+      "loss": 17.1805,
+      "step": 37570
+    },
+    {
+      "epoch": 0.6795030297825021,
+      "grad_norm": 45.84375,
+      "learning_rate": 9.893827831573309e-06,
+      "loss": 17.1783,
+      "step": 37580
+    },
+    {
+      "epoch": 0.6796838448516299,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.89379957926665e-06,
+      "loss": 16.4853,
+      "step": 37590
+    },
+    {
+      "epoch": 0.6798646599207578,
+      "grad_norm": 45.4375,
+      "learning_rate": 9.89377132695999e-06,
+      "loss": 16.8635,
+      "step": 37600
+    },
+    {
+      "epoch": 0.6800454749898857,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.893743074653331e-06,
+      "loss": 16.7238,
+      "step": 37610
+    },
+    {
+      "epoch": 0.6802262900590135,
+      "grad_norm": 44.0,
+      "learning_rate": 9.893714822346672e-06,
+      "loss": 16.7515,
+      "step": 37620
+    },
+    {
+      "epoch": 0.6804071051281414,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.893686570040012e-06,
+      "loss": 16.5855,
+      "step": 37630
+    },
+    {
+      "epoch": 0.6805879201972692,
+      "grad_norm": 45.28125,
+      "learning_rate": 9.893658317733353e-06,
+      "loss": 16.6717,
+      "step": 37640
+    },
+    {
+      "epoch": 0.6807687352663971,
+      "grad_norm": 41.75,
+      "learning_rate": 9.893630065426693e-06,
+      "loss": 16.5964,
+      "step": 37650
+    },
+    {
+      "epoch": 0.680949550335525,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.893601813120032e-06,
+      "loss": 17.0906,
+      "step": 37660
+    },
+    {
+      "epoch": 0.6811303654046528,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.893573560813373e-06,
+      "loss": 16.9528,
+      "step": 37670
+    },
+    {
+      "epoch": 0.6813111804737807,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.893545308506714e-06,
+      "loss": 16.7316,
+      "step": 37680
+    },
+    {
+      "epoch": 0.6814919955429085,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.893517056200054e-06,
+      "loss": 17.0295,
+      "step": 37690
+    },
+    {
+      "epoch": 0.6816728106120364,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.893488803893395e-06,
+      "loss": 16.8099,
+      "step": 37700
+    },
+    {
+      "epoch": 0.6818536256811643,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.893460551586735e-06,
+      "loss": 17.2334,
+      "step": 37710
+    },
+    {
+      "epoch": 0.6820344407502922,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.893432299280076e-06,
+      "loss": 17.0922,
+      "step": 37720
+    },
+    {
+      "epoch": 0.68221525581942,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.893404046973417e-06,
+      "loss": 16.5968,
+      "step": 37730
+    },
+    {
+      "epoch": 0.6823960708885478,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.893375794666756e-06,
+      "loss": 16.5034,
+      "step": 37740
+    },
+    {
+      "epoch": 0.6825768859576757,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.893347542360096e-06,
+      "loss": 16.4676,
+      "step": 37750
+    },
+    {
+      "epoch": 0.6827577010268036,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.893319290053437e-06,
+      "loss": 17.131,
+      "step": 37760
+    },
+    {
+      "epoch": 0.6829385160959315,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.893291037746777e-06,
+      "loss": 16.6227,
+      "step": 37770
+    },
+    {
+      "epoch": 0.6831193311650593,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.893262785440118e-06,
+      "loss": 17.1105,
+      "step": 37780
+    },
+    {
+      "epoch": 0.6833001462341871,
+      "grad_norm": 46.03125,
+      "learning_rate": 9.893234533133459e-06,
+      "loss": 16.6947,
+      "step": 37790
+    },
+    {
+      "epoch": 0.683480961303315,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.8932062808268e-06,
+      "loss": 16.7799,
+      "step": 37800
+    },
+    {
+      "epoch": 0.6836617763724429,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.89317802852014e-06,
+      "loss": 16.9629,
+      "step": 37810
+    },
+    {
+      "epoch": 0.6838425914415708,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.89314977621348e-06,
+      "loss": 16.9785,
+      "step": 37820
+    },
+    {
+      "epoch": 0.6840234065106986,
+      "grad_norm": 47.46875,
+      "learning_rate": 9.89312152390682e-06,
+      "loss": 17.1893,
+      "step": 37830
+    },
+    {
+      "epoch": 0.6842042215798264,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.89309327160016e-06,
+      "loss": 16.745,
+      "step": 37840
+    },
+    {
+      "epoch": 0.6843850366489543,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.8930650192935e-06,
+      "loss": 16.7386,
+      "step": 37850
+    },
+    {
+      "epoch": 0.6845658517180822,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.893036766986841e-06,
+      "loss": 17.1872,
+      "step": 37860
+    },
+    {
+      "epoch": 0.6847466667872101,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.893008514680182e-06,
+      "loss": 17.1807,
+      "step": 37870
+    },
+    {
+      "epoch": 0.6849274818563379,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.892980262373523e-06,
+      "loss": 16.3967,
+      "step": 37880
+    },
+    {
+      "epoch": 0.6851082969254658,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.892952010066863e-06,
+      "loss": 17.071,
+      "step": 37890
+    },
+    {
+      "epoch": 0.6852891119945936,
+      "grad_norm": 40.75,
+      "learning_rate": 9.892923757760204e-06,
+      "loss": 16.4375,
+      "step": 37900
+    },
+    {
+      "epoch": 0.6854699270637215,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.892895505453543e-06,
+      "loss": 16.9052,
+      "step": 37910
+    },
+    {
+      "epoch": 0.6856507421328494,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.892867253146883e-06,
+      "loss": 17.2202,
+      "step": 37920
+    },
+    {
+      "epoch": 0.6858315572019772,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.892839000840224e-06,
+      "loss": 16.2978,
+      "step": 37930
+    },
+    {
+      "epoch": 0.6860123722711051,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.892810748533565e-06,
+      "loss": 16.8237,
+      "step": 37940
+    },
+    {
+      "epoch": 0.6861931873402329,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.892782496226905e-06,
+      "loss": 16.7933,
+      "step": 37950
+    },
+    {
+      "epoch": 0.6863740024093608,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.892754243920246e-06,
+      "loss": 16.5888,
+      "step": 37960
+    },
+    {
+      "epoch": 0.6865548174784887,
+      "grad_norm": 42.625,
+      "learning_rate": 9.892725991613587e-06,
+      "loss": 17.1759,
+      "step": 37970
+    },
+    {
+      "epoch": 0.6867356325476165,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.892697739306927e-06,
+      "loss": 16.5377,
+      "step": 37980
+    },
+    {
+      "epoch": 0.6869164476167444,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.892669487000268e-06,
+      "loss": 16.3997,
+      "step": 37990
+    },
+    {
+      "epoch": 0.6870972626858722,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.892641234693607e-06,
+      "loss": 16.5976,
+      "step": 38000
+    },
+    {
+      "epoch": 0.687278077755,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.892612982386947e-06,
+      "loss": 16.7352,
+      "step": 38010
+    },
+    {
+      "epoch": 0.687458892824128,
+      "grad_norm": 44.0,
+      "learning_rate": 9.892584730080288e-06,
+      "loss": 17.0146,
+      "step": 38020
+    },
+    {
+      "epoch": 0.6876397078932558,
+      "grad_norm": 41.5,
+      "learning_rate": 9.892556477773629e-06,
+      "loss": 16.9501,
+      "step": 38030
+    },
+    {
+      "epoch": 0.6878205229623837,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.89252822546697e-06,
+      "loss": 16.6916,
+      "step": 38040
+    },
+    {
+      "epoch": 0.6880013380315115,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.89249997316031e-06,
+      "loss": 16.8109,
+      "step": 38050
+    },
+    {
+      "epoch": 0.6881821531006395,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.89247172085365e-06,
+      "loss": 16.1281,
+      "step": 38060
+    },
+    {
+      "epoch": 0.6883629681697673,
+      "grad_norm": 39.875,
+      "learning_rate": 9.892443468546991e-06,
+      "loss": 16.5465,
+      "step": 38070
+    },
+    {
+      "epoch": 0.6885437832388951,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.892415216240332e-06,
+      "loss": 16.8856,
+      "step": 38080
+    },
+    {
+      "epoch": 0.688724598308023,
+      "grad_norm": 44.875,
+      "learning_rate": 9.89238696393367e-06,
+      "loss": 17.1266,
+      "step": 38090
+    },
+    {
+      "epoch": 0.6889054133771508,
+      "grad_norm": 48.3125,
+      "learning_rate": 9.892358711627011e-06,
+      "loss": 17.2518,
+      "step": 38100
+    },
+    {
+      "epoch": 0.6890862284462788,
+      "grad_norm": 45.1875,
+      "learning_rate": 9.892330459320352e-06,
+      "loss": 16.6606,
+      "step": 38110
+    },
+    {
+      "epoch": 0.6892670435154066,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.892302207013692e-06,
+      "loss": 17.1191,
+      "step": 38120
+    },
+    {
+      "epoch": 0.6894478585845344,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.892273954707033e-06,
+      "loss": 16.8843,
+      "step": 38130
+    },
+    {
+      "epoch": 0.6896286736536623,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.892245702400374e-06,
+      "loss": 17.0698,
+      "step": 38140
+    },
+    {
+      "epoch": 0.6898094887227901,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.892217450093714e-06,
+      "loss": 16.8898,
+      "step": 38150
+    },
+    {
+      "epoch": 0.6899903037919181,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.892189197787055e-06,
+      "loss": 16.6293,
+      "step": 38160
+    },
+    {
+      "epoch": 0.6901711188610459,
+      "grad_norm": 40.1875,
+      "learning_rate": 9.892160945480394e-06,
+      "loss": 16.8833,
+      "step": 38170
+    },
+    {
+      "epoch": 0.6903519339301737,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.892132693173735e-06,
+      "loss": 16.8337,
+      "step": 38180
+    },
+    {
+      "epoch": 0.6905327489993016,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.892104440867075e-06,
+      "loss": 16.8272,
+      "step": 38190
+    },
+    {
+      "epoch": 0.6907135640684294,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.892076188560416e-06,
+      "loss": 16.7386,
+      "step": 38200
+    },
+    {
+      "epoch": 0.6908943791375574,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.892047936253756e-06,
+      "loss": 17.0917,
+      "step": 38210
+    },
+    {
+      "epoch": 0.6910751942066852,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.892019683947097e-06,
+      "loss": 16.8651,
+      "step": 38220
+    },
+    {
+      "epoch": 0.6912560092758131,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.891991431640438e-06,
+      "loss": 17.1524,
+      "step": 38230
+    },
+    {
+      "epoch": 0.6914368243449409,
+      "grad_norm": 42.25,
+      "learning_rate": 9.891963179333778e-06,
+      "loss": 16.482,
+      "step": 38240
+    },
+    {
+      "epoch": 0.6916176394140687,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.891934927027119e-06,
+      "loss": 16.8363,
+      "step": 38250
+    },
+    {
+      "epoch": 0.6917984544831967,
+      "grad_norm": 39.59375,
+      "learning_rate": 9.891906674720458e-06,
+      "loss": 16.6902,
+      "step": 38260
+    },
+    {
+      "epoch": 0.6919792695523245,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.891878422413798e-06,
+      "loss": 17.1459,
+      "step": 38270
+    },
+    {
+      "epoch": 0.6921600846214524,
+      "grad_norm": 41.75,
+      "learning_rate": 9.891850170107139e-06,
+      "loss": 16.5241,
+      "step": 38280
+    },
+    {
+      "epoch": 0.6923408996905802,
+      "grad_norm": 43.375,
+      "learning_rate": 9.89182191780048e-06,
+      "loss": 16.95,
+      "step": 38290
+    },
+    {
+      "epoch": 0.692521714759708,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.89179366549382e-06,
+      "loss": 16.9044,
+      "step": 38300
+    },
+    {
+      "epoch": 0.692702529828836,
+      "grad_norm": 41.375,
+      "learning_rate": 9.891765413187161e-06,
+      "loss": 17.0597,
+      "step": 38310
+    },
+    {
+      "epoch": 0.6928833448979638,
+      "grad_norm": 39.875,
+      "learning_rate": 9.891737160880502e-06,
+      "loss": 16.8813,
+      "step": 38320
+    },
+    {
+      "epoch": 0.6930641599670917,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.891708908573842e-06,
+      "loss": 17.1459,
+      "step": 38330
+    },
+    {
+      "epoch": 0.6932449750362195,
+      "grad_norm": 42.5,
+      "learning_rate": 9.891680656267181e-06,
+      "loss": 16.6943,
+      "step": 38340
+    },
+    {
+      "epoch": 0.6934257901053473,
+      "grad_norm": 42.625,
+      "learning_rate": 9.891652403960522e-06,
+      "loss": 16.8816,
+      "step": 38350
+    },
+    {
+      "epoch": 0.6936066051744753,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.891624151653862e-06,
+      "loss": 16.8835,
+      "step": 38360
+    },
+    {
+      "epoch": 0.6937874202436031,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.891595899347203e-06,
+      "loss": 17.0272,
+      "step": 38370
+    },
+    {
+      "epoch": 0.693968235312731,
+      "grad_norm": 40.53125,
+      "learning_rate": 9.891567647040544e-06,
+      "loss": 16.7614,
+      "step": 38380
+    },
+    {
+      "epoch": 0.6941490503818588,
+      "grad_norm": 42.125,
+      "learning_rate": 9.891539394733883e-06,
+      "loss": 16.4766,
+      "step": 38390
+    },
+    {
+      "epoch": 0.6943298654509867,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.891511142427225e-06,
+      "loss": 17.3731,
+      "step": 38400
+    },
+    {
+      "epoch": 0.6945106805201146,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.891482890120565e-06,
+      "loss": 16.6749,
+      "step": 38410
+    },
+    {
+      "epoch": 0.6946914955892424,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.891454637813906e-06,
+      "loss": 16.6835,
+      "step": 38420
+    },
+    {
+      "epoch": 0.6948723106583703,
+      "grad_norm": 44.875,
+      "learning_rate": 9.891426385507245e-06,
+      "loss": 16.9625,
+      "step": 38430
+    },
+    {
+      "epoch": 0.6950531257274981,
+      "grad_norm": 43.0,
+      "learning_rate": 9.891398133200586e-06,
+      "loss": 16.6631,
+      "step": 38440
+    },
+    {
+      "epoch": 0.695233940796626,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.891369880893926e-06,
+      "loss": 16.7724,
+      "step": 38450
+    },
+    {
+      "epoch": 0.6954147558657539,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.891341628587267e-06,
+      "loss": 16.9028,
+      "step": 38460
+    },
+    {
+      "epoch": 0.6955955709348817,
+      "grad_norm": 42.5,
+      "learning_rate": 9.891313376280607e-06,
+      "loss": 17.0701,
+      "step": 38470
+    },
+    {
+      "epoch": 0.6957763860040096,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.891285123973946e-06,
+      "loss": 16.3494,
+      "step": 38480
+    },
+    {
+      "epoch": 0.6959572010731374,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.891256871667289e-06,
+      "loss": 16.4693,
+      "step": 38490
+    },
+    {
+      "epoch": 0.6961380161422653,
+      "grad_norm": 43.25,
+      "learning_rate": 9.89122861936063e-06,
+      "loss": 16.8401,
+      "step": 38500
+    },
+    {
+      "epoch": 0.6963188312113932,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.891200367053968e-06,
+      "loss": 16.548,
+      "step": 38510
+    },
+    {
+      "epoch": 0.696499646280521,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.891172114747309e-06,
+      "loss": 17.2683,
+      "step": 38520
+    },
+    {
+      "epoch": 0.6966804613496489,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.89114386244065e-06,
+      "loss": 17.3941,
+      "step": 38530
+    },
+    {
+      "epoch": 0.6968612764187767,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.89111561013399e-06,
+      "loss": 17.3078,
+      "step": 38540
+    },
+    {
+      "epoch": 0.6970420914879046,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.89108735782733e-06,
+      "loss": 16.5073,
+      "step": 38550
+    },
+    {
+      "epoch": 0.6972229065570325,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.891059105520671e-06,
+      "loss": 16.5629,
+      "step": 38560
+    },
+    {
+      "epoch": 0.6974037216261604,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.891030853214012e-06,
+      "loss": 16.6385,
+      "step": 38570
+    },
+    {
+      "epoch": 0.6975845366952882,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.891002600907353e-06,
+      "loss": 16.8539,
+      "step": 38580
+    },
+    {
+      "epoch": 0.697765351764416,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.890974348600693e-06,
+      "loss": 16.9434,
+      "step": 38590
+    },
+    {
+      "epoch": 0.6979461668335439,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.890946096294032e-06,
+      "loss": 16.9127,
+      "step": 38600
+    },
+    {
+      "epoch": 0.6981269819026718,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.890917843987373e-06,
+      "loss": 16.9669,
+      "step": 38610
+    },
+    {
+      "epoch": 0.6983077969717997,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.890889591680713e-06,
+      "loss": 16.6195,
+      "step": 38620
+    },
+    {
+      "epoch": 0.6984886120409275,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.890861339374054e-06,
+      "loss": 16.8106,
+      "step": 38630
+    },
+    {
+      "epoch": 0.6986694271100553,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.890833087067395e-06,
+      "loss": 16.8928,
+      "step": 38640
+    },
+    {
+      "epoch": 0.6988502421791832,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.890804834760734e-06,
+      "loss": 16.7577,
+      "step": 38650
+    },
+    {
+      "epoch": 0.699031057248311,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.890776582454076e-06,
+      "loss": 16.6293,
+      "step": 38660
+    },
+    {
+      "epoch": 0.699211872317439,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.890748330147417e-06,
+      "loss": 16.8936,
+      "step": 38670
+    },
+    {
+      "epoch": 0.6993926873865668,
+      "grad_norm": 43.125,
+      "learning_rate": 9.890720077840757e-06,
+      "loss": 16.863,
+      "step": 38680
+    },
+    {
+      "epoch": 0.6995735024556946,
+      "grad_norm": 44.125,
+      "learning_rate": 9.890691825534096e-06,
+      "loss": 16.9209,
+      "step": 38690
+    },
+    {
+      "epoch": 0.6997543175248225,
+      "grad_norm": 47.375,
+      "learning_rate": 9.890663573227437e-06,
+      "loss": 16.7577,
+      "step": 38700
+    },
+    {
+      "epoch": 0.6999351325939503,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.890635320920777e-06,
+      "loss": 16.7093,
+      "step": 38710
+    },
+    {
+      "epoch": 0.7001159476630783,
+      "grad_norm": 43.75,
+      "learning_rate": 9.890607068614118e-06,
+      "loss": 16.6643,
+      "step": 38720
+    },
+    {
+      "epoch": 0.7002967627322061,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.890578816307459e-06,
+      "loss": 16.9856,
+      "step": 38730
+    },
+    {
+      "epoch": 0.700477577801334,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.890550564000798e-06,
+      "loss": 16.7979,
+      "step": 38740
+    },
+    {
+      "epoch": 0.7006583928704618,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.89052231169414e-06,
+      "loss": 17.2538,
+      "step": 38750
+    },
+    {
+      "epoch": 0.7008392079395896,
+      "grad_norm": 46.84375,
+      "learning_rate": 9.89049405938748e-06,
+      "loss": 16.9212,
+      "step": 38760
+    },
+    {
+      "epoch": 0.7010200230087176,
+      "grad_norm": 46.71875,
+      "learning_rate": 9.89046580708082e-06,
+      "loss": 17.0221,
+      "step": 38770
+    },
+    {
+      "epoch": 0.7012008380778454,
+      "grad_norm": 43.25,
+      "learning_rate": 9.89043755477416e-06,
+      "loss": 17.1704,
+      "step": 38780
+    },
+    {
+      "epoch": 0.7013816531469733,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.8904093024675e-06,
+      "loss": 17.0322,
+      "step": 38790
+    },
+    {
+      "epoch": 0.7015624682161011,
+      "grad_norm": 42.75,
+      "learning_rate": 9.890381050160841e-06,
+      "loss": 16.8613,
+      "step": 38800
+    },
+    {
+      "epoch": 0.701743283285229,
+      "grad_norm": 40.375,
+      "learning_rate": 9.890352797854182e-06,
+      "loss": 16.7168,
+      "step": 38810
+    },
+    {
+      "epoch": 0.7019240983543569,
+      "grad_norm": 40.125,
+      "learning_rate": 9.89032454554752e-06,
+      "loss": 16.8869,
+      "step": 38820
+    },
+    {
+      "epoch": 0.7021049134234847,
+      "grad_norm": 42.0,
+      "learning_rate": 9.890296293240861e-06,
+      "loss": 17.2882,
+      "step": 38830
+    },
+    {
+      "epoch": 0.7022857284926126,
+      "grad_norm": 40.0,
+      "learning_rate": 9.890268040934204e-06,
+      "loss": 16.2854,
+      "step": 38840
+    },
+    {
+      "epoch": 0.7024665435617404,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.890239788627544e-06,
+      "loss": 16.8362,
+      "step": 38850
+    },
+    {
+      "epoch": 0.7026473586308682,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.890211536320883e-06,
+      "loss": 16.914,
+      "step": 38860
+    },
+    {
+      "epoch": 0.7028281736999962,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.890183284014224e-06,
+      "loss": 16.5401,
+      "step": 38870
+    },
+    {
+      "epoch": 0.703008988769124,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.890155031707565e-06,
+      "loss": 16.6829,
+      "step": 38880
+    },
+    {
+      "epoch": 0.7031898038382519,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.890126779400905e-06,
+      "loss": 16.6298,
+      "step": 38890
+    },
+    {
+      "epoch": 0.7033706189073797,
+      "grad_norm": 40.875,
+      "learning_rate": 9.890098527094246e-06,
+      "loss": 16.8991,
+      "step": 38900
+    },
+    {
+      "epoch": 0.7035514339765077,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.890070274787585e-06,
+      "loss": 16.7942,
+      "step": 38910
+    },
+    {
+      "epoch": 0.7037322490456355,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.890042022480927e-06,
+      "loss": 16.8306,
+      "step": 38920
+    },
+    {
+      "epoch": 0.7039130641147633,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.890013770174268e-06,
+      "loss": 16.7857,
+      "step": 38930
+    },
+    {
+      "epoch": 0.7040938791838912,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.889985517867607e-06,
+      "loss": 17.0068,
+      "step": 38940
+    },
+    {
+      "epoch": 0.704274694253019,
+      "grad_norm": 45.0,
+      "learning_rate": 9.889957265560947e-06,
+      "loss": 17.0016,
+      "step": 38950
+    },
+    {
+      "epoch": 0.704455509322147,
+      "grad_norm": 44.0,
+      "learning_rate": 9.889929013254288e-06,
+      "loss": 16.9943,
+      "step": 38960
+    },
+    {
+      "epoch": 0.7046363243912748,
+      "grad_norm": 42.25,
+      "learning_rate": 9.889900760947628e-06,
+      "loss": 17.1728,
+      "step": 38970
+    },
+    {
+      "epoch": 0.7048171394604026,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.889872508640969e-06,
+      "loss": 16.7221,
+      "step": 38980
+    },
+    {
+      "epoch": 0.7049979545295305,
+      "grad_norm": 42.125,
+      "learning_rate": 9.88984425633431e-06,
+      "loss": 17.306,
+      "step": 38990
+    },
+    {
+      "epoch": 0.7051787695986583,
+      "grad_norm": 46.5625,
+      "learning_rate": 9.889816004027649e-06,
+      "loss": 16.8126,
+      "step": 39000
+    },
+    {
+      "epoch": 0.7053595846677863,
+      "grad_norm": 47.84375,
+      "learning_rate": 9.889787751720991e-06,
+      "loss": 17.1228,
+      "step": 39010
+    },
+    {
+      "epoch": 0.7055403997369141,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.889759499414332e-06,
+      "loss": 16.9815,
+      "step": 39020
+    },
+    {
+      "epoch": 0.7057212148060419,
+      "grad_norm": 46.75,
+      "learning_rate": 9.88973124710767e-06,
+      "loss": 16.8795,
+      "step": 39030
+    },
+    {
+      "epoch": 0.7059020298751698,
+      "grad_norm": 41.25,
+      "learning_rate": 9.889702994801011e-06,
+      "loss": 16.5388,
+      "step": 39040
+    },
+    {
+      "epoch": 0.7060828449442976,
+      "grad_norm": 38.8125,
+      "learning_rate": 9.889674742494352e-06,
+      "loss": 17.3997,
+      "step": 39050
+    },
+    {
+      "epoch": 0.7062636600134256,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.889646490187692e-06,
+      "loss": 17.334,
+      "step": 39060
+    },
+    {
+      "epoch": 0.7064444750825534,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.889618237881033e-06,
+      "loss": 16.7021,
+      "step": 39070
+    },
+    {
+      "epoch": 0.7066252901516813,
+      "grad_norm": 42.75,
+      "learning_rate": 9.889589985574372e-06,
+      "loss": 16.7919,
+      "step": 39080
+    },
+    {
+      "epoch": 0.7068061052208091,
+      "grad_norm": 42.375,
+      "learning_rate": 9.889561733267713e-06,
+      "loss": 16.9483,
+      "step": 39090
+    },
+    {
+      "epoch": 0.7069869202899369,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.889533480961055e-06,
+      "loss": 17.155,
+      "step": 39100
+    },
+    {
+      "epoch": 0.7071677353590649,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.889505228654395e-06,
+      "loss": 16.7376,
+      "step": 39110
+    },
+    {
+      "epoch": 0.7073485504281927,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.889476976347734e-06,
+      "loss": 16.9099,
+      "step": 39120
+    },
+    {
+      "epoch": 0.7075293654973206,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.889448724041075e-06,
+      "loss": 16.8503,
+      "step": 39130
+    },
+    {
+      "epoch": 0.7077101805664484,
+      "grad_norm": 41.125,
+      "learning_rate": 9.889420471734416e-06,
+      "loss": 17.4304,
+      "step": 39140
+    },
+    {
+      "epoch": 0.7078909956355762,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.889392219427756e-06,
+      "loss": 16.8346,
+      "step": 39150
+    },
+    {
+      "epoch": 0.7080718107047042,
+      "grad_norm": 40.125,
+      "learning_rate": 9.889363967121097e-06,
+      "loss": 17.1053,
+      "step": 39160
+    },
+    {
+      "epoch": 0.708252625773832,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.889335714814436e-06,
+      "loss": 17.1629,
+      "step": 39170
+    },
+    {
+      "epoch": 0.7084334408429599,
+      "grad_norm": 40.5,
+      "learning_rate": 9.889307462507776e-06,
+      "loss": 16.7123,
+      "step": 39180
+    },
+    {
+      "epoch": 0.7086142559120877,
+      "grad_norm": 41.75,
+      "learning_rate": 9.889279210201119e-06,
+      "loss": 16.7231,
+      "step": 39190
+    },
+    {
+      "epoch": 0.7087950709812155,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.889250957894458e-06,
+      "loss": 16.8257,
+      "step": 39200
+    },
+    {
+      "epoch": 0.7089758860503435,
+      "grad_norm": 41.625,
+      "learning_rate": 9.889222705587798e-06,
+      "loss": 16.6544,
+      "step": 39210
+    },
+    {
+      "epoch": 0.7091567011194713,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.889194453281139e-06,
+      "loss": 16.5631,
+      "step": 39220
+    },
+    {
+      "epoch": 0.7093375161885992,
+      "grad_norm": 45.625,
+      "learning_rate": 9.88916620097448e-06,
+      "loss": 16.7361,
+      "step": 39230
+    },
+    {
+      "epoch": 0.709518331257727,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.88913794866782e-06,
+      "loss": 16.61,
+      "step": 39240
+    },
+    {
+      "epoch": 0.7096991463268549,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.889109696361159e-06,
+      "loss": 16.7179,
+      "step": 39250
+    },
+    {
+      "epoch": 0.7098799613959828,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.8890814440545e-06,
+      "loss": 16.5795,
+      "step": 39260
+    },
+    {
+      "epoch": 0.7100607764651106,
+      "grad_norm": 42.25,
+      "learning_rate": 9.88905319174784e-06,
+      "loss": 16.9953,
+      "step": 39270
+    },
+    {
+      "epoch": 0.7102415915342385,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.889024939441183e-06,
+      "loss": 17.0232,
+      "step": 39280
+    },
+    {
+      "epoch": 0.7104224066033663,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.888996687134522e-06,
+      "loss": 16.6105,
+      "step": 39290
+    },
+    {
+      "epoch": 0.7106032216724942,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.888968434827862e-06,
+      "loss": 16.4751,
+      "step": 39300
+    },
+    {
+      "epoch": 0.710784036741622,
+      "grad_norm": 44.875,
+      "learning_rate": 9.888940182521203e-06,
+      "loss": 16.6902,
+      "step": 39310
+    },
+    {
+      "epoch": 0.7109648518107499,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.888911930214543e-06,
+      "loss": 16.8739,
+      "step": 39320
+    },
+    {
+      "epoch": 0.7111456668798778,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.888883677907884e-06,
+      "loss": 16.8996,
+      "step": 39330
+    },
+    {
+      "epoch": 0.7113264819490056,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.888855425601223e-06,
+      "loss": 17.1528,
+      "step": 39340
+    },
+    {
+      "epoch": 0.7115072970181335,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.888827173294564e-06,
+      "loss": 16.411,
+      "step": 39350
+    },
+    {
+      "epoch": 0.7116881120872613,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.888798920987906e-06,
+      "loss": 17.1764,
+      "step": 39360
+    },
+    {
+      "epoch": 0.7118689271563892,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.888770668681245e-06,
+      "loss": 16.745,
+      "step": 39370
+    },
+    {
+      "epoch": 0.7120497422255171,
+      "grad_norm": 42.875,
+      "learning_rate": 9.888742416374585e-06,
+      "loss": 16.6895,
+      "step": 39380
+    },
+    {
+      "epoch": 0.7122305572946449,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.888714164067926e-06,
+      "loss": 16.6713,
+      "step": 39390
+    },
+    {
+      "epoch": 0.7124113723637728,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.888685911761267e-06,
+      "loss": 16.8277,
+      "step": 39400
+    },
+    {
+      "epoch": 0.7125921874329006,
+      "grad_norm": 44.25,
+      "learning_rate": 9.888657659454607e-06,
+      "loss": 16.4465,
+      "step": 39410
+    },
+    {
+      "epoch": 0.7127730025020286,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.888629407147948e-06,
+      "loss": 17.0582,
+      "step": 39420
+    },
+    {
+      "epoch": 0.7129538175711564,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.888601154841287e-06,
+      "loss": 17.0685,
+      "step": 39430
+    },
+    {
+      "epoch": 0.7131346326402842,
+      "grad_norm": 43.5,
+      "learning_rate": 9.888572902534628e-06,
+      "loss": 16.3677,
+      "step": 39440
+    },
+    {
+      "epoch": 0.7133154477094121,
+      "grad_norm": 43.25,
+      "learning_rate": 9.88854465022797e-06,
+      "loss": 16.704,
+      "step": 39450
+    },
+    {
+      "epoch": 0.71349626277854,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.888516397921309e-06,
+      "loss": 17.0324,
+      "step": 39460
+    },
+    {
+      "epoch": 0.7136770778476679,
+      "grad_norm": 42.125,
+      "learning_rate": 9.88848814561465e-06,
+      "loss": 17.0155,
+      "step": 39470
+    },
+    {
+      "epoch": 0.7138578929167957,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.88845989330799e-06,
+      "loss": 17.1789,
+      "step": 39480
+    },
+    {
+      "epoch": 0.7140387079859235,
+      "grad_norm": 45.875,
+      "learning_rate": 9.88843164100133e-06,
+      "loss": 16.8495,
+      "step": 39490
+    },
+    {
+      "epoch": 0.7142195230550514,
+      "grad_norm": 44.125,
+      "learning_rate": 9.888403388694671e-06,
+      "loss": 16.6137,
+      "step": 39500
+    },
+    {
+      "epoch": 0.7144003381241792,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.88837513638801e-06,
+      "loss": 16.8421,
+      "step": 39510
+    },
+    {
+      "epoch": 0.7145811531933072,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.88834688408135e-06,
+      "loss": 16.8688,
+      "step": 39520
+    },
+    {
+      "epoch": 0.714761968262435,
+      "grad_norm": 42.75,
+      "learning_rate": 9.888318631774691e-06,
+      "loss": 16.5882,
+      "step": 39530
+    },
+    {
+      "epoch": 0.7149427833315628,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.888290379468034e-06,
+      "loss": 16.5181,
+      "step": 39540
+    },
+    {
+      "epoch": 0.7151235984006907,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.888262127161373e-06,
+      "loss": 17.2262,
+      "step": 39550
+    },
+    {
+      "epoch": 0.7153044134698185,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.888233874854713e-06,
+      "loss": 16.6752,
+      "step": 39560
+    },
+    {
+      "epoch": 0.7154852285389465,
+      "grad_norm": 46.625,
+      "learning_rate": 9.888205622548054e-06,
+      "loss": 17.0998,
+      "step": 39570
+    },
+    {
+      "epoch": 0.7156660436080743,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.888177370241395e-06,
+      "loss": 16.7745,
+      "step": 39580
+    },
+    {
+      "epoch": 0.7158468586772022,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.888149117934735e-06,
+      "loss": 16.4374,
+      "step": 39590
+    },
+    {
+      "epoch": 0.71602767374633,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.888120865628074e-06,
+      "loss": 16.4207,
+      "step": 39600
+    },
+    {
+      "epoch": 0.7162084888154578,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.888092613321415e-06,
+      "loss": 16.7451,
+      "step": 39610
+    },
+    {
+      "epoch": 0.7163893038845858,
+      "grad_norm": 39.90625,
+      "learning_rate": 9.888064361014755e-06,
+      "loss": 17.1233,
+      "step": 39620
+    },
+    {
+      "epoch": 0.7165701189537136,
+      "grad_norm": 44.5,
+      "learning_rate": 9.888036108708096e-06,
+      "loss": 16.855,
+      "step": 39630
+    },
+    {
+      "epoch": 0.7167509340228415,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.888007856401437e-06,
+      "loss": 16.5099,
+      "step": 39640
+    },
+    {
+      "epoch": 0.7169317490919693,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.887979604094777e-06,
+      "loss": 17.1891,
+      "step": 39650
+    },
+    {
+      "epoch": 0.7171125641610971,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.887951351788118e-06,
+      "loss": 16.7949,
+      "step": 39660
+    },
+    {
+      "epoch": 0.7172933792302251,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.887923099481458e-06,
+      "loss": 16.9468,
+      "step": 39670
+    },
+    {
+      "epoch": 0.7174741942993529,
+      "grad_norm": 42.0,
+      "learning_rate": 9.887894847174797e-06,
+      "loss": 16.891,
+      "step": 39680
+    },
+    {
+      "epoch": 0.7176550093684808,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.887866594868138e-06,
+      "loss": 17.1045,
+      "step": 39690
+    },
+    {
+      "epoch": 0.7178358244376086,
+      "grad_norm": 43.125,
+      "learning_rate": 9.887838342561479e-06,
+      "loss": 17.314,
+      "step": 39700
+    },
+    {
+      "epoch": 0.7180166395067364,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.887810090254821e-06,
+      "loss": 16.633,
+      "step": 39710
+    },
+    {
+      "epoch": 0.7181974545758644,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.88778183794816e-06,
+      "loss": 16.8595,
+      "step": 39720
+    },
+    {
+      "epoch": 0.7183782696449922,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.8877535856415e-06,
+      "loss": 16.9203,
+      "step": 39730
+    },
+    {
+      "epoch": 0.7185590847141201,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.887725333334841e-06,
+      "loss": 16.9117,
+      "step": 39740
+    },
+    {
+      "epoch": 0.7187398997832479,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.887697081028182e-06,
+      "loss": 17.2357,
+      "step": 39750
+    },
+    {
+      "epoch": 0.7189207148523757,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.887668828721522e-06,
+      "loss": 16.8096,
+      "step": 39760
+    },
+    {
+      "epoch": 0.7191015299215037,
+      "grad_norm": 42.625,
+      "learning_rate": 9.887640576414861e-06,
+      "loss": 16.8199,
+      "step": 39770
+    },
+    {
+      "epoch": 0.7192823449906315,
+      "grad_norm": 44.71875,
+      "learning_rate": 9.887612324108202e-06,
+      "loss": 16.8323,
+      "step": 39780
+    },
+    {
+      "epoch": 0.7194631600597594,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.887584071801543e-06,
+      "loss": 16.9276,
+      "step": 39790
+    },
+    {
+      "epoch": 0.7196439751288872,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.887555819494883e-06,
+      "loss": 16.6748,
+      "step": 39800
+    },
+    {
+      "epoch": 0.7198247901980152,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.887527567188224e-06,
+      "loss": 16.8066,
+      "step": 39810
+    },
+    {
+      "epoch": 0.720005605267143,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.887499314881564e-06,
+      "loss": 17.2768,
+      "step": 39820
+    },
+    {
+      "epoch": 0.7201864203362708,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.887471062574905e-06,
+      "loss": 16.4379,
+      "step": 39830
+    },
+    {
+      "epoch": 0.7203672354053987,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.887442810268246e-06,
+      "loss": 16.8064,
+      "step": 39840
+    },
+    {
+      "epoch": 0.7205480504745265,
+      "grad_norm": 41.875,
+      "learning_rate": 9.887414557961586e-06,
+      "loss": 16.7328,
+      "step": 39850
+    },
+    {
+      "epoch": 0.7207288655436545,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.887386305654925e-06,
+      "loss": 16.8048,
+      "step": 39860
+    },
+    {
+      "epoch": 0.7209096806127823,
+      "grad_norm": 44.0,
+      "learning_rate": 9.887358053348266e-06,
+      "loss": 16.5774,
+      "step": 39870
+    },
+    {
+      "epoch": 0.7210904956819101,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.887329801041606e-06,
+      "loss": 16.6305,
+      "step": 39880
+    },
+    {
+      "epoch": 0.721271310751038,
+      "grad_norm": 40.125,
+      "learning_rate": 9.887301548734947e-06,
+      "loss": 16.7518,
+      "step": 39890
+    },
+    {
+      "epoch": 0.7214521258201658,
+      "grad_norm": 42.25,
+      "learning_rate": 9.887273296428288e-06,
+      "loss": 17.4451,
+      "step": 39900
+    },
+    {
+      "epoch": 0.7216329408892938,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.887245044121628e-06,
+      "loss": 16.9063,
+      "step": 39910
+    },
+    {
+      "epoch": 0.7218137559584216,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.887216791814969e-06,
+      "loss": 16.5798,
+      "step": 39920
+    },
+    {
+      "epoch": 0.7219945710275494,
+      "grad_norm": 42.125,
+      "learning_rate": 9.88718853950831e-06,
+      "loss": 16.8749,
+      "step": 39930
+    },
+    {
+      "epoch": 0.7221753860966773,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.887160287201648e-06,
+      "loss": 16.9119,
+      "step": 39940
+    },
+    {
+      "epoch": 0.7223562011658051,
+      "grad_norm": 42.5,
+      "learning_rate": 9.887132034894989e-06,
+      "loss": 16.7371,
+      "step": 39950
+    },
+    {
+      "epoch": 0.722537016234933,
+      "grad_norm": 41.25,
+      "learning_rate": 9.88710378258833e-06,
+      "loss": 17.1396,
+      "step": 39960
+    },
+    {
+      "epoch": 0.7227178313040609,
+      "grad_norm": 43.75,
+      "learning_rate": 9.88707553028167e-06,
+      "loss": 16.7851,
+      "step": 39970
+    },
+    {
+      "epoch": 0.7228986463731888,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.887047277975011e-06,
+      "loss": 17.002,
+      "step": 39980
+    },
+    {
+      "epoch": 0.7230794614423166,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.887019025668352e-06,
+      "loss": 17.0829,
+      "step": 39990
+    },
+    {
+      "epoch": 0.7232602765114444,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.886990773361692e-06,
+      "loss": 16.9125,
+      "step": 40000
+    },
+    {
+      "epoch": 0.7232602765114444,
+      "eval_loss": 2.107973098754883,
+      "eval_runtime": 229.6504,
+      "eval_samples_per_second": 3161.584,
+      "eval_steps_per_second": 49.401,
+      "step": 40000
+    },
+    {
+      "epoch": 0.7234410915805723,
+      "grad_norm": 41.5,
+      "learning_rate": 9.886962521055033e-06,
+      "loss": 16.4671,
+      "step": 40010
+    },
+    {
+      "epoch": 0.7236219066497002,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.886934268748373e-06,
+      "loss": 16.8505,
+      "step": 40020
+    },
+    {
+      "epoch": 0.7238027217188281,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.886906016441712e-06,
+      "loss": 16.8149,
+      "step": 40030
+    },
+    {
+      "epoch": 0.7239835367879559,
+      "grad_norm": 43.125,
+      "learning_rate": 9.886877764135053e-06,
+      "loss": 16.7169,
+      "step": 40040
+    },
+    {
+      "epoch": 0.7241643518570837,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.886849511828394e-06,
+      "loss": 16.8032,
+      "step": 40050
+    },
+    {
+      "epoch": 0.7243451669262116,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.886821259521734e-06,
+      "loss": 16.7357,
+      "step": 40060
+    },
+    {
+      "epoch": 0.7245259819953395,
+      "grad_norm": 43.5,
+      "learning_rate": 9.886793007215075e-06,
+      "loss": 16.8738,
+      "step": 40070
+    },
+    {
+      "epoch": 0.7247067970644674,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.886764754908416e-06,
+      "loss": 16.9363,
+      "step": 40080
+    },
+    {
+      "epoch": 0.7248876121335952,
+      "grad_norm": 44.875,
+      "learning_rate": 9.886736502601756e-06,
+      "loss": 16.6958,
+      "step": 40090
+    },
+    {
+      "epoch": 0.725068427202723,
+      "grad_norm": 41.625,
+      "learning_rate": 9.886708250295097e-06,
+      "loss": 16.8281,
+      "step": 40100
+    },
+    {
+      "epoch": 0.725249242271851,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.886679997988436e-06,
+      "loss": 16.6294,
+      "step": 40110
+    },
+    {
+      "epoch": 0.7254300573409788,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.886651745681776e-06,
+      "loss": 17.3219,
+      "step": 40120
+    },
+    {
+      "epoch": 0.7256108724101067,
+      "grad_norm": 44.75,
+      "learning_rate": 9.886623493375117e-06,
+      "loss": 17.1288,
+      "step": 40130
+    },
+    {
+      "epoch": 0.7257916874792345,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.886595241068458e-06,
+      "loss": 16.3775,
+      "step": 40140
+    },
+    {
+      "epoch": 0.7259725025483624,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.886566988761798e-06,
+      "loss": 16.2942,
+      "step": 40150
+    },
+    {
+      "epoch": 0.7261533176174902,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.886538736455139e-06,
+      "loss": 16.8152,
+      "step": 40160
+    },
+    {
+      "epoch": 0.7263341326866181,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.88651048414848e-06,
+      "loss": 17.041,
+      "step": 40170
+    },
+    {
+      "epoch": 0.726514947755746,
+      "grad_norm": 43.0,
+      "learning_rate": 9.88648223184182e-06,
+      "loss": 17.1616,
+      "step": 40180
+    },
+    {
+      "epoch": 0.7266957628248738,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.88645397953516e-06,
+      "loss": 16.3568,
+      "step": 40190
+    },
+    {
+      "epoch": 0.7268765778940017,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.8864257272285e-06,
+      "loss": 16.744,
+      "step": 40200
+    },
+    {
+      "epoch": 0.7270573929631295,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.88639747492184e-06,
+      "loss": 16.7687,
+      "step": 40210
+    },
+    {
+      "epoch": 0.7272382080322574,
+      "grad_norm": 42.75,
+      "learning_rate": 9.88636922261518e-06,
+      "loss": 16.7254,
+      "step": 40220
+    },
+    {
+      "epoch": 0.7274190231013853,
+      "grad_norm": 47.21875,
+      "learning_rate": 9.886340970308521e-06,
+      "loss": 16.8152,
+      "step": 40230
+    },
+    {
+      "epoch": 0.7275998381705131,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.886312718001862e-06,
+      "loss": 16.6617,
+      "step": 40240
+    },
+    {
+      "epoch": 0.727780653239641,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.886284465695203e-06,
+      "loss": 16.8171,
+      "step": 40250
+    },
+    {
+      "epoch": 0.7279614683087688,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.886256213388543e-06,
+      "loss": 16.6807,
+      "step": 40260
+    },
+    {
+      "epoch": 0.7281422833778967,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.886227961081884e-06,
+      "loss": 16.8896,
+      "step": 40270
+    },
+    {
+      "epoch": 0.7283230984470246,
+      "grad_norm": 44.75,
+      "learning_rate": 9.886199708775225e-06,
+      "loss": 17.2417,
+      "step": 40280
+    },
+    {
+      "epoch": 0.7285039135161524,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.886171456468563e-06,
+      "loss": 16.7568,
+      "step": 40290
+    },
+    {
+      "epoch": 0.7286847285852803,
+      "grad_norm": 46.21875,
+      "learning_rate": 9.886143204161904e-06,
+      "loss": 16.9543,
+      "step": 40300
+    },
+    {
+      "epoch": 0.7288655436544081,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.886114951855245e-06,
+      "loss": 16.9995,
+      "step": 40310
+    },
+    {
+      "epoch": 0.7290463587235361,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.886086699548585e-06,
+      "loss": 16.5499,
+      "step": 40320
+    },
+    {
+      "epoch": 0.7292271737926639,
+      "grad_norm": 45.96875,
+      "learning_rate": 9.886058447241926e-06,
+      "loss": 16.7233,
+      "step": 40330
+    },
+    {
+      "epoch": 0.7294079888617917,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.886030194935267e-06,
+      "loss": 17.1275,
+      "step": 40340
+    },
+    {
+      "epoch": 0.7295888039309196,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.886001942628607e-06,
+      "loss": 16.9403,
+      "step": 40350
+    },
+    {
+      "epoch": 0.7297696190000474,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.885973690321948e-06,
+      "loss": 16.5147,
+      "step": 40360
+    },
+    {
+      "epoch": 0.7299504340691754,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.885945438015287e-06,
+      "loss": 16.7906,
+      "step": 40370
+    },
+    {
+      "epoch": 0.7301312491383032,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.885917185708627e-06,
+      "loss": 17.0495,
+      "step": 40380
+    },
+    {
+      "epoch": 0.730312064207431,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.885888933401968e-06,
+      "loss": 17.0379,
+      "step": 40390
+    },
+    {
+      "epoch": 0.7304928792765589,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.885860681095309e-06,
+      "loss": 16.7473,
+      "step": 40400
+    },
+    {
+      "epoch": 0.7306736943456867,
+      "grad_norm": 40.375,
+      "learning_rate": 9.88583242878865e-06,
+      "loss": 16.9922,
+      "step": 40410
+    },
+    {
+      "epoch": 0.7308545094148147,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.88580417648199e-06,
+      "loss": 16.7063,
+      "step": 40420
+    },
+    {
+      "epoch": 0.7310353244839425,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.88577592417533e-06,
+      "loss": 17.0254,
+      "step": 40430
+    },
+    {
+      "epoch": 0.7312161395530703,
+      "grad_norm": 42.5,
+      "learning_rate": 9.885747671868671e-06,
+      "loss": 16.7434,
+      "step": 40440
+    },
+    {
+      "epoch": 0.7313969546221982,
+      "grad_norm": 44.5,
+      "learning_rate": 9.885719419562012e-06,
+      "loss": 16.8146,
+      "step": 40450
+    },
+    {
+      "epoch": 0.731577769691326,
+      "grad_norm": 41.625,
+      "learning_rate": 9.88569116725535e-06,
+      "loss": 17.4253,
+      "step": 40460
+    },
+    {
+      "epoch": 0.731758584760454,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.885662914948691e-06,
+      "loss": 16.398,
+      "step": 40470
+    },
+    {
+      "epoch": 0.7319393998295818,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.885634662642032e-06,
+      "loss": 16.5052,
+      "step": 40480
+    },
+    {
+      "epoch": 0.7321202148987097,
+      "grad_norm": 42.75,
+      "learning_rate": 9.885606410335373e-06,
+      "loss": 16.7789,
+      "step": 40490
+    },
+    {
+      "epoch": 0.7323010299678375,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.885578158028713e-06,
+      "loss": 16.5454,
+      "step": 40500
+    },
+    {
+      "epoch": 0.7324818450369653,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.885549905722054e-06,
+      "loss": 16.8163,
+      "step": 40510
+    },
+    {
+      "epoch": 0.7326626601060933,
+      "grad_norm": 42.125,
+      "learning_rate": 9.885521653415394e-06,
+      "loss": 17.143,
+      "step": 40520
+    },
+    {
+      "epoch": 0.7328434751752211,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.885493401108735e-06,
+      "loss": 17.0652,
+      "step": 40530
+    },
+    {
+      "epoch": 0.733024290244349,
+      "grad_norm": 45.71875,
+      "learning_rate": 9.885465148802074e-06,
+      "loss": 17.1131,
+      "step": 40540
+    },
+    {
+      "epoch": 0.7332051053134768,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.885436896495415e-06,
+      "loss": 17.2092,
+      "step": 40550
+    },
+    {
+      "epoch": 0.7333859203826046,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.885408644188755e-06,
+      "loss": 17.1851,
+      "step": 40560
+    },
+    {
+      "epoch": 0.7335667354517326,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.885380391882096e-06,
+      "loss": 17.2057,
+      "step": 40570
+    },
+    {
+      "epoch": 0.7337475505208604,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.885352139575436e-06,
+      "loss": 16.8595,
+      "step": 40580
+    },
+    {
+      "epoch": 0.7339283655899883,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.885323887268777e-06,
+      "loss": 16.6233,
+      "step": 40590
+    },
+    {
+      "epoch": 0.7341091806591161,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.885295634962118e-06,
+      "loss": 16.6906,
+      "step": 40600
+    },
+    {
+      "epoch": 0.7342899957282439,
+      "grad_norm": 41.75,
+      "learning_rate": 9.885267382655458e-06,
+      "loss": 16.5386,
+      "step": 40610
+    },
+    {
+      "epoch": 0.7344708107973719,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.885239130348799e-06,
+      "loss": 16.8199,
+      "step": 40620
+    },
+    {
+      "epoch": 0.7346516258664997,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.885210878042138e-06,
+      "loss": 16.6419,
+      "step": 40630
+    },
+    {
+      "epoch": 0.7348324409356276,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.885182625735478e-06,
+      "loss": 16.9355,
+      "step": 40640
+    },
+    {
+      "epoch": 0.7350132560047554,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.885154373428819e-06,
+      "loss": 16.7165,
+      "step": 40650
+    },
+    {
+      "epoch": 0.7351940710738833,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.88512612112216e-06,
+      "loss": 16.981,
+      "step": 40660
+    },
+    {
+      "epoch": 0.7353748861430112,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.8850978688155e-06,
+      "loss": 16.8327,
+      "step": 40670
+    },
+    {
+      "epoch": 0.735555701212139,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.885069616508841e-06,
+      "loss": 16.9216,
+      "step": 40680
+    },
+    {
+      "epoch": 0.7357365162812669,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.885041364202182e-06,
+      "loss": 17.251,
+      "step": 40690
+    },
+    {
+      "epoch": 0.7359173313503947,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.885013111895522e-06,
+      "loss": 16.3896,
+      "step": 40700
+    },
+    {
+      "epoch": 0.7360981464195226,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.884984859588861e-06,
+      "loss": 16.8334,
+      "step": 40710
+    },
+    {
+      "epoch": 0.7362789614886505,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.884956607282202e-06,
+      "loss": 16.6666,
+      "step": 40720
+    },
+    {
+      "epoch": 0.7364597765577783,
+      "grad_norm": 45.0,
+      "learning_rate": 9.884928354975542e-06,
+      "loss": 17.068,
+      "step": 40730
+    },
+    {
+      "epoch": 0.7366405916269062,
+      "grad_norm": 41.5,
+      "learning_rate": 9.884900102668883e-06,
+      "loss": 16.5892,
+      "step": 40740
+    },
+    {
+      "epoch": 0.736821406696034,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.884871850362224e-06,
+      "loss": 16.977,
+      "step": 40750
+    },
+    {
+      "epoch": 0.737002221765162,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.884843598055564e-06,
+      "loss": 17.0296,
+      "step": 40760
+    },
+    {
+      "epoch": 0.7371830368342898,
+      "grad_norm": 41.75,
+      "learning_rate": 9.884815345748905e-06,
+      "loss": 16.9444,
+      "step": 40770
+    },
+    {
+      "epoch": 0.7373638519034176,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.884787093442246e-06,
+      "loss": 16.7186,
+      "step": 40780
+    },
+    {
+      "epoch": 0.7375446669725455,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.884758841135586e-06,
+      "loss": 16.5658,
+      "step": 40790
+    },
+    {
+      "epoch": 0.7377254820416733,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.884730588828925e-06,
+      "loss": 16.9532,
+      "step": 40800
+    },
+    {
+      "epoch": 0.7379062971108012,
+      "grad_norm": 43.5,
+      "learning_rate": 9.884702336522266e-06,
+      "loss": 16.7144,
+      "step": 40810
+    },
+    {
+      "epoch": 0.7380871121799291,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.884674084215606e-06,
+      "loss": 16.8067,
+      "step": 40820
+    },
+    {
+      "epoch": 0.738267927249057,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.884645831908947e-06,
+      "loss": 16.7855,
+      "step": 40830
+    },
+    {
+      "epoch": 0.7384487423181848,
+      "grad_norm": 40.875,
+      "learning_rate": 9.884617579602288e-06,
+      "loss": 16.8532,
+      "step": 40840
+    },
+    {
+      "epoch": 0.7386295573873126,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.884589327295628e-06,
+      "loss": 16.8848,
+      "step": 40850
+    },
+    {
+      "epoch": 0.7388103724564405,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.884561074988969e-06,
+      "loss": 16.41,
+      "step": 40860
+    },
+    {
+      "epoch": 0.7389911875255684,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.88453282268231e-06,
+      "loss": 16.931,
+      "step": 40870
+    },
+    {
+      "epoch": 0.7391720025946963,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.88450457037565e-06,
+      "loss": 16.4792,
+      "step": 40880
+    },
+    {
+      "epoch": 0.7393528176638241,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.884476318068989e-06,
+      "loss": 16.5623,
+      "step": 40890
+    },
+    {
+      "epoch": 0.7395336327329519,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.88444806576233e-06,
+      "loss": 16.9786,
+      "step": 40900
+    },
+    {
+      "epoch": 0.7397144478020798,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.88441981345567e-06,
+      "loss": 16.7459,
+      "step": 40910
+    },
+    {
+      "epoch": 0.7398952628712077,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.88439156114901e-06,
+      "loss": 17.09,
+      "step": 40920
+    },
+    {
+      "epoch": 0.7400760779403356,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.884363308842351e-06,
+      "loss": 17.1282,
+      "step": 40930
+    },
+    {
+      "epoch": 0.7402568930094634,
+      "grad_norm": 40.875,
+      "learning_rate": 9.884335056535692e-06,
+      "loss": 16.7603,
+      "step": 40940
+    },
+    {
+      "epoch": 0.7404377080785912,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.884306804229033e-06,
+      "loss": 16.8682,
+      "step": 40950
+    },
+    {
+      "epoch": 0.7406185231477191,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.884278551922373e-06,
+      "loss": 17.0364,
+      "step": 40960
+    },
+    {
+      "epoch": 0.740799338216847,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.884250299615712e-06,
+      "loss": 17.0327,
+      "step": 40970
+    },
+    {
+      "epoch": 0.7409801532859749,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.884222047309053e-06,
+      "loss": 16.9583,
+      "step": 40980
+    },
+    {
+      "epoch": 0.7411609683551027,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.884193795002394e-06,
+      "loss": 16.6937,
+      "step": 40990
+    },
+    {
+      "epoch": 0.7413417834242306,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.884165542695734e-06,
+      "loss": 16.5979,
+      "step": 41000
+    },
+    {
+      "epoch": 0.7415225984933584,
+      "grad_norm": 42.875,
+      "learning_rate": 9.884137290389075e-06,
+      "loss": 16.5445,
+      "step": 41010
+    },
+    {
+      "epoch": 0.7417034135624863,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.884109038082414e-06,
+      "loss": 16.9713,
+      "step": 41020
+    },
+    {
+      "epoch": 0.7418842286316142,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.884080785775756e-06,
+      "loss": 16.3581,
+      "step": 41030
+    },
+    {
+      "epoch": 0.742065043700742,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.884052533469097e-06,
+      "loss": 16.7363,
+      "step": 41040
+    },
+    {
+      "epoch": 0.7422458587698699,
+      "grad_norm": 39.625,
+      "learning_rate": 9.884024281162437e-06,
+      "loss": 16.7999,
+      "step": 41050
+    },
+    {
+      "epoch": 0.7424266738389977,
+      "grad_norm": 39.53125,
+      "learning_rate": 9.883996028855776e-06,
+      "loss": 16.8929,
+      "step": 41060
+    },
+    {
+      "epoch": 0.7426074889081256,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.883967776549117e-06,
+      "loss": 16.6393,
+      "step": 41070
+    },
+    {
+      "epoch": 0.7427883039772535,
+      "grad_norm": 43.375,
+      "learning_rate": 9.883939524242457e-06,
+      "loss": 16.7282,
+      "step": 41080
+    },
+    {
+      "epoch": 0.7429691190463813,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.883911271935798e-06,
+      "loss": 17.0336,
+      "step": 41090
+    },
+    {
+      "epoch": 0.7431499341155092,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.883883019629139e-06,
+      "loss": 16.9114,
+      "step": 41100
+    },
+    {
+      "epoch": 0.743330749184637,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.88385476732248e-06,
+      "loss": 17.1745,
+      "step": 41110
+    },
+    {
+      "epoch": 0.7435115642537649,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.88382651501582e-06,
+      "loss": 16.4395,
+      "step": 41120
+    },
+    {
+      "epoch": 0.7436923793228928,
+      "grad_norm": 44.25,
+      "learning_rate": 9.88379826270916e-06,
+      "loss": 16.8133,
+      "step": 41130
+    },
+    {
+      "epoch": 0.7438731943920206,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.8837700104025e-06,
+      "loss": 16.6626,
+      "step": 41140
+    },
+    {
+      "epoch": 0.7440540094611485,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.88374175809584e-06,
+      "loss": 16.7532,
+      "step": 41150
+    },
+    {
+      "epoch": 0.7442348245302763,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.88371350578918e-06,
+      "loss": 16.8214,
+      "step": 41160
+    },
+    {
+      "epoch": 0.7444156395994043,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.883685253482521e-06,
+      "loss": 16.7015,
+      "step": 41170
+    },
+    {
+      "epoch": 0.7445964546685321,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.883657001175862e-06,
+      "loss": 17.0512,
+      "step": 41180
+    },
+    {
+      "epoch": 0.7447772697376599,
+      "grad_norm": 44.5,
+      "learning_rate": 9.883628748869203e-06,
+      "loss": 17.1258,
+      "step": 41190
+    },
+    {
+      "epoch": 0.7449580848067878,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.883600496562543e-06,
+      "loss": 16.8723,
+      "step": 41200
+    },
+    {
+      "epoch": 0.7451388998759156,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.883572244255884e-06,
+      "loss": 16.7487,
+      "step": 41210
+    },
+    {
+      "epoch": 0.7453197149450436,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.883543991949224e-06,
+      "loss": 16.6667,
+      "step": 41220
+    },
+    {
+      "epoch": 0.7455005300141714,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.883515739642563e-06,
+      "loss": 16.8356,
+      "step": 41230
+    },
+    {
+      "epoch": 0.7456813450832992,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.883487487335904e-06,
+      "loss": 16.8372,
+      "step": 41240
+    },
+    {
+      "epoch": 0.7458621601524271,
+      "grad_norm": 44.84375,
+      "learning_rate": 9.883459235029245e-06,
+      "loss": 16.2635,
+      "step": 41250
+    },
+    {
+      "epoch": 0.7460429752215549,
+      "grad_norm": 44.375,
+      "learning_rate": 9.883430982722585e-06,
+      "loss": 16.6561,
+      "step": 41260
+    },
+    {
+      "epoch": 0.7462237902906829,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.883402730415926e-06,
+      "loss": 16.6502,
+      "step": 41270
+    },
+    {
+      "epoch": 0.7464046053598107,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.883374478109265e-06,
+      "loss": 16.8927,
+      "step": 41280
+    },
+    {
+      "epoch": 0.7465854204289385,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.883346225802607e-06,
+      "loss": 16.7709,
+      "step": 41290
+    },
+    {
+      "epoch": 0.7467662354980664,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.883317973495948e-06,
+      "loss": 16.8387,
+      "step": 41300
+    },
+    {
+      "epoch": 0.7469470505671942,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.883289721189288e-06,
+      "loss": 16.9989,
+      "step": 41310
+    },
+    {
+      "epoch": 0.7471278656363222,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.883261468882627e-06,
+      "loss": 16.9145,
+      "step": 41320
+    },
+    {
+      "epoch": 0.74730868070545,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.883233216575968e-06,
+      "loss": 16.7206,
+      "step": 41330
+    },
+    {
+      "epoch": 0.7474894957745779,
+      "grad_norm": 42.625,
+      "learning_rate": 9.883204964269309e-06,
+      "loss": 16.3599,
+      "step": 41340
+    },
+    {
+      "epoch": 0.7476703108437057,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.883176711962649e-06,
+      "loss": 16.4318,
+      "step": 41350
+    },
+    {
+      "epoch": 0.7478511259128335,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.88314845965599e-06,
+      "loss": 16.9316,
+      "step": 41360
+    },
+    {
+      "epoch": 0.7480319409819615,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.883120207349329e-06,
+      "loss": 16.7205,
+      "step": 41370
+    },
+    {
+      "epoch": 0.7482127560510893,
+      "grad_norm": 40.25,
+      "learning_rate": 9.883091955042671e-06,
+      "loss": 16.3273,
+      "step": 41380
+    },
+    {
+      "epoch": 0.7483935711202172,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.883063702736012e-06,
+      "loss": 17.1665,
+      "step": 41390
+    },
+    {
+      "epoch": 0.748574386189345,
+      "grad_norm": 43.59375,
+      "learning_rate": 9.88303545042935e-06,
+      "loss": 16.7555,
+      "step": 41400
+    },
+    {
+      "epoch": 0.7487552012584728,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.883007198122691e-06,
+      "loss": 16.8557,
+      "step": 41410
+    },
+    {
+      "epoch": 0.7489360163276008,
+      "grad_norm": 40.375,
+      "learning_rate": 9.882978945816032e-06,
+      "loss": 16.3214,
+      "step": 41420
+    },
+    {
+      "epoch": 0.7491168313967286,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.882950693509372e-06,
+      "loss": 17.0458,
+      "step": 41430
+    },
+    {
+      "epoch": 0.7492976464658565,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.882922441202713e-06,
+      "loss": 17.0682,
+      "step": 41440
+    },
+    {
+      "epoch": 0.7494784615349843,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.882894188896052e-06,
+      "loss": 16.79,
+      "step": 41450
+    },
+    {
+      "epoch": 0.7496592766041121,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.882865936589394e-06,
+      "loss": 16.8722,
+      "step": 41460
+    },
+    {
+      "epoch": 0.7498400916732401,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.882837684282735e-06,
+      "loss": 16.351,
+      "step": 41470
+    },
+    {
+      "epoch": 0.7500209067423679,
+      "grad_norm": 45.6875,
+      "learning_rate": 9.882809431976076e-06,
+      "loss": 16.8484,
+      "step": 41480
+    },
+    {
+      "epoch": 0.7502017218114958,
+      "grad_norm": 43.125,
+      "learning_rate": 9.882781179669414e-06,
+      "loss": 16.6952,
+      "step": 41490
+    },
+    {
+      "epoch": 0.7503825368806236,
+      "grad_norm": 46.5,
+      "learning_rate": 9.882752927362755e-06,
+      "loss": 16.8167,
+      "step": 41500
+    },
+    {
+      "epoch": 0.7505633519497515,
+      "grad_norm": 40.0,
+      "learning_rate": 9.882724675056096e-06,
+      "loss": 16.6396,
+      "step": 41510
+    },
+    {
+      "epoch": 0.7507441670188794,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.882696422749436e-06,
+      "loss": 16.5496,
+      "step": 41520
+    },
+    {
+      "epoch": 0.7509249820880072,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.882668170442777e-06,
+      "loss": 17.0983,
+      "step": 41530
+    },
+    {
+      "epoch": 0.7511057971571351,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.882639918136116e-06,
+      "loss": 16.601,
+      "step": 41540
+    },
+    {
+      "epoch": 0.7512866122262629,
+      "grad_norm": 41.75,
+      "learning_rate": 9.882611665829458e-06,
+      "loss": 16.7391,
+      "step": 41550
+    },
+    {
+      "epoch": 0.7514674272953908,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.882583413522799e-06,
+      "loss": 17.1427,
+      "step": 41560
+    },
+    {
+      "epoch": 0.7516482423645187,
+      "grad_norm": 42.25,
+      "learning_rate": 9.882555161216138e-06,
+      "loss": 16.784,
+      "step": 41570
+    },
+    {
+      "epoch": 0.7518290574336465,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.882526908909478e-06,
+      "loss": 16.6273,
+      "step": 41580
+    },
+    {
+      "epoch": 0.7520098725027744,
+      "grad_norm": 39.6875,
+      "learning_rate": 9.882498656602819e-06,
+      "loss": 16.5671,
+      "step": 41590
+    },
+    {
+      "epoch": 0.7521906875719022,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.88247040429616e-06,
+      "loss": 16.9705,
+      "step": 41600
+    },
+    {
+      "epoch": 0.7523715026410301,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.8824421519895e-06,
+      "loss": 16.6402,
+      "step": 41610
+    },
+    {
+      "epoch": 0.752552317710158,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.882413899682841e-06,
+      "loss": 16.7703,
+      "step": 41620
+    },
+    {
+      "epoch": 0.7527331327792858,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.88238564737618e-06,
+      "loss": 17.2249,
+      "step": 41630
+    },
+    {
+      "epoch": 0.7529139478484137,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.882357395069522e-06,
+      "loss": 16.6053,
+      "step": 41640
+    },
+    {
+      "epoch": 0.7530947629175415,
+      "grad_norm": 46.1875,
+      "learning_rate": 9.882329142762863e-06,
+      "loss": 17.1259,
+      "step": 41650
+    },
+    {
+      "epoch": 0.7532755779866694,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.882300890456202e-06,
+      "loss": 16.6679,
+      "step": 41660
+    },
+    {
+      "epoch": 0.7534563930557973,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.882272638149542e-06,
+      "loss": 17.3633,
+      "step": 41670
+    },
+    {
+      "epoch": 0.7536372081249252,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.882244385842883e-06,
+      "loss": 16.7871,
+      "step": 41680
+    },
+    {
+      "epoch": 0.753818023194053,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.882216133536224e-06,
+      "loss": 16.3583,
+      "step": 41690
+    },
+    {
+      "epoch": 0.7539988382631808,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.882187881229564e-06,
+      "loss": 16.692,
+      "step": 41700
+    },
+    {
+      "epoch": 0.7541796533323087,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.882159628922903e-06,
+      "loss": 16.7707,
+      "step": 41710
+    },
+    {
+      "epoch": 0.7543604684014366,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.882131376616244e-06,
+      "loss": 17.0561,
+      "step": 41720
+    },
+    {
+      "epoch": 0.7545412834705645,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.882103124309586e-06,
+      "loss": 17.1578,
+      "step": 41730
+    },
+    {
+      "epoch": 0.7547220985396923,
+      "grad_norm": 45.0,
+      "learning_rate": 9.882074872002927e-06,
+      "loss": 16.8074,
+      "step": 41740
+    },
+    {
+      "epoch": 0.7549029136088201,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.882046619696266e-06,
+      "loss": 16.7429,
+      "step": 41750
+    },
+    {
+      "epoch": 0.755083728677948,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.882018367389606e-06,
+      "loss": 16.8894,
+      "step": 41760
+    },
+    {
+      "epoch": 0.7552645437470759,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.881990115082947e-06,
+      "loss": 16.8988,
+      "step": 41770
+    },
+    {
+      "epoch": 0.7554453588162038,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.881961862776287e-06,
+      "loss": 16.8764,
+      "step": 41780
+    },
+    {
+      "epoch": 0.7556261738853316,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.881933610469628e-06,
+      "loss": 17.0994,
+      "step": 41790
+    },
+    {
+      "epoch": 0.7558069889544594,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.881905358162967e-06,
+      "loss": 16.8171,
+      "step": 41800
+    },
+    {
+      "epoch": 0.7559878040235873,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.88187710585631e-06,
+      "loss": 17.0823,
+      "step": 41810
+    },
+    {
+      "epoch": 0.7561686190927152,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.88184885354965e-06,
+      "loss": 16.7577,
+      "step": 41820
+    },
+    {
+      "epoch": 0.7563494341618431,
+      "grad_norm": 41.0,
+      "learning_rate": 9.881820601242989e-06,
+      "loss": 17.2433,
+      "step": 41830
+    },
+    {
+      "epoch": 0.7565302492309709,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.88179234893633e-06,
+      "loss": 17.117,
+      "step": 41840
+    },
+    {
+      "epoch": 0.7567110643000988,
+      "grad_norm": 42.625,
+      "learning_rate": 9.88176409662967e-06,
+      "loss": 16.6246,
+      "step": 41850
+    },
+    {
+      "epoch": 0.7568918793692266,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.88173584432301e-06,
+      "loss": 17.0378,
+      "step": 41860
+    },
+    {
+      "epoch": 0.7570726944383545,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.881707592016351e-06,
+      "loss": 16.559,
+      "step": 41870
+    },
+    {
+      "epoch": 0.7572535095074824,
+      "grad_norm": 46.1875,
+      "learning_rate": 9.88167933970969e-06,
+      "loss": 16.9933,
+      "step": 41880
+    },
+    {
+      "epoch": 0.7574343245766102,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.881651087403031e-06,
+      "loss": 16.5829,
+      "step": 41890
+    },
+    {
+      "epoch": 0.7576151396457381,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.881622835096373e-06,
+      "loss": 16.7736,
+      "step": 41900
+    },
+    {
+      "epoch": 0.7577959547148659,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.881594582789714e-06,
+      "loss": 16.9979,
+      "step": 41910
+    },
+    {
+      "epoch": 0.7579767697839938,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.881566330483053e-06,
+      "loss": 16.8586,
+      "step": 41920
+    },
+    {
+      "epoch": 0.7581575848531217,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.881538078176393e-06,
+      "loss": 16.416,
+      "step": 41930
+    },
+    {
+      "epoch": 0.7583383999222495,
+      "grad_norm": 43.375,
+      "learning_rate": 9.881509825869734e-06,
+      "loss": 17.0207,
+      "step": 41940
+    },
+    {
+      "epoch": 0.7585192149913774,
+      "grad_norm": 45.6875,
+      "learning_rate": 9.881481573563075e-06,
+      "loss": 16.8822,
+      "step": 41950
+    },
+    {
+      "epoch": 0.7587000300605052,
+      "grad_norm": 41.375,
+      "learning_rate": 9.881453321256415e-06,
+      "loss": 16.8448,
+      "step": 41960
+    },
+    {
+      "epoch": 0.758880845129633,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.881425068949754e-06,
+      "loss": 16.5718,
+      "step": 41970
+    },
+    {
+      "epoch": 0.759061660198761,
+      "grad_norm": 44.125,
+      "learning_rate": 9.881396816643095e-06,
+      "loss": 16.2509,
+      "step": 41980
+    },
+    {
+      "epoch": 0.7592424752678888,
+      "grad_norm": 43.5,
+      "learning_rate": 9.881368564336437e-06,
+      "loss": 16.4995,
+      "step": 41990
+    },
+    {
+      "epoch": 0.7594232903370167,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.881340312029776e-06,
+      "loss": 16.7124,
+      "step": 42000
+    },
+    {
+      "epoch": 0.7596041054061445,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.881312059723117e-06,
+      "loss": 16.6199,
+      "step": 42010
+    },
+    {
+      "epoch": 0.7597849204752725,
+      "grad_norm": 41.125,
+      "learning_rate": 9.881283807416457e-06,
+      "loss": 17.145,
+      "step": 42020
+    },
+    {
+      "epoch": 0.7599657355444003,
+      "grad_norm": 40.96875,
+      "learning_rate": 9.881255555109798e-06,
+      "loss": 16.6874,
+      "step": 42030
+    },
+    {
+      "epoch": 0.7601465506135281,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.881227302803139e-06,
+      "loss": 16.8522,
+      "step": 42040
+    },
+    {
+      "epoch": 0.760327365682656,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.881199050496479e-06,
+      "loss": 16.6918,
+      "step": 42050
+    },
+    {
+      "epoch": 0.7605081807517838,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.881170798189818e-06,
+      "loss": 16.6511,
+      "step": 42060
+    },
+    {
+      "epoch": 0.7606889958209118,
+      "grad_norm": 45.875,
+      "learning_rate": 9.881142545883159e-06,
+      "loss": 16.3789,
+      "step": 42070
+    },
+    {
+      "epoch": 0.7608698108900396,
+      "grad_norm": 45.34375,
+      "learning_rate": 9.881114293576501e-06,
+      "loss": 16.3309,
+      "step": 42080
+    },
+    {
+      "epoch": 0.7610506259591674,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.88108604126984e-06,
+      "loss": 17.0268,
+      "step": 42090
+    },
+    {
+      "epoch": 0.7612314410282953,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.88105778896318e-06,
+      "loss": 16.5562,
+      "step": 42100
+    },
+    {
+      "epoch": 0.7614122560974231,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.881029536656521e-06,
+      "loss": 16.671,
+      "step": 42110
+    },
+    {
+      "epoch": 0.7615930711665511,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.881001284349862e-06,
+      "loss": 16.8582,
+      "step": 42120
+    },
+    {
+      "epoch": 0.7617738862356789,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.880973032043202e-06,
+      "loss": 16.7876,
+      "step": 42130
+    },
+    {
+      "epoch": 0.7619547013048067,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.880944779736541e-06,
+      "loss": 16.9899,
+      "step": 42140
+    },
+    {
+      "epoch": 0.7621355163739346,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.880916527429882e-06,
+      "loss": 16.6254,
+      "step": 42150
+    },
+    {
+      "epoch": 0.7623163314430624,
+      "grad_norm": 45.5,
+      "learning_rate": 9.880888275123224e-06,
+      "loss": 16.5524,
+      "step": 42160
+    },
+    {
+      "epoch": 0.7624971465121904,
+      "grad_norm": 43.25,
+      "learning_rate": 9.880860022816565e-06,
+      "loss": 17.2899,
+      "step": 42170
+    },
+    {
+      "epoch": 0.7626779615813182,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.880831770509904e-06,
+      "loss": 16.8959,
+      "step": 42180
+    },
+    {
+      "epoch": 0.7628587766504461,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.880803518203244e-06,
+      "loss": 16.9736,
+      "step": 42190
+    },
+    {
+      "epoch": 0.7630395917195739,
+      "grad_norm": 46.625,
+      "learning_rate": 9.880775265896585e-06,
+      "loss": 17.254,
+      "step": 42200
+    },
+    {
+      "epoch": 0.7632204067887017,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.880747013589926e-06,
+      "loss": 16.6953,
+      "step": 42210
+    },
+    {
+      "epoch": 0.7634012218578297,
+      "grad_norm": 45.03125,
+      "learning_rate": 9.880718761283266e-06,
+      "loss": 16.8156,
+      "step": 42220
+    },
+    {
+      "epoch": 0.7635820369269575,
+      "grad_norm": 41.25,
+      "learning_rate": 9.880690508976605e-06,
+      "loss": 16.7833,
+      "step": 42230
+    },
+    {
+      "epoch": 0.7637628519960854,
+      "grad_norm": 45.25,
+      "learning_rate": 9.880662256669946e-06,
+      "loss": 17.0329,
+      "step": 42240
+    },
+    {
+      "epoch": 0.7639436670652132,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.880634004363288e-06,
+      "loss": 16.7803,
+      "step": 42250
+    },
+    {
+      "epoch": 0.764124482134341,
+      "grad_norm": 46.65625,
+      "learning_rate": 9.880605752056627e-06,
+      "loss": 16.7741,
+      "step": 42260
+    },
+    {
+      "epoch": 0.764305297203469,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.880577499749968e-06,
+      "loss": 16.4529,
+      "step": 42270
+    },
+    {
+      "epoch": 0.7644861122725968,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.880549247443308e-06,
+      "loss": 16.7497,
+      "step": 42280
+    },
+    {
+      "epoch": 0.7646669273417247,
+      "grad_norm": 40.59375,
+      "learning_rate": 9.880520995136649e-06,
+      "loss": 16.7111,
+      "step": 42290
+    },
+    {
+      "epoch": 0.7648477424108525,
+      "grad_norm": 43.75,
+      "learning_rate": 9.88049274282999e-06,
+      "loss": 16.9267,
+      "step": 42300
+    },
+    {
+      "epoch": 0.7650285574799803,
+      "grad_norm": 43.25,
+      "learning_rate": 9.880464490523329e-06,
+      "loss": 17.0684,
+      "step": 42310
+    },
+    {
+      "epoch": 0.7652093725491083,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.88043623821667e-06,
+      "loss": 16.6041,
+      "step": 42320
+    },
+    {
+      "epoch": 0.7653901876182361,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.88040798591001e-06,
+      "loss": 16.4228,
+      "step": 42330
+    },
+    {
+      "epoch": 0.765571002687364,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.880379733603352e-06,
+      "loss": 16.712,
+      "step": 42340
+    },
+    {
+      "epoch": 0.7657518177564918,
+      "grad_norm": 42.375,
+      "learning_rate": 9.880351481296691e-06,
+      "loss": 16.6611,
+      "step": 42350
+    },
+    {
+      "epoch": 0.7659326328256197,
+      "grad_norm": 43.0,
+      "learning_rate": 9.880323228990032e-06,
+      "loss": 16.5457,
+      "step": 42360
+    },
+    {
+      "epoch": 0.7661134478947476,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.880294976683372e-06,
+      "loss": 17.0666,
+      "step": 42370
+    },
+    {
+      "epoch": 0.7662942629638754,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.880266724376713e-06,
+      "loss": 16.7831,
+      "step": 42380
+    },
+    {
+      "epoch": 0.7664750780330033,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.880238472070054e-06,
+      "loss": 17.1136,
+      "step": 42390
+    },
+    {
+      "epoch": 0.7666558931021311,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.880210219763392e-06,
+      "loss": 16.8953,
+      "step": 42400
+    },
+    {
+      "epoch": 0.766836708171259,
+      "grad_norm": 41.25,
+      "learning_rate": 9.880181967456733e-06,
+      "loss": 16.8162,
+      "step": 42410
+    },
+    {
+      "epoch": 0.7670175232403869,
+      "grad_norm": 42.0,
+      "learning_rate": 9.880153715150074e-06,
+      "loss": 16.8969,
+      "step": 42420
+    },
+    {
+      "epoch": 0.7671983383095147,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.880125462843414e-06,
+      "loss": 16.7205,
+      "step": 42430
+    },
+    {
+      "epoch": 0.7673791533786426,
+      "grad_norm": 47.90625,
+      "learning_rate": 9.880097210536755e-06,
+      "loss": 16.8168,
+      "step": 42440
+    },
+    {
+      "epoch": 0.7675599684477704,
+      "grad_norm": 41.5,
+      "learning_rate": 9.880068958230096e-06,
+      "loss": 17.1686,
+      "step": 42450
+    },
+    {
+      "epoch": 0.7677407835168983,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.880040705923436e-06,
+      "loss": 16.6928,
+      "step": 42460
+    },
+    {
+      "epoch": 0.7679215985860262,
+      "grad_norm": 40.46875,
+      "learning_rate": 9.880012453616777e-06,
+      "loss": 17.0065,
+      "step": 42470
+    },
+    {
+      "epoch": 0.768102413655154,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.879984201310117e-06,
+      "loss": 16.5089,
+      "step": 42480
+    },
+    {
+      "epoch": 0.7682832287242819,
+      "grad_norm": 46.0,
+      "learning_rate": 9.879955949003456e-06,
+      "loss": 16.6789,
+      "step": 42490
+    },
+    {
+      "epoch": 0.7684640437934097,
+      "grad_norm": 43.75,
+      "learning_rate": 9.879927696696797e-06,
+      "loss": 17.1577,
+      "step": 42500
+    },
+    {
+      "epoch": 0.7686448588625376,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.87989944439014e-06,
+      "loss": 16.6613,
+      "step": 42510
+    },
+    {
+      "epoch": 0.7688256739316655,
+      "grad_norm": 42.625,
+      "learning_rate": 9.879871192083478e-06,
+      "loss": 16.8114,
+      "step": 42520
+    },
+    {
+      "epoch": 0.7690064890007934,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.879842939776819e-06,
+      "loss": 17.2868,
+      "step": 42530
+    },
+    {
+      "epoch": 0.7691873040699212,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.87981468747016e-06,
+      "loss": 16.7327,
+      "step": 42540
+    },
+    {
+      "epoch": 0.769368119139049,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.8797864351635e-06,
+      "loss": 16.9727,
+      "step": 42550
+    },
+    {
+      "epoch": 0.7695489342081769,
+      "grad_norm": 42.375,
+      "learning_rate": 9.87975818285684e-06,
+      "loss": 16.9383,
+      "step": 42560
+    },
+    {
+      "epoch": 0.7697297492773048,
+      "grad_norm": 43.375,
+      "learning_rate": 9.87972993055018e-06,
+      "loss": 16.681,
+      "step": 42570
+    },
+    {
+      "epoch": 0.7699105643464327,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.87970167824352e-06,
+      "loss": 17.1889,
+      "step": 42580
+    },
+    {
+      "epoch": 0.7700913794155605,
+      "grad_norm": 42.25,
+      "learning_rate": 9.879673425936861e-06,
+      "loss": 16.4792,
+      "step": 42590
+    },
+    {
+      "epoch": 0.7702721944846883,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.879645173630203e-06,
+      "loss": 16.7619,
+      "step": 42600
+    },
+    {
+      "epoch": 0.7704530095538162,
+      "grad_norm": 42.125,
+      "learning_rate": 9.879616921323542e-06,
+      "loss": 16.4408,
+      "step": 42610
+    },
+    {
+      "epoch": 0.770633824622944,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.879588669016883e-06,
+      "loss": 16.6894,
+      "step": 42620
+    },
+    {
+      "epoch": 0.770814639692072,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.879560416710223e-06,
+      "loss": 16.7997,
+      "step": 42630
+    },
+    {
+      "epoch": 0.7709954547611998,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.879532164403564e-06,
+      "loss": 17.1275,
+      "step": 42640
+    },
+    {
+      "epoch": 0.7711762698303276,
+      "grad_norm": 41.75,
+      "learning_rate": 9.879503912096905e-06,
+      "loss": 16.7961,
+      "step": 42650
+    },
+    {
+      "epoch": 0.7713570848994555,
+      "grad_norm": 45.59375,
+      "learning_rate": 9.879475659790244e-06,
+      "loss": 16.8425,
+      "step": 42660
+    },
+    {
+      "epoch": 0.7715378999685834,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.879447407483584e-06,
+      "loss": 16.9238,
+      "step": 42670
+    },
+    {
+      "epoch": 0.7717187150377113,
+      "grad_norm": 44.0,
+      "learning_rate": 9.879419155176925e-06,
+      "loss": 16.7722,
+      "step": 42680
+    },
+    {
+      "epoch": 0.7718995301068391,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.879390902870265e-06,
+      "loss": 16.8247,
+      "step": 42690
+    },
+    {
+      "epoch": 0.772080345175967,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.879362650563606e-06,
+      "loss": 16.495,
+      "step": 42700
+    },
+    {
+      "epoch": 0.7722611602450948,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.879334398256947e-06,
+      "loss": 16.5344,
+      "step": 42710
+    },
+    {
+      "epoch": 0.7724419753142227,
+      "grad_norm": 44.625,
+      "learning_rate": 9.879306145950287e-06,
+      "loss": 16.6817,
+      "step": 42720
+    },
+    {
+      "epoch": 0.7726227903833506,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.879277893643628e-06,
+      "loss": 16.883,
+      "step": 42730
+    },
+    {
+      "epoch": 0.7728036054524784,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.879249641336967e-06,
+      "loss": 16.7871,
+      "step": 42740
+    },
+    {
+      "epoch": 0.7729844205216063,
+      "grad_norm": 43.625,
+      "learning_rate": 9.879221389030307e-06,
+      "loss": 16.6971,
+      "step": 42750
+    },
+    {
+      "epoch": 0.7731652355907341,
+      "grad_norm": 44.53125,
+      "learning_rate": 9.879193136723648e-06,
+      "loss": 16.6874,
+      "step": 42760
+    },
+    {
+      "epoch": 0.773346050659862,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.879164884416989e-06,
+      "loss": 16.8721,
+      "step": 42770
+    },
+    {
+      "epoch": 0.7735268657289899,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.87913663211033e-06,
+      "loss": 16.7498,
+      "step": 42780
+    },
+    {
+      "epoch": 0.7737076807981177,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.87910837980367e-06,
+      "loss": 16.7373,
+      "step": 42790
+    },
+    {
+      "epoch": 0.7738884958672456,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.87908012749701e-06,
+      "loss": 16.5815,
+      "step": 42800
+    },
+    {
+      "epoch": 0.7740693109363734,
+      "grad_norm": 40.15625,
+      "learning_rate": 9.879051875190351e-06,
+      "loss": 16.9589,
+      "step": 42810
+    },
+    {
+      "epoch": 0.7742501260055012,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.879023622883692e-06,
+      "loss": 17.089,
+      "step": 42820
+    },
+    {
+      "epoch": 0.7744309410746292,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.87899537057703e-06,
+      "loss": 16.7203,
+      "step": 42830
+    },
+    {
+      "epoch": 0.774611756143757,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.878967118270371e-06,
+      "loss": 16.8916,
+      "step": 42840
+    },
+    {
+      "epoch": 0.7747925712128849,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.878938865963712e-06,
+      "loss": 16.8227,
+      "step": 42850
+    },
+    {
+      "epoch": 0.7749733862820127,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.878910613657053e-06,
+      "loss": 16.8217,
+      "step": 42860
+    },
+    {
+      "epoch": 0.7751542013511407,
+      "grad_norm": 42.375,
+      "learning_rate": 9.878882361350393e-06,
+      "loss": 16.9417,
+      "step": 42870
+    },
+    {
+      "epoch": 0.7753350164202685,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.878854109043734e-06,
+      "loss": 16.5462,
+      "step": 42880
+    },
+    {
+      "epoch": 0.7755158314893963,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.878825856737074e-06,
+      "loss": 16.7315,
+      "step": 42890
+    },
+    {
+      "epoch": 0.7756966465585242,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.878797604430415e-06,
+      "loss": 16.9322,
+      "step": 42900
+    },
+    {
+      "epoch": 0.775877461627652,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.878769352123756e-06,
+      "loss": 16.8821,
+      "step": 42910
+    },
+    {
+      "epoch": 0.77605827669678,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.878741099817095e-06,
+      "loss": 17.2593,
+      "step": 42920
+    },
+    {
+      "epoch": 0.7762390917659078,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.878712847510435e-06,
+      "loss": 16.8446,
+      "step": 42930
+    },
+    {
+      "epoch": 0.7764199068350356,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.878684595203776e-06,
+      "loss": 16.921,
+      "step": 42940
+    },
+    {
+      "epoch": 0.7766007219041635,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.878656342897117e-06,
+      "loss": 17.1003,
+      "step": 42950
+    },
+    {
+      "epoch": 0.7767815369732913,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.878628090590457e-06,
+      "loss": 16.8286,
+      "step": 42960
+    },
+    {
+      "epoch": 0.7769623520424193,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.878599838283798e-06,
+      "loss": 17.0732,
+      "step": 42970
+    },
+    {
+      "epoch": 0.7771431671115471,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.878571585977138e-06,
+      "loss": 16.7076,
+      "step": 42980
+    },
+    {
+      "epoch": 0.7773239821806749,
+      "grad_norm": 42.5,
+      "learning_rate": 9.878543333670479e-06,
+      "loss": 16.9669,
+      "step": 42990
+    },
+    {
+      "epoch": 0.7775047972498028,
+      "grad_norm": 46.78125,
+      "learning_rate": 9.878515081363818e-06,
+      "loss": 16.9687,
+      "step": 43000
+    },
+    {
+      "epoch": 0.7776856123189306,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.878486829057159e-06,
+      "loss": 16.8253,
+      "step": 43010
+    },
+    {
+      "epoch": 0.7778664273880586,
+      "grad_norm": 41.25,
+      "learning_rate": 9.8784585767505e-06,
+      "loss": 16.6976,
+      "step": 43020
+    },
+    {
+      "epoch": 0.7780472424571864,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.87843032444384e-06,
+      "loss": 17.3396,
+      "step": 43030
+    },
+    {
+      "epoch": 0.7782280575263142,
+      "grad_norm": 44.25,
+      "learning_rate": 9.87840207213718e-06,
+      "loss": 16.4274,
+      "step": 43040
+    },
+    {
+      "epoch": 0.7784088725954421,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.878373819830521e-06,
+      "loss": 16.6991,
+      "step": 43050
+    },
+    {
+      "epoch": 0.7785896876645699,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.878345567523862e-06,
+      "loss": 17.0445,
+      "step": 43060
+    },
+    {
+      "epoch": 0.7787705027336979,
+      "grad_norm": 39.5,
+      "learning_rate": 9.878317315217202e-06,
+      "loss": 16.6232,
+      "step": 43070
+    },
+    {
+      "epoch": 0.7789513178028257,
+      "grad_norm": 40.9375,
+      "learning_rate": 9.878289062910543e-06,
+      "loss": 16.6776,
+      "step": 43080
+    },
+    {
+      "epoch": 0.7791321328719536,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.878260810603882e-06,
+      "loss": 16.6483,
+      "step": 43090
+    },
+    {
+      "epoch": 0.7793129479410814,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.878232558297222e-06,
+      "loss": 17.0843,
+      "step": 43100
+    },
+    {
+      "epoch": 0.7794937630102092,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.878204305990563e-06,
+      "loss": 16.4456,
+      "step": 43110
+    },
+    {
+      "epoch": 0.7796745780793372,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.878176053683904e-06,
+      "loss": 16.9825,
+      "step": 43120
+    },
+    {
+      "epoch": 0.779855393148465,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.878147801377244e-06,
+      "loss": 17.033,
+      "step": 43130
+    },
+    {
+      "epoch": 0.7800362082175929,
+      "grad_norm": 42.0,
+      "learning_rate": 9.878119549070585e-06,
+      "loss": 16.6665,
+      "step": 43140
+    },
+    {
+      "epoch": 0.7802170232867207,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.878091296763926e-06,
+      "loss": 16.7691,
+      "step": 43150
+    },
+    {
+      "epoch": 0.7803978383558485,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.878063044457266e-06,
+      "loss": 16.9256,
+      "step": 43160
+    },
+    {
+      "epoch": 0.7805786534249765,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.878034792150605e-06,
+      "loss": 16.8972,
+      "step": 43170
+    },
+    {
+      "epoch": 0.7807594684941043,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.878006539843946e-06,
+      "loss": 16.2868,
+      "step": 43180
+    },
+    {
+      "epoch": 0.7809402835632322,
+      "grad_norm": 43.375,
+      "learning_rate": 9.877978287537286e-06,
+      "loss": 16.5691,
+      "step": 43190
+    },
+    {
+      "epoch": 0.78112109863236,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.877950035230627e-06,
+      "loss": 16.9558,
+      "step": 43200
+    },
+    {
+      "epoch": 0.7813019137014878,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.877921782923968e-06,
+      "loss": 16.5978,
+      "step": 43210
+    },
+    {
+      "epoch": 0.7814827287706158,
+      "grad_norm": 43.5,
+      "learning_rate": 9.877893530617308e-06,
+      "loss": 16.3195,
+      "step": 43220
+    },
+    {
+      "epoch": 0.7816635438397436,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.877865278310649e-06,
+      "loss": 16.8766,
+      "step": 43230
+    },
+    {
+      "epoch": 0.7818443589088715,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.87783702600399e-06,
+      "loss": 16.7194,
+      "step": 43240
+    },
+    {
+      "epoch": 0.7820251739779993,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.87780877369733e-06,
+      "loss": 16.5889,
+      "step": 43250
+    },
+    {
+      "epoch": 0.7822059890471272,
+      "grad_norm": 41.75,
+      "learning_rate": 9.877780521390669e-06,
+      "loss": 16.5907,
+      "step": 43260
+    },
+    {
+      "epoch": 0.782386804116255,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.87775226908401e-06,
+      "loss": 16.5148,
+      "step": 43270
+    },
+    {
+      "epoch": 0.7825676191853829,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.87772401677735e-06,
+      "loss": 17.0262,
+      "step": 43280
+    },
+    {
+      "epoch": 0.7827484342545108,
+      "grad_norm": 40.875,
+      "learning_rate": 9.877695764470691e-06,
+      "loss": 16.7053,
+      "step": 43290
+    },
+    {
+      "epoch": 0.7829292493236386,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.877667512164032e-06,
+      "loss": 16.8643,
+      "step": 43300
+    },
+    {
+      "epoch": 0.7831100643927665,
+      "grad_norm": 46.46875,
+      "learning_rate": 9.877639259857372e-06,
+      "loss": 16.8704,
+      "step": 43310
+    },
+    {
+      "epoch": 0.7832908794618944,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.877611007550713e-06,
+      "loss": 16.2131,
+      "step": 43320
+    },
+    {
+      "epoch": 0.7834716945310222,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.877582755244053e-06,
+      "loss": 17.0028,
+      "step": 43330
+    },
+    {
+      "epoch": 0.7836525096001501,
+      "grad_norm": 43.5,
+      "learning_rate": 9.877554502937392e-06,
+      "loss": 17.1055,
+      "step": 43340
+    },
+    {
+      "epoch": 0.7838333246692779,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.877526250630733e-06,
+      "loss": 16.8805,
+      "step": 43350
+    },
+    {
+      "epoch": 0.7840141397384058,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.877497998324074e-06,
+      "loss": 16.7048,
+      "step": 43360
+    },
+    {
+      "epoch": 0.7841949548075337,
+      "grad_norm": 40.875,
+      "learning_rate": 9.877469746017414e-06,
+      "loss": 16.7925,
+      "step": 43370
+    },
+    {
+      "epoch": 0.7843757698766615,
+      "grad_norm": 43.75,
+      "learning_rate": 9.877441493710755e-06,
+      "loss": 16.9311,
+      "step": 43380
+    },
+    {
+      "epoch": 0.7845565849457894,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.877413241404095e-06,
+      "loss": 16.5517,
+      "step": 43390
+    },
+    {
+      "epoch": 0.7847374000149172,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.877384989097436e-06,
+      "loss": 16.7184,
+      "step": 43400
+    },
+    {
+      "epoch": 0.7849182150840451,
+      "grad_norm": 43.625,
+      "learning_rate": 9.877356736790777e-06,
+      "loss": 16.433,
+      "step": 43410
+    },
+    {
+      "epoch": 0.785099030153173,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.877328484484117e-06,
+      "loss": 16.5557,
+      "step": 43420
+    },
+    {
+      "epoch": 0.7852798452223009,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.877300232177456e-06,
+      "loss": 17.1738,
+      "step": 43430
+    },
+    {
+      "epoch": 0.7854606602914287,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.877271979870797e-06,
+      "loss": 16.7636,
+      "step": 43440
+    },
+    {
+      "epoch": 0.7856414753605565,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.877243727564137e-06,
+      "loss": 16.6188,
+      "step": 43450
+    },
+    {
+      "epoch": 0.7858222904296844,
+      "grad_norm": 46.0,
+      "learning_rate": 9.877215475257478e-06,
+      "loss": 16.8037,
+      "step": 43460
+    },
+    {
+      "epoch": 0.7860031054988122,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.877187222950819e-06,
+      "loss": 16.84,
+      "step": 43470
+    },
+    {
+      "epoch": 0.7861839205679402,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.87715897064416e-06,
+      "loss": 16.7468,
+      "step": 43480
+    },
+    {
+      "epoch": 0.786364735637068,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.8771307183375e-06,
+      "loss": 16.8372,
+      "step": 43490
+    },
+    {
+      "epoch": 0.7865455507061958,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.87710246603084e-06,
+      "loss": 16.5974,
+      "step": 43500
+    },
+    {
+      "epoch": 0.7867263657753237,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.877074213724181e-06,
+      "loss": 16.6528,
+      "step": 43510
+    },
+    {
+      "epoch": 0.7869071808444515,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.87704596141752e-06,
+      "loss": 16.7555,
+      "step": 43520
+    },
+    {
+      "epoch": 0.7870879959135795,
+      "grad_norm": 41.5,
+      "learning_rate": 9.87701770911086e-06,
+      "loss": 17.2377,
+      "step": 43530
+    },
+    {
+      "epoch": 0.7872688109827073,
+      "grad_norm": 43.375,
+      "learning_rate": 9.876989456804201e-06,
+      "loss": 17.0131,
+      "step": 43540
+    },
+    {
+      "epoch": 0.7874496260518351,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.876961204497542e-06,
+      "loss": 16.7978,
+      "step": 43550
+    },
+    {
+      "epoch": 0.787630441120963,
+      "grad_norm": 44.0,
+      "learning_rate": 9.876932952190883e-06,
+      "loss": 17.0134,
+      "step": 43560
+    },
+    {
+      "epoch": 0.7878112561900908,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.876904699884223e-06,
+      "loss": 16.7749,
+      "step": 43570
+    },
+    {
+      "epoch": 0.7879920712592188,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.876876447577564e-06,
+      "loss": 16.8705,
+      "step": 43580
+    },
+    {
+      "epoch": 0.7881728863283466,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.876848195270904e-06,
+      "loss": 16.4949,
+      "step": 43590
+    },
+    {
+      "epoch": 0.7883537013974745,
+      "grad_norm": 46.65625,
+      "learning_rate": 9.876819942964243e-06,
+      "loss": 17.022,
+      "step": 43600
+    },
+    {
+      "epoch": 0.7885345164666023,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.876791690657584e-06,
+      "loss": 16.8378,
+      "step": 43610
+    },
+    {
+      "epoch": 0.7887153315357301,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.876763438350925e-06,
+      "loss": 16.9004,
+      "step": 43620
+    },
+    {
+      "epoch": 0.7888961466048581,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.876735186044265e-06,
+      "loss": 16.8231,
+      "step": 43630
+    },
+    {
+      "epoch": 0.7890769616739859,
+      "grad_norm": 44.375,
+      "learning_rate": 9.876706933737606e-06,
+      "loss": 16.7193,
+      "step": 43640
+    },
+    {
+      "epoch": 0.7892577767431138,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.876678681430947e-06,
+      "loss": 16.8568,
+      "step": 43650
+    },
+    {
+      "epoch": 0.7894385918122416,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.876650429124287e-06,
+      "loss": 16.5435,
+      "step": 43660
+    },
+    {
+      "epoch": 0.7896194068813694,
+      "grad_norm": 43.5,
+      "learning_rate": 9.876622176817628e-06,
+      "loss": 16.9701,
+      "step": 43670
+    },
+    {
+      "epoch": 0.7898002219504974,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.876593924510968e-06,
+      "loss": 16.5421,
+      "step": 43680
+    },
+    {
+      "epoch": 0.7899810370196252,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.876565672204307e-06,
+      "loss": 16.8361,
+      "step": 43690
+    },
+    {
+      "epoch": 0.7901618520887531,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.876537419897648e-06,
+      "loss": 17.0753,
+      "step": 43700
+    },
+    {
+      "epoch": 0.7903426671578809,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.876509167590989e-06,
+      "loss": 16.8627,
+      "step": 43710
+    },
+    {
+      "epoch": 0.7905234822270087,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.87648091528433e-06,
+      "loss": 16.7451,
+      "step": 43720
+    },
+    {
+      "epoch": 0.7907042972961367,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.87645266297767e-06,
+      "loss": 16.8212,
+      "step": 43730
+    },
+    {
+      "epoch": 0.7908851123652645,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.87642441067101e-06,
+      "loss": 16.9218,
+      "step": 43740
+    },
+    {
+      "epoch": 0.7910659274343924,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.876396158364351e-06,
+      "loss": 16.8543,
+      "step": 43750
+    },
+    {
+      "epoch": 0.7912467425035202,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.876367906057692e-06,
+      "loss": 16.5328,
+      "step": 43760
+    },
+    {
+      "epoch": 0.7914275575726482,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.87633965375103e-06,
+      "loss": 16.5861,
+      "step": 43770
+    },
+    {
+      "epoch": 0.791608372641776,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.876311401444371e-06,
+      "loss": 16.7647,
+      "step": 43780
+    },
+    {
+      "epoch": 0.7917891877109038,
+      "grad_norm": 41.5,
+      "learning_rate": 9.876283149137712e-06,
+      "loss": 16.6035,
+      "step": 43790
+    },
+    {
+      "epoch": 0.7919700027800317,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.876254896831052e-06,
+      "loss": 16.502,
+      "step": 43800
+    },
+    {
+      "epoch": 0.7921508178491595,
+      "grad_norm": 42.5,
+      "learning_rate": 9.876226644524393e-06,
+      "loss": 16.6553,
+      "step": 43810
+    },
+    {
+      "epoch": 0.7923316329182875,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.876198392217734e-06,
+      "loss": 16.2923,
+      "step": 43820
+    },
+    {
+      "epoch": 0.7925124479874153,
+      "grad_norm": 47.375,
+      "learning_rate": 9.876170139911074e-06,
+      "loss": 16.8679,
+      "step": 43830
+    },
+    {
+      "epoch": 0.7926932630565431,
+      "grad_norm": 46.5,
+      "learning_rate": 9.876141887604415e-06,
+      "loss": 17.1426,
+      "step": 43840
+    },
+    {
+      "epoch": 0.792874078125671,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.876113635297756e-06,
+      "loss": 16.7616,
+      "step": 43850
+    },
+    {
+      "epoch": 0.7930548931947988,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.876085382991095e-06,
+      "loss": 16.7291,
+      "step": 43860
+    },
+    {
+      "epoch": 0.7932357082639268,
+      "grad_norm": 41.75,
+      "learning_rate": 9.876057130684435e-06,
+      "loss": 16.9794,
+      "step": 43870
+    },
+    {
+      "epoch": 0.7934165233330546,
+      "grad_norm": 45.90625,
+      "learning_rate": 9.876028878377776e-06,
+      "loss": 16.7668,
+      "step": 43880
+    },
+    {
+      "epoch": 0.7935973384021824,
+      "grad_norm": 44.6875,
+      "learning_rate": 9.876000626071116e-06,
+      "loss": 17.1568,
+      "step": 43890
+    },
+    {
+      "epoch": 0.7937781534713103,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.875972373764457e-06,
+      "loss": 16.3285,
+      "step": 43900
+    },
+    {
+      "epoch": 0.7939589685404381,
+      "grad_norm": 45.6875,
+      "learning_rate": 9.875944121457796e-06,
+      "loss": 16.5321,
+      "step": 43910
+    },
+    {
+      "epoch": 0.794139783609566,
+      "grad_norm": 43.75,
+      "learning_rate": 9.875915869151138e-06,
+      "loss": 16.8059,
+      "step": 43920
+    },
+    {
+      "epoch": 0.7943205986786939,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.875887616844479e-06,
+      "loss": 16.7762,
+      "step": 43930
+    },
+    {
+      "epoch": 0.7945014137478218,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.87585936453782e-06,
+      "loss": 16.8121,
+      "step": 43940
+    },
+    {
+      "epoch": 0.7946822288169496,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.875831112231158e-06,
+      "loss": 16.7487,
+      "step": 43950
+    },
+    {
+      "epoch": 0.7948630438860774,
+      "grad_norm": 46.40625,
+      "learning_rate": 9.875802859924499e-06,
+      "loss": 16.681,
+      "step": 43960
+    },
+    {
+      "epoch": 0.7950438589552054,
+      "grad_norm": 44.375,
+      "learning_rate": 9.87577460761784e-06,
+      "loss": 16.903,
+      "step": 43970
+    },
+    {
+      "epoch": 0.7952246740243332,
+      "grad_norm": 46.59375,
+      "learning_rate": 9.87574635531118e-06,
+      "loss": 17.0115,
+      "step": 43980
+    },
+    {
+      "epoch": 0.7954054890934611,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.875718103004521e-06,
+      "loss": 16.7663,
+      "step": 43990
+    },
+    {
+      "epoch": 0.7955863041625889,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.875689850697862e-06,
+      "loss": 16.5117,
+      "step": 44000
+    },
+    {
+      "epoch": 0.7957671192317167,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.875661598391202e-06,
+      "loss": 16.799,
+      "step": 44010
+    },
+    {
+      "epoch": 0.7959479343008447,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.875633346084543e-06,
+      "loss": 16.8954,
+      "step": 44020
+    },
+    {
+      "epoch": 0.7961287493699725,
+      "grad_norm": 44.625,
+      "learning_rate": 9.875605093777882e-06,
+      "loss": 16.9782,
+      "step": 44030
+    },
+    {
+      "epoch": 0.7963095644391004,
+      "grad_norm": 42.75,
+      "learning_rate": 9.875576841471222e-06,
+      "loss": 16.9463,
+      "step": 44040
+    },
+    {
+      "epoch": 0.7964903795082282,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.875548589164563e-06,
+      "loss": 16.6091,
+      "step": 44050
+    },
+    {
+      "epoch": 0.796671194577356,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.875520336857904e-06,
+      "loss": 16.6663,
+      "step": 44060
+    },
+    {
+      "epoch": 0.796852009646484,
+      "grad_norm": 44.125,
+      "learning_rate": 9.875492084551244e-06,
+      "loss": 17.0986,
+      "step": 44070
+    },
+    {
+      "epoch": 0.7970328247156118,
+      "grad_norm": 41.46875,
+      "learning_rate": 9.875463832244583e-06,
+      "loss": 17.1912,
+      "step": 44080
+    },
+    {
+      "epoch": 0.7972136397847397,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.875435579937925e-06,
+      "loss": 17.1274,
+      "step": 44090
+    },
+    {
+      "epoch": 0.7973944548538675,
+      "grad_norm": 44.375,
+      "learning_rate": 9.875407327631266e-06,
+      "loss": 16.9897,
+      "step": 44100
+    },
+    {
+      "epoch": 0.7975752699229954,
+      "grad_norm": 42.125,
+      "learning_rate": 9.875379075324607e-06,
+      "loss": 16.8096,
+      "step": 44110
+    },
+    {
+      "epoch": 0.7977560849921232,
+      "grad_norm": 47.8125,
+      "learning_rate": 9.875350823017946e-06,
+      "loss": 17.1615,
+      "step": 44120
+    },
+    {
+      "epoch": 0.7979369000612511,
+      "grad_norm": 43.75,
+      "learning_rate": 9.875322570711286e-06,
+      "loss": 16.8355,
+      "step": 44130
+    },
+    {
+      "epoch": 0.798117715130379,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.875294318404627e-06,
+      "loss": 17.0292,
+      "step": 44140
+    },
+    {
+      "epoch": 0.7982985301995068,
+      "grad_norm": 43.5,
+      "learning_rate": 9.875266066097967e-06,
+      "loss": 16.3573,
+      "step": 44150
+    },
+    {
+      "epoch": 0.7984793452686347,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.875237813791308e-06,
+      "loss": 16.6646,
+      "step": 44160
+    },
+    {
+      "epoch": 0.7986601603377625,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.875209561484647e-06,
+      "loss": 16.396,
+      "step": 44170
+    },
+    {
+      "epoch": 0.7988409754068904,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.87518130917799e-06,
+      "loss": 16.947,
+      "step": 44180
+    },
+    {
+      "epoch": 0.7990217904760183,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.87515305687133e-06,
+      "loss": 16.8843,
+      "step": 44190
+    },
+    {
+      "epoch": 0.7992026055451461,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.875124804564669e-06,
+      "loss": 16.7936,
+      "step": 44200
+    },
+    {
+      "epoch": 0.799383420614274,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.87509655225801e-06,
+      "loss": 16.9738,
+      "step": 44210
+    },
+    {
+      "epoch": 0.7995642356834018,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.87506829995135e-06,
+      "loss": 17.1478,
+      "step": 44220
+    },
+    {
+      "epoch": 0.7997450507525297,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.87504004764469e-06,
+      "loss": 16.736,
+      "step": 44230
+    },
+    {
+      "epoch": 0.7999258658216576,
+      "grad_norm": 46.125,
+      "learning_rate": 9.875011795338031e-06,
+      "loss": 16.9612,
+      "step": 44240
+    },
+    {
+      "epoch": 0.8001066808907854,
+      "grad_norm": 44.625,
+      "learning_rate": 9.874983543031372e-06,
+      "loss": 17.1204,
+      "step": 44250
+    },
+    {
+      "epoch": 0.8002874959599133,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.874955290724711e-06,
+      "loss": 16.6089,
+      "step": 44260
+    },
+    {
+      "epoch": 0.8004683110290411,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.874927038418053e-06,
+      "loss": 16.6859,
+      "step": 44270
+    },
+    {
+      "epoch": 0.8006491260981691,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.874898786111394e-06,
+      "loss": 16.8503,
+      "step": 44280
+    },
+    {
+      "epoch": 0.8008299411672969,
+      "grad_norm": 43.75,
+      "learning_rate": 9.874870533804733e-06,
+      "loss": 16.7074,
+      "step": 44290
+    },
+    {
+      "epoch": 0.8010107562364247,
+      "grad_norm": 40.625,
+      "learning_rate": 9.874842281498073e-06,
+      "loss": 16.9856,
+      "step": 44300
+    },
+    {
+      "epoch": 0.8011915713055526,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.874814029191414e-06,
+      "loss": 17.1389,
+      "step": 44310
+    },
+    {
+      "epoch": 0.8013723863746804,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.874785776884755e-06,
+      "loss": 16.9076,
+      "step": 44320
+    },
+    {
+      "epoch": 0.8015532014438084,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.874757524578095e-06,
+      "loss": 16.6759,
+      "step": 44330
+    },
+    {
+      "epoch": 0.8017340165129362,
+      "grad_norm": 39.84375,
+      "learning_rate": 9.874729272271434e-06,
+      "loss": 16.5483,
+      "step": 44340
+    },
+    {
+      "epoch": 0.801914831582064,
+      "grad_norm": 43.125,
+      "learning_rate": 9.874701019964777e-06,
+      "loss": 16.9525,
+      "step": 44350
+    },
+    {
+      "epoch": 0.8020956466511919,
+      "grad_norm": 45.78125,
+      "learning_rate": 9.874672767658117e-06,
+      "loss": 16.641,
+      "step": 44360
+    },
+    {
+      "epoch": 0.8022764617203197,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.874644515351458e-06,
+      "loss": 16.7242,
+      "step": 44370
+    },
+    {
+      "epoch": 0.8024572767894477,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.874616263044797e-06,
+      "loss": 16.9407,
+      "step": 44380
+    },
+    {
+      "epoch": 0.8026380918585755,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.874588010738137e-06,
+      "loss": 17.0624,
+      "step": 44390
+    },
+    {
+      "epoch": 0.8028189069277033,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.874559758431478e-06,
+      "loss": 16.4757,
+      "step": 44400
+    },
+    {
+      "epoch": 0.8029997219968312,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.874531506124819e-06,
+      "loss": 17.0101,
+      "step": 44410
+    },
+    {
+      "epoch": 0.803180537065959,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.87450325381816e-06,
+      "loss": 16.7205,
+      "step": 44420
+    },
+    {
+      "epoch": 0.803361352135087,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.874475001511498e-06,
+      "loss": 17.1587,
+      "step": 44430
+    },
+    {
+      "epoch": 0.8035421672042148,
+      "grad_norm": 42.25,
+      "learning_rate": 9.87444674920484e-06,
+      "loss": 16.4366,
+      "step": 44440
+    },
+    {
+      "epoch": 0.8037229822733427,
+      "grad_norm": 46.9375,
+      "learning_rate": 9.874418496898181e-06,
+      "loss": 17.0698,
+      "step": 44450
+    },
+    {
+      "epoch": 0.8039037973424705,
+      "grad_norm": 42.5,
+      "learning_rate": 9.87439024459152e-06,
+      "loss": 16.8598,
+      "step": 44460
+    },
+    {
+      "epoch": 0.8040846124115983,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.87436199228486e-06,
+      "loss": 16.9069,
+      "step": 44470
+    },
+    {
+      "epoch": 0.8042654274807263,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.874333739978201e-06,
+      "loss": 16.4355,
+      "step": 44480
+    },
+    {
+      "epoch": 0.8044462425498541,
+      "grad_norm": 47.53125,
+      "learning_rate": 9.874305487671542e-06,
+      "loss": 16.8449,
+      "step": 44490
+    },
+    {
+      "epoch": 0.804627057618982,
+      "grad_norm": 42.75,
+      "learning_rate": 9.874277235364882e-06,
+      "loss": 17.022,
+      "step": 44500
+    },
+    {
+      "epoch": 0.8048078726881098,
+      "grad_norm": 41.875,
+      "learning_rate": 9.874248983058221e-06,
+      "loss": 17.0746,
+      "step": 44510
+    },
+    {
+      "epoch": 0.8049886877572376,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.874220730751562e-06,
+      "loss": 17.0565,
+      "step": 44520
+    },
+    {
+      "epoch": 0.8051695028263656,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.874192478444904e-06,
+      "loss": 16.7014,
+      "step": 44530
+    },
+    {
+      "epoch": 0.8053503178954934,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.874164226138245e-06,
+      "loss": 16.5769,
+      "step": 44540
+    },
+    {
+      "epoch": 0.8055311329646213,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.874135973831584e-06,
+      "loss": 16.7823,
+      "step": 44550
+    },
+    {
+      "epoch": 0.8057119480337491,
+      "grad_norm": 46.71875,
+      "learning_rate": 9.874107721524925e-06,
+      "loss": 16.4381,
+      "step": 44560
+    },
+    {
+      "epoch": 0.8058927631028769,
+      "grad_norm": 48.0625,
+      "learning_rate": 9.874079469218265e-06,
+      "loss": 16.2223,
+      "step": 44570
+    },
+    {
+      "epoch": 0.8060735781720049,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.874051216911606e-06,
+      "loss": 16.6863,
+      "step": 44580
+    },
+    {
+      "epoch": 0.8062543932411327,
+      "grad_norm": 41.875,
+      "learning_rate": 9.874022964604946e-06,
+      "loss": 16.6746,
+      "step": 44590
+    },
+    {
+      "epoch": 0.8064352083102606,
+      "grad_norm": 40.25,
+      "learning_rate": 9.873994712298285e-06,
+      "loss": 16.7305,
+      "step": 44600
+    },
+    {
+      "epoch": 0.8066160233793884,
+      "grad_norm": 47.125,
+      "learning_rate": 9.873966459991626e-06,
+      "loss": 16.9804,
+      "step": 44610
+    },
+    {
+      "epoch": 0.8067968384485164,
+      "grad_norm": 46.21875,
+      "learning_rate": 9.873938207684968e-06,
+      "loss": 16.7462,
+      "step": 44620
+    },
+    {
+      "epoch": 0.8069776535176442,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.873909955378307e-06,
+      "loss": 16.8084,
+      "step": 44630
+    },
+    {
+      "epoch": 0.807158468586772,
+      "grad_norm": 46.125,
+      "learning_rate": 9.873881703071648e-06,
+      "loss": 16.5409,
+      "step": 44640
+    },
+    {
+      "epoch": 0.8073392836558999,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.873853450764988e-06,
+      "loss": 16.5718,
+      "step": 44650
+    },
+    {
+      "epoch": 0.8075200987250277,
+      "grad_norm": 38.34375,
+      "learning_rate": 9.873825198458329e-06,
+      "loss": 16.6985,
+      "step": 44660
+    },
+    {
+      "epoch": 0.8077009137941557,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.87379694615167e-06,
+      "loss": 16.5292,
+      "step": 44670
+    },
+    {
+      "epoch": 0.8078817288632835,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.87376869384501e-06,
+      "loss": 16.7325,
+      "step": 44680
+    },
+    {
+      "epoch": 0.8080625439324113,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.87374044153835e-06,
+      "loss": 17.2133,
+      "step": 44690
+    },
+    {
+      "epoch": 0.8082433590015392,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.873712189231692e-06,
+      "loss": 16.6545,
+      "step": 44700
+    },
+    {
+      "epoch": 0.808424174070667,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.873683936925032e-06,
+      "loss": 16.7823,
+      "step": 44710
+    },
+    {
+      "epoch": 0.808604989139795,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.873655684618371e-06,
+      "loss": 16.7078,
+      "step": 44720
+    },
+    {
+      "epoch": 0.8087858042089228,
+      "grad_norm": 44.625,
+      "learning_rate": 9.873627432311712e-06,
+      "loss": 16.8342,
+      "step": 44730
+    },
+    {
+      "epoch": 0.8089666192780506,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.873599180005052e-06,
+      "loss": 17.0291,
+      "step": 44740
+    },
+    {
+      "epoch": 0.8091474343471785,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.873570927698393e-06,
+      "loss": 16.6447,
+      "step": 44750
+    },
+    {
+      "epoch": 0.8093282494163063,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.873542675391734e-06,
+      "loss": 17.0995,
+      "step": 44760
+    },
+    {
+      "epoch": 0.8095090644854342,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.873514423085073e-06,
+      "loss": 16.6301,
+      "step": 44770
+    },
+    {
+      "epoch": 0.8096898795545621,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.873486170778413e-06,
+      "loss": 17.1241,
+      "step": 44780
+    },
+    {
+      "epoch": 0.80987069462369,
+      "grad_norm": 45.625,
+      "learning_rate": 9.873457918471755e-06,
+      "loss": 16.5711,
+      "step": 44790
+    },
+    {
+      "epoch": 0.8100515096928178,
+      "grad_norm": 43.375,
+      "learning_rate": 9.873429666165096e-06,
+      "loss": 16.494,
+      "step": 44800
+    },
+    {
+      "epoch": 0.8102323247619456,
+      "grad_norm": 43.875,
+      "learning_rate": 9.873401413858435e-06,
+      "loss": 16.9783,
+      "step": 44810
+    },
+    {
+      "epoch": 0.8104131398310735,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.873373161551776e-06,
+      "loss": 16.9719,
+      "step": 44820
+    },
+    {
+      "epoch": 0.8105939549002014,
+      "grad_norm": 41.125,
+      "learning_rate": 9.873344909245116e-06,
+      "loss": 16.8303,
+      "step": 44830
+    },
+    {
+      "epoch": 0.8107747699693293,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.873316656938457e-06,
+      "loss": 16.6135,
+      "step": 44840
+    },
+    {
+      "epoch": 0.8109555850384571,
+      "grad_norm": 47.21875,
+      "learning_rate": 9.873288404631797e-06,
+      "loss": 17.2661,
+      "step": 44850
+    },
+    {
+      "epoch": 0.8111364001075849,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.873260152325136e-06,
+      "loss": 16.436,
+      "step": 44860
+    },
+    {
+      "epoch": 0.8113172151767128,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.873231900018477e-06,
+      "loss": 16.6702,
+      "step": 44870
+    },
+    {
+      "epoch": 0.8114980302458407,
+      "grad_norm": 42.25,
+      "learning_rate": 9.87320364771182e-06,
+      "loss": 16.866,
+      "step": 44880
+    },
+    {
+      "epoch": 0.8116788453149686,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.873175395405158e-06,
+      "loss": 16.8068,
+      "step": 44890
+    },
+    {
+      "epoch": 0.8118596603840964,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.873147143098499e-06,
+      "loss": 16.5403,
+      "step": 44900
+    },
+    {
+      "epoch": 0.8120404754532242,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.87311889079184e-06,
+      "loss": 17.0165,
+      "step": 44910
+    },
+    {
+      "epoch": 0.8122212905223521,
+      "grad_norm": 44.25,
+      "learning_rate": 9.87309063848518e-06,
+      "loss": 16.3663,
+      "step": 44920
+    },
+    {
+      "epoch": 0.81240210559148,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.87306238617852e-06,
+      "loss": 16.5132,
+      "step": 44930
+    },
+    {
+      "epoch": 0.8125829206606079,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.87303413387186e-06,
+      "loss": 16.7811,
+      "step": 44940
+    },
+    {
+      "epoch": 0.8127637357297357,
+      "grad_norm": 47.28125,
+      "learning_rate": 9.8730058815652e-06,
+      "loss": 17.3034,
+      "step": 44950
+    },
+    {
+      "epoch": 0.8129445507988636,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.872977629258541e-06,
+      "loss": 16.4767,
+      "step": 44960
+    },
+    {
+      "epoch": 0.8131253658679914,
+      "grad_norm": 42.125,
+      "learning_rate": 9.872949376951883e-06,
+      "loss": 17.1173,
+      "step": 44970
+    },
+    {
+      "epoch": 0.8133061809371193,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.872921124645222e-06,
+      "loss": 16.4008,
+      "step": 44980
+    },
+    {
+      "epoch": 0.8134869960062472,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.872892872338563e-06,
+      "loss": 16.4453,
+      "step": 44990
+    },
+    {
+      "epoch": 0.813667811075375,
+      "grad_norm": 45.5,
+      "learning_rate": 9.872864620031903e-06,
+      "loss": 16.954,
+      "step": 45000
+    },
+    {
+      "epoch": 0.813667811075375,
+      "eval_loss": 2.1011931896209717,
+      "eval_runtime": 229.8768,
+      "eval_samples_per_second": 3158.471,
+      "eval_steps_per_second": 49.353,
+      "step": 45000
+    },
+    {
+      "epoch": 0.8138486261445029,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.872836367725244e-06,
+      "loss": 16.1717,
+      "step": 45010
+    },
+    {
+      "epoch": 0.8140294412136307,
+      "grad_norm": 40.21875,
+      "learning_rate": 9.872808115418585e-06,
+      "loss": 16.7626,
+      "step": 45020
+    },
+    {
+      "epoch": 0.8142102562827586,
+      "grad_norm": 41.25,
+      "learning_rate": 9.872779863111924e-06,
+      "loss": 17.0021,
+      "step": 45030
+    },
+    {
+      "epoch": 0.8143910713518865,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.872751610805264e-06,
+      "loss": 16.884,
+      "step": 45040
+    },
+    {
+      "epoch": 0.8145718864210143,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.872723358498607e-06,
+      "loss": 16.9262,
+      "step": 45050
+    },
+    {
+      "epoch": 0.8147527014901422,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.872695106191945e-06,
+      "loss": 17.088,
+      "step": 45060
+    },
+    {
+      "epoch": 0.81493351655927,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.872666853885286e-06,
+      "loss": 16.5677,
+      "step": 45070
+    },
+    {
+      "epoch": 0.8151143316283979,
+      "grad_norm": 44.6875,
+      "learning_rate": 9.872638601578627e-06,
+      "loss": 16.9741,
+      "step": 45080
+    },
+    {
+      "epoch": 0.8152951466975258,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.872610349271967e-06,
+      "loss": 16.5187,
+      "step": 45090
+    },
+    {
+      "epoch": 0.8154759617666536,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.872582096965308e-06,
+      "loss": 16.83,
+      "step": 45100
+    },
+    {
+      "epoch": 0.8156567768357815,
+      "grad_norm": 45.0,
+      "learning_rate": 9.872553844658649e-06,
+      "loss": 16.7083,
+      "step": 45110
+    },
+    {
+      "epoch": 0.8158375919049093,
+      "grad_norm": 46.1875,
+      "learning_rate": 9.872525592351988e-06,
+      "loss": 17.0987,
+      "step": 45120
+    },
+    {
+      "epoch": 0.8160184069740373,
+      "grad_norm": 45.625,
+      "learning_rate": 9.872497340045328e-06,
+      "loss": 16.7546,
+      "step": 45130
+    },
+    {
+      "epoch": 0.8161992220431651,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.87246908773867e-06,
+      "loss": 17.1085,
+      "step": 45140
+    },
+    {
+      "epoch": 0.8163800371122929,
+      "grad_norm": 43.5,
+      "learning_rate": 9.87244083543201e-06,
+      "loss": 16.7036,
+      "step": 45150
+    },
+    {
+      "epoch": 0.8165608521814208,
+      "grad_norm": 45.78125,
+      "learning_rate": 9.87241258312535e-06,
+      "loss": 16.5432,
+      "step": 45160
+    },
+    {
+      "epoch": 0.8167416672505486,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.87238433081869e-06,
+      "loss": 16.7783,
+      "step": 45170
+    },
+    {
+      "epoch": 0.8169224823196766,
+      "grad_norm": 44.875,
+      "learning_rate": 9.872356078512031e-06,
+      "loss": 17.3186,
+      "step": 45180
+    },
+    {
+      "epoch": 0.8171032973888044,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.872327826205372e-06,
+      "loss": 17.2104,
+      "step": 45190
+    },
+    {
+      "epoch": 0.8172841124579322,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.87229957389871e-06,
+      "loss": 16.3376,
+      "step": 45200
+    },
+    {
+      "epoch": 0.8174649275270601,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.872271321592051e-06,
+      "loss": 16.9167,
+      "step": 45210
+    },
+    {
+      "epoch": 0.8176457425961879,
+      "grad_norm": 47.34375,
+      "learning_rate": 9.872243069285392e-06,
+      "loss": 17.1726,
+      "step": 45220
+    },
+    {
+      "epoch": 0.8178265576653159,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.872214816978734e-06,
+      "loss": 16.9629,
+      "step": 45230
+    },
+    {
+      "epoch": 0.8180073727344437,
+      "grad_norm": 46.375,
+      "learning_rate": 9.872186564672073e-06,
+      "loss": 16.6152,
+      "step": 45240
+    },
+    {
+      "epoch": 0.8181881878035715,
+      "grad_norm": 41.53125,
+      "learning_rate": 9.872158312365414e-06,
+      "loss": 16.5411,
+      "step": 45250
+    },
+    {
+      "epoch": 0.8183690028726994,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.872130060058755e-06,
+      "loss": 16.7691,
+      "step": 45260
+    },
+    {
+      "epoch": 0.8185498179418272,
+      "grad_norm": 43.5,
+      "learning_rate": 9.872101807752095e-06,
+      "loss": 16.5727,
+      "step": 45270
+    },
+    {
+      "epoch": 0.8187306330109552,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.872073555445436e-06,
+      "loss": 16.6411,
+      "step": 45280
+    },
+    {
+      "epoch": 0.818911448080083,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.872045303138775e-06,
+      "loss": 16.804,
+      "step": 45290
+    },
+    {
+      "epoch": 0.8190922631492109,
+      "grad_norm": 44.71875,
+      "learning_rate": 9.872017050832115e-06,
+      "loss": 16.286,
+      "step": 45300
+    },
+    {
+      "epoch": 0.8192730782183387,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.871988798525456e-06,
+      "loss": 16.8301,
+      "step": 45310
+    },
+    {
+      "epoch": 0.8194538932874665,
+      "grad_norm": 40.5625,
+      "learning_rate": 9.871960546218797e-06,
+      "loss": 16.8716,
+      "step": 45320
+    },
+    {
+      "epoch": 0.8196347083565945,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.871932293912137e-06,
+      "loss": 16.4862,
+      "step": 45330
+    },
+    {
+      "epoch": 0.8198155234257223,
+      "grad_norm": 46.46875,
+      "learning_rate": 9.871904041605478e-06,
+      "loss": 16.9789,
+      "step": 45340
+    },
+    {
+      "epoch": 0.8199963384948502,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.871875789298818e-06,
+      "loss": 16.7286,
+      "step": 45350
+    },
+    {
+      "epoch": 0.820177153563978,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.871847536992159e-06,
+      "loss": 16.843,
+      "step": 45360
+    },
+    {
+      "epoch": 0.8203579686331058,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.871819284685498e-06,
+      "loss": 16.6393,
+      "step": 45370
+    },
+    {
+      "epoch": 0.8205387837022338,
+      "grad_norm": 48.25,
+      "learning_rate": 9.871791032378839e-06,
+      "loss": 16.6034,
+      "step": 45380
+    },
+    {
+      "epoch": 0.8207195987713616,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.87176278007218e-06,
+      "loss": 17.0632,
+      "step": 45390
+    },
+    {
+      "epoch": 0.8209004138404895,
+      "grad_norm": 44.375,
+      "learning_rate": 9.871734527765522e-06,
+      "loss": 16.6226,
+      "step": 45400
+    },
+    {
+      "epoch": 0.8210812289096173,
+      "grad_norm": 45.71875,
+      "learning_rate": 9.87170627545886e-06,
+      "loss": 16.8551,
+      "step": 45410
+    },
+    {
+      "epoch": 0.8212620439787451,
+      "grad_norm": 43.25,
+      "learning_rate": 9.871678023152201e-06,
+      "loss": 17.2683,
+      "step": 45420
+    },
+    {
+      "epoch": 0.8214428590478731,
+      "grad_norm": 39.3125,
+      "learning_rate": 9.871649770845542e-06,
+      "loss": 17.3095,
+      "step": 45430
+    },
+    {
+      "epoch": 0.8216236741170009,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.871621518538882e-06,
+      "loss": 16.9099,
+      "step": 45440
+    },
+    {
+      "epoch": 0.8218044891861288,
+      "grad_norm": 42.25,
+      "learning_rate": 9.871593266232223e-06,
+      "loss": 16.8983,
+      "step": 45450
+    },
+    {
+      "epoch": 0.8219853042552566,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.871565013925562e-06,
+      "loss": 16.3534,
+      "step": 45460
+    },
+    {
+      "epoch": 0.8221661193243845,
+      "grad_norm": 42.5,
+      "learning_rate": 9.871536761618903e-06,
+      "loss": 16.9496,
+      "step": 45470
+    },
+    {
+      "epoch": 0.8223469343935124,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.871508509312243e-06,
+      "loss": 16.7329,
+      "step": 45480
+    },
+    {
+      "epoch": 0.8225277494626402,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.871480257005584e-06,
+      "loss": 16.4478,
+      "step": 45490
+    },
+    {
+      "epoch": 0.8227085645317681,
+      "grad_norm": 46.59375,
+      "learning_rate": 9.871452004698924e-06,
+      "loss": 16.8804,
+      "step": 45500
+    },
+    {
+      "epoch": 0.8228893796008959,
+      "grad_norm": 41.25,
+      "learning_rate": 9.871423752392265e-06,
+      "loss": 16.4803,
+      "step": 45510
+    },
+    {
+      "epoch": 0.8230701946700238,
+      "grad_norm": 42.375,
+      "learning_rate": 9.871395500085606e-06,
+      "loss": 17.0666,
+      "step": 45520
+    },
+    {
+      "epoch": 0.8232510097391517,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.871367247778946e-06,
+      "loss": 17.1151,
+      "step": 45530
+    },
+    {
+      "epoch": 0.8234318248082795,
+      "grad_norm": 41.0625,
+      "learning_rate": 9.871338995472285e-06,
+      "loss": 17.0039,
+      "step": 45540
+    },
+    {
+      "epoch": 0.8236126398774074,
+      "grad_norm": 45.0,
+      "learning_rate": 9.871310743165626e-06,
+      "loss": 16.9728,
+      "step": 45550
+    },
+    {
+      "epoch": 0.8237934549465352,
+      "grad_norm": 48.59375,
+      "learning_rate": 9.871282490858966e-06,
+      "loss": 17.0749,
+      "step": 45560
+    },
+    {
+      "epoch": 0.8239742700156631,
+      "grad_norm": 43.375,
+      "learning_rate": 9.871254238552307e-06,
+      "loss": 17.18,
+      "step": 45570
+    },
+    {
+      "epoch": 0.824155085084791,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.871225986245648e-06,
+      "loss": 17.0381,
+      "step": 45580
+    },
+    {
+      "epoch": 0.8243359001539188,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.871197733938988e-06,
+      "loss": 16.5253,
+      "step": 45590
+    },
+    {
+      "epoch": 0.8245167152230467,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.871169481632329e-06,
+      "loss": 16.8323,
+      "step": 45600
+    },
+    {
+      "epoch": 0.8246975302921745,
+      "grad_norm": 40.5,
+      "learning_rate": 9.87114122932567e-06,
+      "loss": 16.5359,
+      "step": 45610
+    },
+    {
+      "epoch": 0.8248783453613024,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.87111297701901e-06,
+      "loss": 16.7246,
+      "step": 45620
+    },
+    {
+      "epoch": 0.8250591604304303,
+      "grad_norm": 47.375,
+      "learning_rate": 9.871084724712349e-06,
+      "loss": 16.5078,
+      "step": 45630
+    },
+    {
+      "epoch": 0.8252399754995582,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.87105647240569e-06,
+      "loss": 16.6933,
+      "step": 45640
+    },
+    {
+      "epoch": 0.825420790568686,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.87102822009903e-06,
+      "loss": 16.3012,
+      "step": 45650
+    },
+    {
+      "epoch": 0.8256016056378138,
+      "grad_norm": 44.125,
+      "learning_rate": 9.870999967792371e-06,
+      "loss": 16.8273,
+      "step": 45660
+    },
+    {
+      "epoch": 0.8257824207069417,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.870971715485712e-06,
+      "loss": 17.0442,
+      "step": 45670
+    },
+    {
+      "epoch": 0.8259632357760696,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.870943463179052e-06,
+      "loss": 16.7705,
+      "step": 45680
+    },
+    {
+      "epoch": 0.8261440508451975,
+      "grad_norm": 42.0,
+      "learning_rate": 9.870915210872393e-06,
+      "loss": 16.8474,
+      "step": 45690
+    },
+    {
+      "epoch": 0.8263248659143253,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.870886958565733e-06,
+      "loss": 16.8813,
+      "step": 45700
+    },
+    {
+      "epoch": 0.8265056809834531,
+      "grad_norm": 44.375,
+      "learning_rate": 9.870858706259074e-06,
+      "loss": 16.8363,
+      "step": 45710
+    },
+    {
+      "epoch": 0.826686496052581,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.870830453952413e-06,
+      "loss": 16.4847,
+      "step": 45720
+    },
+    {
+      "epoch": 0.8268673111217089,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.870802201645754e-06,
+      "loss": 16.3199,
+      "step": 45730
+    },
+    {
+      "epoch": 0.8270481261908368,
+      "grad_norm": 40.875,
+      "learning_rate": 9.870773949339094e-06,
+      "loss": 17.1142,
+      "step": 45740
+    },
+    {
+      "epoch": 0.8272289412599646,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.870745697032435e-06,
+      "loss": 17.0783,
+      "step": 45750
+    },
+    {
+      "epoch": 0.8274097563290924,
+      "grad_norm": 45.25,
+      "learning_rate": 9.870717444725775e-06,
+      "loss": 16.8826,
+      "step": 45760
+    },
+    {
+      "epoch": 0.8275905713982203,
+      "grad_norm": 42.75,
+      "learning_rate": 9.870689192419116e-06,
+      "loss": 16.5429,
+      "step": 45770
+    },
+    {
+      "epoch": 0.8277713864673482,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.870660940112457e-06,
+      "loss": 16.6929,
+      "step": 45780
+    },
+    {
+      "epoch": 0.8279522015364761,
+      "grad_norm": 43.5,
+      "learning_rate": 9.870632687805797e-06,
+      "loss": 17.2222,
+      "step": 45790
+    },
+    {
+      "epoch": 0.8281330166056039,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.870604435499136e-06,
+      "loss": 17.0767,
+      "step": 45800
+    },
+    {
+      "epoch": 0.8283138316747318,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.870576183192477e-06,
+      "loss": 16.9483,
+      "step": 45810
+    },
+    {
+      "epoch": 0.8284946467438596,
+      "grad_norm": 43.25,
+      "learning_rate": 9.870547930885818e-06,
+      "loss": 17.0716,
+      "step": 45820
+    },
+    {
+      "epoch": 0.8286754618129875,
+      "grad_norm": 42.375,
+      "learning_rate": 9.870519678579158e-06,
+      "loss": 16.5698,
+      "step": 45830
+    },
+    {
+      "epoch": 0.8288562768821154,
+      "grad_norm": 40.625,
+      "learning_rate": 9.870491426272499e-06,
+      "loss": 16.5512,
+      "step": 45840
+    },
+    {
+      "epoch": 0.8290370919512432,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.87046317396584e-06,
+      "loss": 17.2864,
+      "step": 45850
+    },
+    {
+      "epoch": 0.8292179070203711,
+      "grad_norm": 43.125,
+      "learning_rate": 9.87043492165918e-06,
+      "loss": 17.0127,
+      "step": 45860
+    },
+    {
+      "epoch": 0.8293987220894989,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.87040666935252e-06,
+      "loss": 16.7168,
+      "step": 45870
+    },
+    {
+      "epoch": 0.8295795371586268,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.870378417045861e-06,
+      "loss": 16.454,
+      "step": 45880
+    },
+    {
+      "epoch": 0.8297603522277547,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.8703501647392e-06,
+      "loss": 16.5775,
+      "step": 45890
+    },
+    {
+      "epoch": 0.8299411672968825,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.87032191243254e-06,
+      "loss": 16.7222,
+      "step": 45900
+    },
+    {
+      "epoch": 0.8301219823660104,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.870293660125881e-06,
+      "loss": 16.7991,
+      "step": 45910
+    },
+    {
+      "epoch": 0.8303027974351382,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.870265407819222e-06,
+      "loss": 16.6125,
+      "step": 45920
+    },
+    {
+      "epoch": 0.830483612504266,
+      "grad_norm": 41.875,
+      "learning_rate": 9.870237155512563e-06,
+      "loss": 16.341,
+      "step": 45930
+    },
+    {
+      "epoch": 0.830664427573394,
+      "grad_norm": 42.25,
+      "learning_rate": 9.870208903205903e-06,
+      "loss": 17.3,
+      "step": 45940
+    },
+    {
+      "epoch": 0.8308452426425218,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.870180650899244e-06,
+      "loss": 16.0941,
+      "step": 45950
+    },
+    {
+      "epoch": 0.8310260577116497,
+      "grad_norm": 44.5,
+      "learning_rate": 9.870152398592585e-06,
+      "loss": 16.6848,
+      "step": 45960
+    },
+    {
+      "epoch": 0.8312068727807775,
+      "grad_norm": 42.0,
+      "learning_rate": 9.870124146285923e-06,
+      "loss": 16.6799,
+      "step": 45970
+    },
+    {
+      "epoch": 0.8313876878499055,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.870095893979264e-06,
+      "loss": 16.3178,
+      "step": 45980
+    },
+    {
+      "epoch": 0.8315685029190333,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.870067641672605e-06,
+      "loss": 16.6089,
+      "step": 45990
+    },
+    {
+      "epoch": 0.8317493179881611,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.870039389365945e-06,
+      "loss": 17.0249,
+      "step": 46000
+    },
+    {
+      "epoch": 0.831930133057289,
+      "grad_norm": 45.96875,
+      "learning_rate": 9.870011137059286e-06,
+      "loss": 16.8498,
+      "step": 46010
+    },
+    {
+      "epoch": 0.8321109481264168,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.869982884752627e-06,
+      "loss": 16.647,
+      "step": 46020
+    },
+    {
+      "epoch": 0.8322917631955448,
+      "grad_norm": 43.25,
+      "learning_rate": 9.869954632445967e-06,
+      "loss": 16.6773,
+      "step": 46030
+    },
+    {
+      "epoch": 0.8324725782646726,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.869926380139308e-06,
+      "loss": 16.4971,
+      "step": 46040
+    },
+    {
+      "epoch": 0.8326533933338004,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.869898127832648e-06,
+      "loss": 16.3478,
+      "step": 46050
+    },
+    {
+      "epoch": 0.8328342084029283,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.869869875525987e-06,
+      "loss": 16.8095,
+      "step": 46060
+    },
+    {
+      "epoch": 0.8330150234720561,
+      "grad_norm": 45.375,
+      "learning_rate": 9.869841623219328e-06,
+      "loss": 17.218,
+      "step": 46070
+    },
+    {
+      "epoch": 0.8331958385411841,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.869813370912669e-06,
+      "loss": 16.6927,
+      "step": 46080
+    },
+    {
+      "epoch": 0.8333766536103119,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.86978511860601e-06,
+      "loss": 16.999,
+      "step": 46090
+    },
+    {
+      "epoch": 0.8335574686794397,
+      "grad_norm": 47.84375,
+      "learning_rate": 9.86975686629935e-06,
+      "loss": 16.9362,
+      "step": 46100
+    },
+    {
+      "epoch": 0.8337382837485676,
+      "grad_norm": 47.09375,
+      "learning_rate": 9.86972861399269e-06,
+      "loss": 16.9018,
+      "step": 46110
+    },
+    {
+      "epoch": 0.8339190988176954,
+      "grad_norm": 49.65625,
+      "learning_rate": 9.869700361686031e-06,
+      "loss": 17.1598,
+      "step": 46120
+    },
+    {
+      "epoch": 0.8340999138868234,
+      "grad_norm": 44.375,
+      "learning_rate": 9.869672109379372e-06,
+      "loss": 16.3532,
+      "step": 46130
+    },
+    {
+      "epoch": 0.8342807289559512,
+      "grad_norm": 42.125,
+      "learning_rate": 9.869643857072712e-06,
+      "loss": 16.8197,
+      "step": 46140
+    },
+    {
+      "epoch": 0.8344615440250791,
+      "grad_norm": 43.375,
+      "learning_rate": 9.869615604766051e-06,
+      "loss": 16.5218,
+      "step": 46150
+    },
+    {
+      "epoch": 0.8346423590942069,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.869587352459392e-06,
+      "loss": 16.7873,
+      "step": 46160
+    },
+    {
+      "epoch": 0.8348231741633347,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.869559100152733e-06,
+      "loss": 16.6171,
+      "step": 46170
+    },
+    {
+      "epoch": 0.8350039892324627,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.869530847846073e-06,
+      "loss": 16.6384,
+      "step": 46180
+    },
+    {
+      "epoch": 0.8351848043015905,
+      "grad_norm": 42.0,
+      "learning_rate": 9.869502595539414e-06,
+      "loss": 16.6035,
+      "step": 46190
+    },
+    {
+      "epoch": 0.8353656193707184,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.869474343232754e-06,
+      "loss": 16.9676,
+      "step": 46200
+    },
+    {
+      "epoch": 0.8355464344398462,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.869446090926095e-06,
+      "loss": 16.6818,
+      "step": 46210
+    },
+    {
+      "epoch": 0.835727249508974,
+      "grad_norm": 41.6875,
+      "learning_rate": 9.869417838619436e-06,
+      "loss": 16.6355,
+      "step": 46220
+    },
+    {
+      "epoch": 0.835908064578102,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.869389586312775e-06,
+      "loss": 16.671,
+      "step": 46230
+    },
+    {
+      "epoch": 0.8360888796472298,
+      "grad_norm": 42.625,
+      "learning_rate": 9.869361334006115e-06,
+      "loss": 16.6258,
+      "step": 46240
+    },
+    {
+      "epoch": 0.8362696947163577,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.869333081699456e-06,
+      "loss": 16.8311,
+      "step": 46250
+    },
+    {
+      "epoch": 0.8364505097854855,
+      "grad_norm": 40.5,
+      "learning_rate": 9.869304829392796e-06,
+      "loss": 16.7952,
+      "step": 46260
+    },
+    {
+      "epoch": 0.8366313248546133,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.869276577086137e-06,
+      "loss": 16.7644,
+      "step": 46270
+    },
+    {
+      "epoch": 0.8368121399237413,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.869248324779478e-06,
+      "loss": 16.9682,
+      "step": 46280
+    },
+    {
+      "epoch": 0.8369929549928691,
+      "grad_norm": 44.375,
+      "learning_rate": 9.869220072472818e-06,
+      "loss": 16.6565,
+      "step": 46290
+    },
+    {
+      "epoch": 0.837173770061997,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.869191820166159e-06,
+      "loss": 16.4278,
+      "step": 46300
+    },
+    {
+      "epoch": 0.8373545851311248,
+      "grad_norm": 44.75,
+      "learning_rate": 9.8691635678595e-06,
+      "loss": 17.0361,
+      "step": 46310
+    },
+    {
+      "epoch": 0.8375354002002527,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.869135315552838e-06,
+      "loss": 16.7528,
+      "step": 46320
+    },
+    {
+      "epoch": 0.8377162152693806,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.869107063246179e-06,
+      "loss": 16.6941,
+      "step": 46330
+    },
+    {
+      "epoch": 0.8378970303385084,
+      "grad_norm": 41.625,
+      "learning_rate": 9.86907881093952e-06,
+      "loss": 16.5204,
+      "step": 46340
+    },
+    {
+      "epoch": 0.8380778454076363,
+      "grad_norm": 41.25,
+      "learning_rate": 9.86905055863286e-06,
+      "loss": 17.0408,
+      "step": 46350
+    },
+    {
+      "epoch": 0.8382586604767641,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.869022306326201e-06,
+      "loss": 16.3497,
+      "step": 46360
+    },
+    {
+      "epoch": 0.838439475545892,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.868994054019542e-06,
+      "loss": 16.5019,
+      "step": 46370
+    },
+    {
+      "epoch": 0.8386202906150199,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.868965801712882e-06,
+      "loss": 16.9828,
+      "step": 46380
+    },
+    {
+      "epoch": 0.8388011056841477,
+      "grad_norm": 45.5,
+      "learning_rate": 9.868937549406223e-06,
+      "loss": 16.5874,
+      "step": 46390
+    },
+    {
+      "epoch": 0.8389819207532756,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.868909297099562e-06,
+      "loss": 17.0737,
+      "step": 46400
+    },
+    {
+      "epoch": 0.8391627358224034,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.868881044792902e-06,
+      "loss": 17.0793,
+      "step": 46410
+    },
+    {
+      "epoch": 0.8393435508915313,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.868852792486243e-06,
+      "loss": 16.9732,
+      "step": 46420
+    },
+    {
+      "epoch": 0.8395243659606592,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.868824540179584e-06,
+      "loss": 17.0003,
+      "step": 46430
+    },
+    {
+      "epoch": 0.839705181029787,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.868796287872924e-06,
+      "loss": 16.6613,
+      "step": 46440
+    },
+    {
+      "epoch": 0.8398859960989149,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.868768035566265e-06,
+      "loss": 16.735,
+      "step": 46450
+    },
+    {
+      "epoch": 0.8400668111680427,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.868739783259605e-06,
+      "loss": 16.3271,
+      "step": 46460
+    },
+    {
+      "epoch": 0.8402476262371706,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.868711530952946e-06,
+      "loss": 16.5614,
+      "step": 46470
+    },
+    {
+      "epoch": 0.8404284413062985,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.868683278646287e-06,
+      "loss": 16.7678,
+      "step": 46480
+    },
+    {
+      "epoch": 0.8406092563754263,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.868655026339626e-06,
+      "loss": 16.8397,
+      "step": 46490
+    },
+    {
+      "epoch": 0.8407900714445542,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.868626774032966e-06,
+      "loss": 16.9057,
+      "step": 46500
+    },
+    {
+      "epoch": 0.840970886513682,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.868598521726307e-06,
+      "loss": 16.3798,
+      "step": 46510
+    },
+    {
+      "epoch": 0.8411517015828099,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.868570269419648e-06,
+      "loss": 16.7016,
+      "step": 46520
+    },
+    {
+      "epoch": 0.8413325166519378,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.868542017112988e-06,
+      "loss": 16.4862,
+      "step": 46530
+    },
+    {
+      "epoch": 0.8415133317210657,
+      "grad_norm": 45.59375,
+      "learning_rate": 9.868513764806329e-06,
+      "loss": 16.9339,
+      "step": 46540
+    },
+    {
+      "epoch": 0.8416941467901935,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.86848551249967e-06,
+      "loss": 17.1162,
+      "step": 46550
+    },
+    {
+      "epoch": 0.8418749618593213,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.86845726019301e-06,
+      "loss": 17.1667,
+      "step": 46560
+    },
+    {
+      "epoch": 0.8420557769284492,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.86842900788635e-06,
+      "loss": 16.5697,
+      "step": 46570
+    },
+    {
+      "epoch": 0.842236591997577,
+      "grad_norm": 41.125,
+      "learning_rate": 9.86840075557969e-06,
+      "loss": 16.86,
+      "step": 46580
+    },
+    {
+      "epoch": 0.842417407066705,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.86837250327303e-06,
+      "loss": 16.8571,
+      "step": 46590
+    },
+    {
+      "epoch": 0.8425982221358328,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.86834425096637e-06,
+      "loss": 17.0934,
+      "step": 46600
+    },
+    {
+      "epoch": 0.8427790372049606,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.868315998659711e-06,
+      "loss": 17.0367,
+      "step": 46610
+    },
+    {
+      "epoch": 0.8429598522740885,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.868287746353052e-06,
+      "loss": 16.933,
+      "step": 46620
+    },
+    {
+      "epoch": 0.8431406673432164,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.868259494046393e-06,
+      "loss": 17.0869,
+      "step": 46630
+    },
+    {
+      "epoch": 0.8433214824123443,
+      "grad_norm": 42.625,
+      "learning_rate": 9.868231241739733e-06,
+      "loss": 16.5878,
+      "step": 46640
+    },
+    {
+      "epoch": 0.8435022974814721,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.868202989433074e-06,
+      "loss": 16.6963,
+      "step": 46650
+    },
+    {
+      "epoch": 0.8436831125505999,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.868174737126413e-06,
+      "loss": 17.0261,
+      "step": 46660
+    },
+    {
+      "epoch": 0.8438639276197278,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.868146484819753e-06,
+      "loss": 16.5386,
+      "step": 46670
+    },
+    {
+      "epoch": 0.8440447426888557,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.868118232513094e-06,
+      "loss": 16.9873,
+      "step": 46680
+    },
+    {
+      "epoch": 0.8442255577579836,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.868089980206435e-06,
+      "loss": 16.8286,
+      "step": 46690
+    },
+    {
+      "epoch": 0.8444063728271114,
+      "grad_norm": 40.84375,
+      "learning_rate": 9.868061727899775e-06,
+      "loss": 16.9142,
+      "step": 46700
+    },
+    {
+      "epoch": 0.8445871878962393,
+      "grad_norm": 41.625,
+      "learning_rate": 9.868033475593114e-06,
+      "loss": 16.7071,
+      "step": 46710
+    },
+    {
+      "epoch": 0.8447680029653671,
+      "grad_norm": 41.03125,
+      "learning_rate": 9.868005223286457e-06,
+      "loss": 17.1076,
+      "step": 46720
+    },
+    {
+      "epoch": 0.844948818034495,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.867976970979797e-06,
+      "loss": 16.5466,
+      "step": 46730
+    },
+    {
+      "epoch": 0.8451296331036229,
+      "grad_norm": 44.875,
+      "learning_rate": 9.867948718673138e-06,
+      "loss": 17.1935,
+      "step": 46740
+    },
+    {
+      "epoch": 0.8453104481727507,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.867920466366477e-06,
+      "loss": 16.4253,
+      "step": 46750
+    },
+    {
+      "epoch": 0.8454912632418786,
+      "grad_norm": 47.0,
+      "learning_rate": 9.867892214059817e-06,
+      "loss": 16.9154,
+      "step": 46760
+    },
+    {
+      "epoch": 0.8456720783110064,
+      "grad_norm": 45.59375,
+      "learning_rate": 9.867863961753158e-06,
+      "loss": 16.6339,
+      "step": 46770
+    },
+    {
+      "epoch": 0.8458528933801343,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.867835709446499e-06,
+      "loss": 16.3904,
+      "step": 46780
+    },
+    {
+      "epoch": 0.8460337084492622,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.86780745713984e-06,
+      "loss": 16.8356,
+      "step": 46790
+    },
+    {
+      "epoch": 0.84621452351839,
+      "grad_norm": 45.6875,
+      "learning_rate": 9.867779204833178e-06,
+      "loss": 16.9262,
+      "step": 46800
+    },
+    {
+      "epoch": 0.8463953385875179,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.86775095252652e-06,
+      "loss": 17.0825,
+      "step": 46810
+    },
+    {
+      "epoch": 0.8465761536566457,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.867722700219861e-06,
+      "loss": 16.8047,
+      "step": 46820
+    },
+    {
+      "epoch": 0.8467569687257736,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.8676944479132e-06,
+      "loss": 16.5422,
+      "step": 46830
+    },
+    {
+      "epoch": 0.8469377837949015,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.86766619560654e-06,
+      "loss": 16.8578,
+      "step": 46840
+    },
+    {
+      "epoch": 0.8471185988640293,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.867637943299881e-06,
+      "loss": 16.5696,
+      "step": 46850
+    },
+    {
+      "epoch": 0.8472994139331572,
+      "grad_norm": 47.4375,
+      "learning_rate": 9.867609690993222e-06,
+      "loss": 16.7412,
+      "step": 46860
+    },
+    {
+      "epoch": 0.847480229002285,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.867581438686563e-06,
+      "loss": 16.9719,
+      "step": 46870
+    },
+    {
+      "epoch": 0.847661044071413,
+      "grad_norm": 43.75,
+      "learning_rate": 9.867553186379903e-06,
+      "loss": 17.0033,
+      "step": 46880
+    },
+    {
+      "epoch": 0.8478418591405408,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.867524934073244e-06,
+      "loss": 16.7468,
+      "step": 46890
+    },
+    {
+      "epoch": 0.8480226742096686,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.867496681766584e-06,
+      "loss": 16.8236,
+      "step": 46900
+    },
+    {
+      "epoch": 0.8482034892787965,
+      "grad_norm": 44.125,
+      "learning_rate": 9.867468429459925e-06,
+      "loss": 16.9039,
+      "step": 46910
+    },
+    {
+      "epoch": 0.8483843043479243,
+      "grad_norm": 42.0,
+      "learning_rate": 9.867440177153264e-06,
+      "loss": 16.8074,
+      "step": 46920
+    },
+    {
+      "epoch": 0.8485651194170523,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.867411924846605e-06,
+      "loss": 16.4367,
+      "step": 46930
+    },
+    {
+      "epoch": 0.8487459344861801,
+      "grad_norm": 45.90625,
+      "learning_rate": 9.867383672539945e-06,
+      "loss": 17.0261,
+      "step": 46940
+    },
+    {
+      "epoch": 0.8489267495553079,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.867355420233286e-06,
+      "loss": 16.4444,
+      "step": 46950
+    },
+    {
+      "epoch": 0.8491075646244358,
+      "grad_norm": 42.25,
+      "learning_rate": 9.867327167926626e-06,
+      "loss": 16.2641,
+      "step": 46960
+    },
+    {
+      "epoch": 0.8492883796935636,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.867298915619965e-06,
+      "loss": 16.3875,
+      "step": 46970
+    },
+    {
+      "epoch": 0.8494691947626916,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.867270663313308e-06,
+      "loss": 16.8093,
+      "step": 46980
+    },
+    {
+      "epoch": 0.8496500098318194,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.867242411006648e-06,
+      "loss": 16.3896,
+      "step": 46990
+    },
+    {
+      "epoch": 0.8498308249009472,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.867214158699989e-06,
+      "loss": 16.6276,
+      "step": 47000
+    },
+    {
+      "epoch": 0.8500116399700751,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.867185906393328e-06,
+      "loss": 16.7573,
+      "step": 47010
+    },
+    {
+      "epoch": 0.8501924550392029,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.867157654086668e-06,
+      "loss": 17.2952,
+      "step": 47020
+    },
+    {
+      "epoch": 0.8503732701083309,
+      "grad_norm": 42.25,
+      "learning_rate": 9.867129401780009e-06,
+      "loss": 17.0923,
+      "step": 47030
+    },
+    {
+      "epoch": 0.8505540851774587,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.86710114947335e-06,
+      "loss": 16.9049,
+      "step": 47040
+    },
+    {
+      "epoch": 0.8507349002465866,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.86707289716669e-06,
+      "loss": 16.667,
+      "step": 47050
+    },
+    {
+      "epoch": 0.8509157153157144,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.86704464486003e-06,
+      "loss": 16.6055,
+      "step": 47060
+    },
+    {
+      "epoch": 0.8510965303848422,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.867016392553372e-06,
+      "loss": 17.369,
+      "step": 47070
+    },
+    {
+      "epoch": 0.8512773454539702,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.866988140246712e-06,
+      "loss": 16.3215,
+      "step": 47080
+    },
+    {
+      "epoch": 0.851458160523098,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.866959887940051e-06,
+      "loss": 16.7921,
+      "step": 47090
+    },
+    {
+      "epoch": 0.8516389755922259,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.866931635633392e-06,
+      "loss": 16.9453,
+      "step": 47100
+    },
+    {
+      "epoch": 0.8518197906613537,
+      "grad_norm": 44.0,
+      "learning_rate": 9.866903383326732e-06,
+      "loss": 17.1338,
+      "step": 47110
+    },
+    {
+      "epoch": 0.8520006057304815,
+      "grad_norm": 40.0625,
+      "learning_rate": 9.866875131020073e-06,
+      "loss": 16.6403,
+      "step": 47120
+    },
+    {
+      "epoch": 0.8521814207996095,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.866846878713414e-06,
+      "loss": 16.5787,
+      "step": 47130
+    },
+    {
+      "epoch": 0.8523622358687373,
+      "grad_norm": 47.53125,
+      "learning_rate": 9.866818626406753e-06,
+      "loss": 16.6966,
+      "step": 47140
+    },
+    {
+      "epoch": 0.8525430509378652,
+      "grad_norm": 45.90625,
+      "learning_rate": 9.866790374100093e-06,
+      "loss": 16.2386,
+      "step": 47150
+    },
+    {
+      "epoch": 0.852723866006993,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.866762121793436e-06,
+      "loss": 16.7593,
+      "step": 47160
+    },
+    {
+      "epoch": 0.8529046810761208,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.866733869486776e-06,
+      "loss": 16.6689,
+      "step": 47170
+    },
+    {
+      "epoch": 0.8530854961452488,
+      "grad_norm": 46.3125,
+      "learning_rate": 9.866705617180115e-06,
+      "loss": 16.603,
+      "step": 47180
+    },
+    {
+      "epoch": 0.8532663112143766,
+      "grad_norm": 42.625,
+      "learning_rate": 9.866677364873456e-06,
+      "loss": 16.7621,
+      "step": 47190
+    },
+    {
+      "epoch": 0.8534471262835045,
+      "grad_norm": 43.25,
+      "learning_rate": 9.866649112566796e-06,
+      "loss": 16.6054,
+      "step": 47200
+    },
+    {
+      "epoch": 0.8536279413526323,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.866620860260137e-06,
+      "loss": 16.8067,
+      "step": 47210
+    },
+    {
+      "epoch": 0.8538087564217602,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.866592607953478e-06,
+      "loss": 16.8855,
+      "step": 47220
+    },
+    {
+      "epoch": 0.853989571490888,
+      "grad_norm": 41.1875,
+      "learning_rate": 9.866564355646816e-06,
+      "loss": 16.7962,
+      "step": 47230
+    },
+    {
+      "epoch": 0.8541703865600159,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.866536103340159e-06,
+      "loss": 16.797,
+      "step": 47240
+    },
+    {
+      "epoch": 0.8543512016291438,
+      "grad_norm": 46.96875,
+      "learning_rate": 9.8665078510335e-06,
+      "loss": 16.5223,
+      "step": 47250
+    },
+    {
+      "epoch": 0.8545320166982716,
+      "grad_norm": 45.8125,
+      "learning_rate": 9.866479598726838e-06,
+      "loss": 16.5319,
+      "step": 47260
+    },
+    {
+      "epoch": 0.8547128317673995,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.866451346420179e-06,
+      "loss": 16.6102,
+      "step": 47270
+    },
+    {
+      "epoch": 0.8548936468365274,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.86642309411352e-06,
+      "loss": 16.6636,
+      "step": 47280
+    },
+    {
+      "epoch": 0.8550744619056552,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.86639484180686e-06,
+      "loss": 16.6404,
+      "step": 47290
+    },
+    {
+      "epoch": 0.8552552769747831,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.8663665895002e-06,
+      "loss": 16.7613,
+      "step": 47300
+    },
+    {
+      "epoch": 0.8554360920439109,
+      "grad_norm": 41.75,
+      "learning_rate": 9.866338337193541e-06,
+      "loss": 17.157,
+      "step": 47310
+    },
+    {
+      "epoch": 0.8556169071130388,
+      "grad_norm": 46.71875,
+      "learning_rate": 9.86631008488688e-06,
+      "loss": 17.1103,
+      "step": 47320
+    },
+    {
+      "epoch": 0.8557977221821667,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.866281832580223e-06,
+      "loss": 16.653,
+      "step": 47330
+    },
+    {
+      "epoch": 0.8559785372512945,
+      "grad_norm": 43.75,
+      "learning_rate": 9.866253580273563e-06,
+      "loss": 17.0226,
+      "step": 47340
+    },
+    {
+      "epoch": 0.8561593523204224,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.866225327966902e-06,
+      "loss": 16.7507,
+      "step": 47350
+    },
+    {
+      "epoch": 0.8563401673895502,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.866197075660243e-06,
+      "loss": 16.1328,
+      "step": 47360
+    },
+    {
+      "epoch": 0.8565209824586781,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.866168823353583e-06,
+      "loss": 16.8316,
+      "step": 47370
+    },
+    {
+      "epoch": 0.856701797527806,
+      "grad_norm": 41.5,
+      "learning_rate": 9.866140571046924e-06,
+      "loss": 16.9856,
+      "step": 47380
+    },
+    {
+      "epoch": 0.8568826125969339,
+      "grad_norm": 45.125,
+      "learning_rate": 9.866112318740265e-06,
+      "loss": 17.0456,
+      "step": 47390
+    },
+    {
+      "epoch": 0.8570634276660617,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.866084066433604e-06,
+      "loss": 17.0106,
+      "step": 47400
+    },
+    {
+      "epoch": 0.8572442427351895,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.866055814126944e-06,
+      "loss": 16.3447,
+      "step": 47410
+    },
+    {
+      "epoch": 0.8574250578043174,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.866027561820287e-06,
+      "loss": 17.0897,
+      "step": 47420
+    },
+    {
+      "epoch": 0.8576058728734453,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.865999309513627e-06,
+      "loss": 16.8414,
+      "step": 47430
+    },
+    {
+      "epoch": 0.8577866879425732,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.865971057206966e-06,
+      "loss": 16.7726,
+      "step": 47440
+    },
+    {
+      "epoch": 0.857967503011701,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.865942804900307e-06,
+      "loss": 16.839,
+      "step": 47450
+    },
+    {
+      "epoch": 0.8581483180808288,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.865914552593647e-06,
+      "loss": 16.4288,
+      "step": 47460
+    },
+    {
+      "epoch": 0.8583291331499567,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.865886300286988e-06,
+      "loss": 16.6905,
+      "step": 47470
+    },
+    {
+      "epoch": 0.8585099482190846,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.865858047980329e-06,
+      "loss": 16.5618,
+      "step": 47480
+    },
+    {
+      "epoch": 0.8586907632882125,
+      "grad_norm": 46.125,
+      "learning_rate": 9.865829795673668e-06,
+      "loss": 16.7685,
+      "step": 47490
+    },
+    {
+      "epoch": 0.8588715783573403,
+      "grad_norm": 42.5,
+      "learning_rate": 9.865801543367008e-06,
+      "loss": 16.5453,
+      "step": 47500
+    },
+    {
+      "epoch": 0.8590523934264681,
+      "grad_norm": 40.78125,
+      "learning_rate": 9.86577329106035e-06,
+      "loss": 16.8157,
+      "step": 47510
+    },
+    {
+      "epoch": 0.859233208495596,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.86574503875369e-06,
+      "loss": 16.924,
+      "step": 47520
+    },
+    {
+      "epoch": 0.8594140235647239,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.86571678644703e-06,
+      "loss": 16.8491,
+      "step": 47530
+    },
+    {
+      "epoch": 0.8595948386338518,
+      "grad_norm": 42.875,
+      "learning_rate": 9.86568853414037e-06,
+      "loss": 16.7195,
+      "step": 47540
+    },
+    {
+      "epoch": 0.8597756537029796,
+      "grad_norm": 47.21875,
+      "learning_rate": 9.865660281833711e-06,
+      "loss": 16.9705,
+      "step": 47550
+    },
+    {
+      "epoch": 0.8599564687721075,
+      "grad_norm": 43.375,
+      "learning_rate": 9.865632029527052e-06,
+      "loss": 16.7851,
+      "step": 47560
+    },
+    {
+      "epoch": 0.8601372838412353,
+      "grad_norm": 43.625,
+      "learning_rate": 9.865603777220391e-06,
+      "loss": 17.0512,
+      "step": 47570
+    },
+    {
+      "epoch": 0.8603180989103631,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.865575524913731e-06,
+      "loss": 16.9528,
+      "step": 47580
+    },
+    {
+      "epoch": 0.8604989139794911,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.865547272607074e-06,
+      "loss": 16.8987,
+      "step": 47590
+    },
+    {
+      "epoch": 0.8606797290486189,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.865519020300414e-06,
+      "loss": 16.7751,
+      "step": 47600
+    },
+    {
+      "epoch": 0.8608605441177468,
+      "grad_norm": 45.34375,
+      "learning_rate": 9.865490767993753e-06,
+      "loss": 16.5027,
+      "step": 47610
+    },
+    {
+      "epoch": 0.8610413591868746,
+      "grad_norm": 45.4375,
+      "learning_rate": 9.865462515687094e-06,
+      "loss": 17.264,
+      "step": 47620
+    },
+    {
+      "epoch": 0.8612221742560024,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.865434263380435e-06,
+      "loss": 16.5418,
+      "step": 47630
+    },
+    {
+      "epoch": 0.8614029893251304,
+      "grad_norm": 44.875,
+      "learning_rate": 9.865406011073775e-06,
+      "loss": 16.6337,
+      "step": 47640
+    },
+    {
+      "epoch": 0.8615838043942582,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.865377758767116e-06,
+      "loss": 16.822,
+      "step": 47650
+    },
+    {
+      "epoch": 0.8617646194633861,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.865349506460455e-06,
+      "loss": 17.0387,
+      "step": 47660
+    },
+    {
+      "epoch": 0.8619454345325139,
+      "grad_norm": 40.4375,
+      "learning_rate": 9.865321254153795e-06,
+      "loss": 16.4247,
+      "step": 47670
+    },
+    {
+      "epoch": 0.8621262496016417,
+      "grad_norm": 40.40625,
+      "learning_rate": 9.865293001847138e-06,
+      "loss": 16.9979,
+      "step": 47680
+    },
+    {
+      "epoch": 0.8623070646707697,
+      "grad_norm": 42.25,
+      "learning_rate": 9.865264749540477e-06,
+      "loss": 16.5828,
+      "step": 47690
+    },
+    {
+      "epoch": 0.8624878797398975,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.865236497233817e-06,
+      "loss": 16.8847,
+      "step": 47700
+    },
+    {
+      "epoch": 0.8626686948090254,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.865208244927158e-06,
+      "loss": 16.9782,
+      "step": 47710
+    },
+    {
+      "epoch": 0.8628495098781532,
+      "grad_norm": 44.875,
+      "learning_rate": 9.865179992620498e-06,
+      "loss": 16.5684,
+      "step": 47720
+    },
+    {
+      "epoch": 0.8630303249472812,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.865151740313839e-06,
+      "loss": 16.2784,
+      "step": 47730
+    },
+    {
+      "epoch": 0.863211140016409,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.86512348800718e-06,
+      "loss": 17.0211,
+      "step": 47740
+    },
+    {
+      "epoch": 0.8633919550855368,
+      "grad_norm": 46.625,
+      "learning_rate": 9.865095235700519e-06,
+      "loss": 16.7793,
+      "step": 47750
+    },
+    {
+      "epoch": 0.8635727701546647,
+      "grad_norm": 38.5625,
+      "learning_rate": 9.86506698339386e-06,
+      "loss": 16.9268,
+      "step": 47760
+    },
+    {
+      "epoch": 0.8637535852237925,
+      "grad_norm": 39.9375,
+      "learning_rate": 9.865038731087202e-06,
+      "loss": 16.7704,
+      "step": 47770
+    },
+    {
+      "epoch": 0.8639344002929205,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.86501047878054e-06,
+      "loss": 16.9899,
+      "step": 47780
+    },
+    {
+      "epoch": 0.8641152153620483,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.864982226473881e-06,
+      "loss": 16.8435,
+      "step": 47790
+    },
+    {
+      "epoch": 0.8642960304311761,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.864953974167222e-06,
+      "loss": 16.5593,
+      "step": 47800
+    },
+    {
+      "epoch": 0.864476845500304,
+      "grad_norm": 46.1875,
+      "learning_rate": 9.864925721860562e-06,
+      "loss": 16.7038,
+      "step": 47810
+    },
+    {
+      "epoch": 0.8646576605694318,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.864897469553903e-06,
+      "loss": 16.8785,
+      "step": 47820
+    },
+    {
+      "epoch": 0.8648384756385598,
+      "grad_norm": 41.28125,
+      "learning_rate": 9.864869217247242e-06,
+      "loss": 16.9381,
+      "step": 47830
+    },
+    {
+      "epoch": 0.8650192907076876,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.864840964940583e-06,
+      "loss": 16.3128,
+      "step": 47840
+    },
+    {
+      "epoch": 0.8652001057768154,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.864812712633923e-06,
+      "loss": 16.5548,
+      "step": 47850
+    },
+    {
+      "epoch": 0.8653809208459433,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.864784460327266e-06,
+      "loss": 16.6328,
+      "step": 47860
+    },
+    {
+      "epoch": 0.8655617359150711,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.864756208020604e-06,
+      "loss": 16.7783,
+      "step": 47870
+    },
+    {
+      "epoch": 0.865742550984199,
+      "grad_norm": 43.375,
+      "learning_rate": 9.864727955713945e-06,
+      "loss": 16.5969,
+      "step": 47880
+    },
+    {
+      "epoch": 0.8659233660533269,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.864699703407286e-06,
+      "loss": 16.7332,
+      "step": 47890
+    },
+    {
+      "epoch": 0.8661041811224548,
+      "grad_norm": 44.25,
+      "learning_rate": 9.864671451100626e-06,
+      "loss": 16.8043,
+      "step": 47900
+    },
+    {
+      "epoch": 0.8662849961915826,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.864643198793967e-06,
+      "loss": 16.796,
+      "step": 47910
+    },
+    {
+      "epoch": 0.8664658112607104,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.864614946487306e-06,
+      "loss": 17.0849,
+      "step": 47920
+    },
+    {
+      "epoch": 0.8666466263298384,
+      "grad_norm": 42.625,
+      "learning_rate": 9.864586694180646e-06,
+      "loss": 16.6788,
+      "step": 47930
+    },
+    {
+      "epoch": 0.8668274413989662,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.864558441873989e-06,
+      "loss": 16.5263,
+      "step": 47940
+    },
+    {
+      "epoch": 0.8670082564680941,
+      "grad_norm": 45.59375,
+      "learning_rate": 9.864530189567328e-06,
+      "loss": 16.6813,
+      "step": 47950
+    },
+    {
+      "epoch": 0.8671890715372219,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.864501937260668e-06,
+      "loss": 16.7066,
+      "step": 47960
+    },
+    {
+      "epoch": 0.8673698866063497,
+      "grad_norm": 43.75,
+      "learning_rate": 9.864473684954009e-06,
+      "loss": 16.4457,
+      "step": 47970
+    },
+    {
+      "epoch": 0.8675507016754777,
+      "grad_norm": 44.6875,
+      "learning_rate": 9.86444543264735e-06,
+      "loss": 16.8559,
+      "step": 47980
+    },
+    {
+      "epoch": 0.8677315167446055,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.86441718034069e-06,
+      "loss": 16.6689,
+      "step": 47990
+    },
+    {
+      "epoch": 0.8679123318137334,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.864388928034029e-06,
+      "loss": 17.0778,
+      "step": 48000
+    },
+    {
+      "epoch": 0.8680931468828612,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.86436067572737e-06,
+      "loss": 16.6127,
+      "step": 48010
+    },
+    {
+      "epoch": 0.868273961951989,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.86433242342071e-06,
+      "loss": 16.7316,
+      "step": 48020
+    },
+    {
+      "epoch": 0.868454777021117,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.864304171114053e-06,
+      "loss": 17.0651,
+      "step": 48030
+    },
+    {
+      "epoch": 0.8686355920902448,
+      "grad_norm": 39.875,
+      "learning_rate": 9.864275918807392e-06,
+      "loss": 16.8044,
+      "step": 48040
+    },
+    {
+      "epoch": 0.8688164071593727,
+      "grad_norm": 46.4375,
+      "learning_rate": 9.864247666500732e-06,
+      "loss": 16.662,
+      "step": 48050
+    },
+    {
+      "epoch": 0.8689972222285005,
+      "grad_norm": 46.34375,
+      "learning_rate": 9.864219414194073e-06,
+      "loss": 16.8479,
+      "step": 48060
+    },
+    {
+      "epoch": 0.8691780372976284,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.864191161887413e-06,
+      "loss": 16.6773,
+      "step": 48070
+    },
+    {
+      "epoch": 0.8693588523667563,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.864162909580754e-06,
+      "loss": 16.3427,
+      "step": 48080
+    },
+    {
+      "epoch": 0.8695396674358841,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.864134657274093e-06,
+      "loss": 16.5875,
+      "step": 48090
+    },
+    {
+      "epoch": 0.869720482505012,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.864106404967434e-06,
+      "loss": 16.8144,
+      "step": 48100
+    },
+    {
+      "epoch": 0.8699012975741398,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.864078152660774e-06,
+      "loss": 16.7123,
+      "step": 48110
+    },
+    {
+      "epoch": 0.8700821126432677,
+      "grad_norm": 46.96875,
+      "learning_rate": 9.864049900354115e-06,
+      "loss": 16.5468,
+      "step": 48120
+    },
+    {
+      "epoch": 0.8702629277123956,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.864021648047456e-06,
+      "loss": 16.6123,
+      "step": 48130
+    },
+    {
+      "epoch": 0.8704437427815234,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.863993395740796e-06,
+      "loss": 16.6049,
+      "step": 48140
+    },
+    {
+      "epoch": 0.8706245578506513,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.863965143434137e-06,
+      "loss": 16.6403,
+      "step": 48150
+    },
+    {
+      "epoch": 0.8708053729197791,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.863936891127477e-06,
+      "loss": 16.3105,
+      "step": 48160
+    },
+    {
+      "epoch": 0.870986187988907,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.863908638820816e-06,
+      "loss": 16.4318,
+      "step": 48170
+    },
+    {
+      "epoch": 0.8711670030580349,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.863880386514157e-06,
+      "loss": 16.7314,
+      "step": 48180
+    },
+    {
+      "epoch": 0.8713478181271627,
+      "grad_norm": 45.25,
+      "learning_rate": 9.863852134207498e-06,
+      "loss": 17.3316,
+      "step": 48190
+    },
+    {
+      "epoch": 0.8715286331962906,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.863823881900838e-06,
+      "loss": 16.8433,
+      "step": 48200
+    },
+    {
+      "epoch": 0.8717094482654184,
+      "grad_norm": 42.75,
+      "learning_rate": 9.863795629594179e-06,
+      "loss": 16.7235,
+      "step": 48210
+    },
+    {
+      "epoch": 0.8718902633345463,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.86376737728752e-06,
+      "loss": 16.7464,
+      "step": 48220
+    },
+    {
+      "epoch": 0.8720710784036741,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.86373912498086e-06,
+      "loss": 16.5672,
+      "step": 48230
+    },
+    {
+      "epoch": 0.8722518934728021,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.8637108726742e-06,
+      "loss": 16.292,
+      "step": 48240
+    },
+    {
+      "epoch": 0.8724327085419299,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.863682620367541e-06,
+      "loss": 16.7786,
+      "step": 48250
+    },
+    {
+      "epoch": 0.8726135236110577,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.86365436806088e-06,
+      "loss": 16.6473,
+      "step": 48260
+    },
+    {
+      "epoch": 0.8727943386801856,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.863626115754221e-06,
+      "loss": 16.7557,
+      "step": 48270
+    },
+    {
+      "epoch": 0.8729751537493134,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.863597863447561e-06,
+      "loss": 17.0484,
+      "step": 48280
+    },
+    {
+      "epoch": 0.8731559688184414,
+      "grad_norm": 43.125,
+      "learning_rate": 9.863569611140902e-06,
+      "loss": 16.5089,
+      "step": 48290
+    },
+    {
+      "epoch": 0.8733367838875692,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.863541358834243e-06,
+      "loss": 16.8514,
+      "step": 48300
+    },
+    {
+      "epoch": 0.873517598956697,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.863513106527583e-06,
+      "loss": 16.7409,
+      "step": 48310
+    },
+    {
+      "epoch": 0.8736984140258249,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.863484854220924e-06,
+      "loss": 16.4951,
+      "step": 48320
+    },
+    {
+      "epoch": 0.8738792290949527,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.863456601914265e-06,
+      "loss": 17.0963,
+      "step": 48330
+    },
+    {
+      "epoch": 0.8740600441640807,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.863428349607605e-06,
+      "loss": 17.1169,
+      "step": 48340
+    },
+    {
+      "epoch": 0.8742408592332085,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.863400097300944e-06,
+      "loss": 16.532,
+      "step": 48350
+    },
+    {
+      "epoch": 0.8744216743023363,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.863371844994285e-06,
+      "loss": 16.4468,
+      "step": 48360
+    },
+    {
+      "epoch": 0.8746024893714642,
+      "grad_norm": 42.875,
+      "learning_rate": 9.863343592687625e-06,
+      "loss": 16.5803,
+      "step": 48370
+    },
+    {
+      "epoch": 0.874783304440592,
+      "grad_norm": 40.125,
+      "learning_rate": 9.863315340380966e-06,
+      "loss": 16.6159,
+      "step": 48380
+    },
+    {
+      "epoch": 0.87496411950972,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.863287088074307e-06,
+      "loss": 16.6872,
+      "step": 48390
+    },
+    {
+      "epoch": 0.8751449345788478,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.863258835767647e-06,
+      "loss": 16.4338,
+      "step": 48400
+    },
+    {
+      "epoch": 0.8753257496479757,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.863230583460988e-06,
+      "loss": 16.7224,
+      "step": 48410
+    },
+    {
+      "epoch": 0.8755065647171035,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.863202331154329e-06,
+      "loss": 16.6712,
+      "step": 48420
+    },
+    {
+      "epoch": 0.8756873797862313,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.863174078847667e-06,
+      "loss": 17.0588,
+      "step": 48430
+    },
+    {
+      "epoch": 0.8758681948553593,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.863145826541008e-06,
+      "loss": 16.8368,
+      "step": 48440
+    },
+    {
+      "epoch": 0.8760490099244871,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.863117574234349e-06,
+      "loss": 16.9934,
+      "step": 48450
+    },
+    {
+      "epoch": 0.876229824993615,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.86308932192769e-06,
+      "loss": 17.0267,
+      "step": 48460
+    },
+    {
+      "epoch": 0.8764106400627428,
+      "grad_norm": 46.5625,
+      "learning_rate": 9.86306106962103e-06,
+      "loss": 16.648,
+      "step": 48470
+    },
+    {
+      "epoch": 0.8765914551318706,
+      "grad_norm": 44.625,
+      "learning_rate": 9.86303281731437e-06,
+      "loss": 16.7814,
+      "step": 48480
+    },
+    {
+      "epoch": 0.8767722702009986,
+      "grad_norm": 43.25,
+      "learning_rate": 9.863004565007711e-06,
+      "loss": 16.8009,
+      "step": 48490
+    },
+    {
+      "epoch": 0.8769530852701264,
+      "grad_norm": 43.625,
+      "learning_rate": 9.862976312701052e-06,
+      "loss": 16.8888,
+      "step": 48500
+    },
+    {
+      "epoch": 0.8771339003392543,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.862948060394392e-06,
+      "loss": 16.9263,
+      "step": 48510
+    },
+    {
+      "epoch": 0.8773147154083821,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.862919808087731e-06,
+      "loss": 16.7748,
+      "step": 48520
+    },
+    {
+      "epoch": 0.8774955304775099,
+      "grad_norm": 46.65625,
+      "learning_rate": 9.862891555781072e-06,
+      "loss": 16.5806,
+      "step": 48530
+    },
+    {
+      "epoch": 0.8776763455466379,
+      "grad_norm": 41.375,
+      "learning_rate": 9.862863303474413e-06,
+      "loss": 16.629,
+      "step": 48540
+    },
+    {
+      "epoch": 0.8778571606157657,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.862835051167753e-06,
+      "loss": 17.0033,
+      "step": 48550
+    },
+    {
+      "epoch": 0.8780379756848936,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.862806798861094e-06,
+      "loss": 17.1855,
+      "step": 48560
+    },
+    {
+      "epoch": 0.8782187907540214,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.862778546554434e-06,
+      "loss": 16.6759,
+      "step": 48570
+    },
+    {
+      "epoch": 0.8783996058231494,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.862750294247775e-06,
+      "loss": 16.7433,
+      "step": 48580
+    },
+    {
+      "epoch": 0.8785804208922772,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.862722041941116e-06,
+      "loss": 16.637,
+      "step": 48590
+    },
+    {
+      "epoch": 0.878761235961405,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.862693789634455e-06,
+      "loss": 16.9434,
+      "step": 48600
+    },
+    {
+      "epoch": 0.8789420510305329,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.862665537327795e-06,
+      "loss": 16.7419,
+      "step": 48610
+    },
+    {
+      "epoch": 0.8791228660996607,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.862637285021136e-06,
+      "loss": 16.7121,
+      "step": 48620
+    },
+    {
+      "epoch": 0.8793036811687887,
+      "grad_norm": 41.40625,
+      "learning_rate": 9.862609032714476e-06,
+      "loss": 17.281,
+      "step": 48630
+    },
+    {
+      "epoch": 0.8794844962379165,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.862580780407817e-06,
+      "loss": 17.0728,
+      "step": 48640
+    },
+    {
+      "epoch": 0.8796653113070443,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.862552528101158e-06,
+      "loss": 17.1695,
+      "step": 48650
+    },
+    {
+      "epoch": 0.8798461263761722,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.862524275794498e-06,
+      "loss": 16.7729,
+      "step": 48660
+    },
+    {
+      "epoch": 0.8800269414453,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.862496023487839e-06,
+      "loss": 16.5975,
+      "step": 48670
+    },
+    {
+      "epoch": 0.880207756514428,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.86246777118118e-06,
+      "loss": 17.0676,
+      "step": 48680
+    },
+    {
+      "epoch": 0.8803885715835558,
+      "grad_norm": 43.625,
+      "learning_rate": 9.862439518874519e-06,
+      "loss": 16.4909,
+      "step": 48690
+    },
+    {
+      "epoch": 0.8805693866526836,
+      "grad_norm": 45.96875,
+      "learning_rate": 9.86241126656786e-06,
+      "loss": 16.2651,
+      "step": 48700
+    },
+    {
+      "epoch": 0.8807502017218115,
+      "grad_norm": 43.59375,
+      "learning_rate": 9.8623830142612e-06,
+      "loss": 16.8944,
+      "step": 48710
+    },
+    {
+      "epoch": 0.8809310167909393,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.86235476195454e-06,
+      "loss": 16.936,
+      "step": 48720
+    },
+    {
+      "epoch": 0.8811118318600673,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.862326509647881e-06,
+      "loss": 16.6326,
+      "step": 48730
+    },
+    {
+      "epoch": 0.8812926469291951,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.862298257341222e-06,
+      "loss": 16.9422,
+      "step": 48740
+    },
+    {
+      "epoch": 0.881473461998323,
+      "grad_norm": 41.375,
+      "learning_rate": 9.862270005034562e-06,
+      "loss": 16.3952,
+      "step": 48750
+    },
+    {
+      "epoch": 0.8816542770674508,
+      "grad_norm": 43.0,
+      "learning_rate": 9.862241752727903e-06,
+      "loss": 16.612,
+      "step": 48760
+    },
+    {
+      "epoch": 0.8818350921365786,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.862213500421244e-06,
+      "loss": 17.0378,
+      "step": 48770
+    },
+    {
+      "epoch": 0.8820159072057066,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.862185248114582e-06,
+      "loss": 16.5489,
+      "step": 48780
+    },
+    {
+      "epoch": 0.8821967222748344,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.862156995807923e-06,
+      "loss": 16.7768,
+      "step": 48790
+    },
+    {
+      "epoch": 0.8823775373439623,
+      "grad_norm": 47.0,
+      "learning_rate": 9.862128743501264e-06,
+      "loss": 16.9206,
+      "step": 48800
+    },
+    {
+      "epoch": 0.8825583524130901,
+      "grad_norm": 39.8125,
+      "learning_rate": 9.862100491194604e-06,
+      "loss": 16.4764,
+      "step": 48810
+    },
+    {
+      "epoch": 0.8827391674822179,
+      "grad_norm": 41.875,
+      "learning_rate": 9.862072238887945e-06,
+      "loss": 16.5945,
+      "step": 48820
+    },
+    {
+      "epoch": 0.8829199825513458,
+      "grad_norm": 45.625,
+      "learning_rate": 9.862043986581286e-06,
+      "loss": 16.5602,
+      "step": 48830
+    },
+    {
+      "epoch": 0.8831007976204737,
+      "grad_norm": 42.0,
+      "learning_rate": 9.862015734274626e-06,
+      "loss": 16.9655,
+      "step": 48840
+    },
+    {
+      "epoch": 0.8832816126896016,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.861987481967967e-06,
+      "loss": 16.6877,
+      "step": 48850
+    },
+    {
+      "epoch": 0.8834624277587294,
+      "grad_norm": 43.375,
+      "learning_rate": 9.861959229661306e-06,
+      "loss": 16.6978,
+      "step": 48860
+    },
+    {
+      "epoch": 0.8836432428278572,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.861930977354646e-06,
+      "loss": 16.9109,
+      "step": 48870
+    },
+    {
+      "epoch": 0.8838240578969851,
+      "grad_norm": 46.0625,
+      "learning_rate": 9.861902725047987e-06,
+      "loss": 16.019,
+      "step": 48880
+    },
+    {
+      "epoch": 0.884004872966113,
+      "grad_norm": 45.90625,
+      "learning_rate": 9.861874472741328e-06,
+      "loss": 17.0793,
+      "step": 48890
+    },
+    {
+      "epoch": 0.8841856880352409,
+      "grad_norm": 44.84375,
+      "learning_rate": 9.861846220434668e-06,
+      "loss": 16.832,
+      "step": 48900
+    },
+    {
+      "epoch": 0.8843665031043687,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.861817968128009e-06,
+      "loss": 16.9307,
+      "step": 48910
+    },
+    {
+      "epoch": 0.8845473181734966,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.86178971582135e-06,
+      "loss": 16.1817,
+      "step": 48920
+    },
+    {
+      "epoch": 0.8847281332426244,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.86176146351469e-06,
+      "loss": 17.1689,
+      "step": 48930
+    },
+    {
+      "epoch": 0.8849089483117523,
+      "grad_norm": 42.375,
+      "learning_rate": 9.86173321120803e-06,
+      "loss": 16.9108,
+      "step": 48940
+    },
+    {
+      "epoch": 0.8850897633808802,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.86170495890137e-06,
+      "loss": 16.6475,
+      "step": 48950
+    },
+    {
+      "epoch": 0.885270578450008,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.86167670659471e-06,
+      "loss": 16.5359,
+      "step": 48960
+    },
+    {
+      "epoch": 0.8854513935191359,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.861648454288051e-06,
+      "loss": 16.8834,
+      "step": 48970
+    },
+    {
+      "epoch": 0.8856322085882637,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.861620201981391e-06,
+      "loss": 16.5986,
+      "step": 48980
+    },
+    {
+      "epoch": 0.8858130236573916,
+      "grad_norm": 47.15625,
+      "learning_rate": 9.861591949674732e-06,
+      "loss": 16.5972,
+      "step": 48990
+    },
+    {
+      "epoch": 0.8859938387265195,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.861563697368073e-06,
+      "loss": 16.8942,
+      "step": 49000
+    },
+    {
+      "epoch": 0.8861746537956473,
+      "grad_norm": 45.34375,
+      "learning_rate": 9.861535445061413e-06,
+      "loss": 16.5526,
+      "step": 49010
+    },
+    {
+      "epoch": 0.8863554688647752,
+      "grad_norm": 41.34375,
+      "learning_rate": 9.861507192754754e-06,
+      "loss": 16.7797,
+      "step": 49020
+    },
+    {
+      "epoch": 0.886536283933903,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.861478940448093e-06,
+      "loss": 16.9787,
+      "step": 49030
+    },
+    {
+      "epoch": 0.8867170990030309,
+      "grad_norm": 42.6875,
+      "learning_rate": 9.861450688141434e-06,
+      "loss": 16.6541,
+      "step": 49040
+    },
+    {
+      "epoch": 0.8868979140721588,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.861422435834774e-06,
+      "loss": 16.7543,
+      "step": 49050
+    },
+    {
+      "epoch": 0.8870787291412866,
+      "grad_norm": 43.0,
+      "learning_rate": 9.861394183528115e-06,
+      "loss": 16.7508,
+      "step": 49060
+    },
+    {
+      "epoch": 0.8872595442104145,
+      "grad_norm": 40.3125,
+      "learning_rate": 9.861365931221455e-06,
+      "loss": 16.7637,
+      "step": 49070
+    },
+    {
+      "epoch": 0.8874403592795423,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.861337678914796e-06,
+      "loss": 16.0263,
+      "step": 49080
+    },
+    {
+      "epoch": 0.8876211743486703,
+      "grad_norm": 42.625,
+      "learning_rate": 9.861309426608137e-06,
+      "loss": 16.8237,
+      "step": 49090
+    },
+    {
+      "epoch": 0.8878019894177981,
+      "grad_norm": 40.6875,
+      "learning_rate": 9.861281174301477e-06,
+      "loss": 17.5541,
+      "step": 49100
+    },
+    {
+      "epoch": 0.8879828044869259,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.861252921994818e-06,
+      "loss": 17.1185,
+      "step": 49110
+    },
+    {
+      "epoch": 0.8881636195560538,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.861224669688157e-06,
+      "loss": 16.7611,
+      "step": 49120
+    },
+    {
+      "epoch": 0.8883444346251816,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.861196417381497e-06,
+      "loss": 16.2662,
+      "step": 49130
+    },
+    {
+      "epoch": 0.8885252496943096,
+      "grad_norm": 43.03125,
+      "learning_rate": 9.861168165074838e-06,
+      "loss": 16.8431,
+      "step": 49140
+    },
+    {
+      "epoch": 0.8887060647634374,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.861139912768179e-06,
+      "loss": 16.5856,
+      "step": 49150
+    },
+    {
+      "epoch": 0.8888868798325652,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.86111166046152e-06,
+      "loss": 17.3069,
+      "step": 49160
+    },
+    {
+      "epoch": 0.8890676949016931,
+      "grad_norm": 43.5,
+      "learning_rate": 9.86108340815486e-06,
+      "loss": 16.948,
+      "step": 49170
+    },
+    {
+      "epoch": 0.8892485099708209,
+      "grad_norm": 44.25,
+      "learning_rate": 9.8610551558482e-06,
+      "loss": 16.3898,
+      "step": 49180
+    },
+    {
+      "epoch": 0.8894293250399489,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.861026903541541e-06,
+      "loss": 16.5265,
+      "step": 49190
+    },
+    {
+      "epoch": 0.8896101401090767,
+      "grad_norm": 45.71875,
+      "learning_rate": 9.860998651234882e-06,
+      "loss": 16.7786,
+      "step": 49200
+    },
+    {
+      "epoch": 0.8897909551782045,
+      "grad_norm": 45.84375,
+      "learning_rate": 9.86097039892822e-06,
+      "loss": 16.3813,
+      "step": 49210
+    },
+    {
+      "epoch": 0.8899717702473324,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.860942146621561e-06,
+      "loss": 16.3776,
+      "step": 49220
+    },
+    {
+      "epoch": 0.8901525853164602,
+      "grad_norm": 42.75,
+      "learning_rate": 9.860913894314902e-06,
+      "loss": 16.5008,
+      "step": 49230
+    },
+    {
+      "epoch": 0.8903334003855882,
+      "grad_norm": 44.0,
+      "learning_rate": 9.860885642008243e-06,
+      "loss": 16.4967,
+      "step": 49240
+    },
+    {
+      "epoch": 0.890514215454716,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.860857389701583e-06,
+      "loss": 16.8697,
+      "step": 49250
+    },
+    {
+      "epoch": 0.8906950305238439,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.860829137394924e-06,
+      "loss": 16.6238,
+      "step": 49260
+    },
+    {
+      "epoch": 0.8908758455929717,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.860800885088264e-06,
+      "loss": 16.7561,
+      "step": 49270
+    },
+    {
+      "epoch": 0.8910566606620995,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.860772632781605e-06,
+      "loss": 17.2425,
+      "step": 49280
+    },
+    {
+      "epoch": 0.8912374757312275,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.860744380474944e-06,
+      "loss": 16.5298,
+      "step": 49290
+    },
+    {
+      "epoch": 0.8914182908003553,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.860716128168285e-06,
+      "loss": 16.7332,
+      "step": 49300
+    },
+    {
+      "epoch": 0.8915991058694832,
+      "grad_norm": 44.5,
+      "learning_rate": 9.860687875861625e-06,
+      "loss": 17.0327,
+      "step": 49310
+    },
+    {
+      "epoch": 0.891779920938611,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.860659623554966e-06,
+      "loss": 16.7635,
+      "step": 49320
+    },
+    {
+      "epoch": 0.8919607360077388,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.860631371248307e-06,
+      "loss": 16.7297,
+      "step": 49330
+    },
+    {
+      "epoch": 0.8921415510768668,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.860603118941647e-06,
+      "loss": 16.9263,
+      "step": 49340
+    },
+    {
+      "epoch": 0.8923223661459946,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.860574866634988e-06,
+      "loss": 17.077,
+      "step": 49350
+    },
+    {
+      "epoch": 0.8925031812151225,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.860546614328328e-06,
+      "loss": 16.648,
+      "step": 49360
+    },
+    {
+      "epoch": 0.8926839962842503,
+      "grad_norm": 42.0625,
+      "learning_rate": 9.860518362021669e-06,
+      "loss": 16.6563,
+      "step": 49370
+    },
+    {
+      "epoch": 0.8928648113533781,
+      "grad_norm": 42.15625,
+      "learning_rate": 9.860490109715008e-06,
+      "loss": 16.6227,
+      "step": 49380
+    },
+    {
+      "epoch": 0.8930456264225061,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.860461857408349e-06,
+      "loss": 16.5707,
+      "step": 49390
+    },
+    {
+      "epoch": 0.8932264414916339,
+      "grad_norm": 39.40625,
+      "learning_rate": 9.86043360510169e-06,
+      "loss": 17.3028,
+      "step": 49400
+    },
+    {
+      "epoch": 0.8934072565607618,
+      "grad_norm": 44.25,
+      "learning_rate": 9.86040535279503e-06,
+      "loss": 16.9679,
+      "step": 49410
+    },
+    {
+      "epoch": 0.8935880716298896,
+      "grad_norm": 46.0,
+      "learning_rate": 9.86037710048837e-06,
+      "loss": 16.7102,
+      "step": 49420
+    },
+    {
+      "epoch": 0.8937688866990176,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.860348848181711e-06,
+      "loss": 16.5409,
+      "step": 49430
+    },
+    {
+      "epoch": 0.8939497017681454,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.860320595875052e-06,
+      "loss": 16.6154,
+      "step": 49440
+    },
+    {
+      "epoch": 0.8941305168372732,
+      "grad_norm": 44.125,
+      "learning_rate": 9.860292343568392e-06,
+      "loss": 16.6337,
+      "step": 49450
+    },
+    {
+      "epoch": 0.8943113319064011,
+      "grad_norm": 43.0,
+      "learning_rate": 9.860264091261731e-06,
+      "loss": 16.6019,
+      "step": 49460
+    },
+    {
+      "epoch": 0.8944921469755289,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.860235838955072e-06,
+      "loss": 16.6053,
+      "step": 49470
+    },
+    {
+      "epoch": 0.8946729620446568,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.860207586648412e-06,
+      "loss": 16.5996,
+      "step": 49480
+    },
+    {
+      "epoch": 0.8948537771137847,
+      "grad_norm": 44.25,
+      "learning_rate": 9.860179334341753e-06,
+      "loss": 16.5798,
+      "step": 49490
+    },
+    {
+      "epoch": 0.8950345921829125,
+      "grad_norm": 43.5,
+      "learning_rate": 9.860151082035094e-06,
+      "loss": 16.9582,
+      "step": 49500
+    },
+    {
+      "epoch": 0.8952154072520404,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.860122829728434e-06,
+      "loss": 16.8579,
+      "step": 49510
+    },
+    {
+      "epoch": 0.8953962223211682,
+      "grad_norm": 45.125,
+      "learning_rate": 9.860094577421775e-06,
+      "loss": 16.5535,
+      "step": 49520
+    },
+    {
+      "epoch": 0.8955770373902961,
+      "grad_norm": 43.875,
+      "learning_rate": 9.860066325115116e-06,
+      "loss": 17.1519,
+      "step": 49530
+    },
+    {
+      "epoch": 0.895757852459424,
+      "grad_norm": 42.375,
+      "learning_rate": 9.860038072808456e-06,
+      "loss": 16.4699,
+      "step": 49540
+    },
+    {
+      "epoch": 0.8959386675285518,
+      "grad_norm": 42.78125,
+      "learning_rate": 9.860009820501795e-06,
+      "loss": 16.67,
+      "step": 49550
+    },
+    {
+      "epoch": 0.8961194825976797,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.859981568195136e-06,
+      "loss": 17.0471,
+      "step": 49560
+    },
+    {
+      "epoch": 0.8963002976668075,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.859953315888476e-06,
+      "loss": 17.0289,
+      "step": 49570
+    },
+    {
+      "epoch": 0.8964811127359354,
+      "grad_norm": 46.0625,
+      "learning_rate": 9.859925063581817e-06,
+      "loss": 16.8848,
+      "step": 49580
+    },
+    {
+      "epoch": 0.8966619278050633,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.859896811275158e-06,
+      "loss": 17.0249,
+      "step": 49590
+    },
+    {
+      "epoch": 0.8968427428741912,
+      "grad_norm": 42.03125,
+      "learning_rate": 9.859868558968497e-06,
+      "loss": 16.5212,
+      "step": 49600
+    },
+    {
+      "epoch": 0.897023557943319,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.859840306661839e-06,
+      "loss": 16.6439,
+      "step": 49610
+    },
+    {
+      "epoch": 0.8972043730124468,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.85981205435518e-06,
+      "loss": 16.4337,
+      "step": 49620
+    },
+    {
+      "epoch": 0.8973851880815747,
+      "grad_norm": 45.1875,
+      "learning_rate": 9.85978380204852e-06,
+      "loss": 16.9367,
+      "step": 49630
+    },
+    {
+      "epoch": 0.8975660031507026,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.859755549741859e-06,
+      "loss": 17.0094,
+      "step": 49640
+    },
+    {
+      "epoch": 0.8977468182198305,
+      "grad_norm": 48.46875,
+      "learning_rate": 9.8597272974352e-06,
+      "loss": 17.0798,
+      "step": 49650
+    },
+    {
+      "epoch": 0.8979276332889583,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.85969904512854e-06,
+      "loss": 16.8546,
+      "step": 49660
+    },
+    {
+      "epoch": 0.8981084483580861,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.859670792821881e-06,
+      "loss": 16.4153,
+      "step": 49670
+    },
+    {
+      "epoch": 0.898289263427214,
+      "grad_norm": 46.1875,
+      "learning_rate": 9.859642540515222e-06,
+      "loss": 16.5445,
+      "step": 49680
+    },
+    {
+      "epoch": 0.8984700784963419,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.859614288208562e-06,
+      "loss": 16.6849,
+      "step": 49690
+    },
+    {
+      "epoch": 0.8986508935654698,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.859586035901903e-06,
+      "loss": 16.7131,
+      "step": 49700
+    },
+    {
+      "epoch": 0.8988317086345976,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.859557783595243e-06,
+      "loss": 16.6575,
+      "step": 49710
+    },
+    {
+      "epoch": 0.8990125237037254,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.859529531288582e-06,
+      "loss": 16.6249,
+      "step": 49720
+    },
+    {
+      "epoch": 0.8991933387728533,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.859501278981923e-06,
+      "loss": 16.9034,
+      "step": 49730
+    },
+    {
+      "epoch": 0.8993741538419812,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.859473026675264e-06,
+      "loss": 16.7501,
+      "step": 49740
+    },
+    {
+      "epoch": 0.8995549689111091,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.859444774368604e-06,
+      "loss": 16.6332,
+      "step": 49750
+    },
+    {
+      "epoch": 0.8997357839802369,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.859416522061945e-06,
+      "loss": 16.6388,
+      "step": 49760
+    },
+    {
+      "epoch": 0.8999165990493648,
+      "grad_norm": 44.75,
+      "learning_rate": 9.859388269755284e-06,
+      "loss": 16.5714,
+      "step": 49770
+    },
+    {
+      "epoch": 0.9000974141184926,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.859360017448626e-06,
+      "loss": 16.5571,
+      "step": 49780
+    },
+    {
+      "epoch": 0.9002782291876205,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.859331765141967e-06,
+      "loss": 16.8097,
+      "step": 49790
+    },
+    {
+      "epoch": 0.9004590442567484,
+      "grad_norm": 39.1875,
+      "learning_rate": 9.859303512835307e-06,
+      "loss": 17.1475,
+      "step": 49800
+    },
+    {
+      "epoch": 0.9006398593258762,
+      "grad_norm": 43.875,
+      "learning_rate": 9.859275260528646e-06,
+      "loss": 17.1851,
+      "step": 49810
+    },
+    {
+      "epoch": 0.9008206743950041,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.859247008221987e-06,
+      "loss": 16.6011,
+      "step": 49820
+    },
+    {
+      "epoch": 0.9010014894641319,
+      "grad_norm": 44.875,
+      "learning_rate": 9.859218755915327e-06,
+      "loss": 16.3234,
+      "step": 49830
+    },
+    {
+      "epoch": 0.9011823045332598,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.859190503608668e-06,
+      "loss": 16.7468,
+      "step": 49840
+    },
+    {
+      "epoch": 0.9013631196023877,
+      "grad_norm": 49.09375,
+      "learning_rate": 9.859162251302009e-06,
+      "loss": 17.3395,
+      "step": 49850
+    },
+    {
+      "epoch": 0.9015439346715155,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.859133998995348e-06,
+      "loss": 16.6495,
+      "step": 49860
+    },
+    {
+      "epoch": 0.9017247497406434,
+      "grad_norm": 44.53125,
+      "learning_rate": 9.85910574668869e-06,
+      "loss": 16.7057,
+      "step": 49870
+    },
+    {
+      "epoch": 0.9019055648097712,
+      "grad_norm": 44.625,
+      "learning_rate": 9.85907749438203e-06,
+      "loss": 16.5376,
+      "step": 49880
+    },
+    {
+      "epoch": 0.9020863798788991,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.85904924207537e-06,
+      "loss": 17.2714,
+      "step": 49890
+    },
+    {
+      "epoch": 0.902267194948027,
+      "grad_norm": 46.78125,
+      "learning_rate": 9.85902098976871e-06,
+      "loss": 16.6724,
+      "step": 49900
+    },
+    {
+      "epoch": 0.9024480100171548,
+      "grad_norm": 43.875,
+      "learning_rate": 9.85899273746205e-06,
+      "loss": 16.9802,
+      "step": 49910
+    },
+    {
+      "epoch": 0.9026288250862827,
+      "grad_norm": 45.1875,
+      "learning_rate": 9.858964485155391e-06,
+      "loss": 17.1209,
+      "step": 49920
+    },
+    {
+      "epoch": 0.9028096401554105,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.858936232848732e-06,
+      "loss": 16.8002,
+      "step": 49930
+    },
+    {
+      "epoch": 0.9029904552245384,
+      "grad_norm": 43.125,
+      "learning_rate": 9.858907980542073e-06,
+      "loss": 16.709,
+      "step": 49940
+    },
+    {
+      "epoch": 0.9031712702936663,
+      "grad_norm": 45.96875,
+      "learning_rate": 9.858879728235412e-06,
+      "loss": 16.9123,
+      "step": 49950
+    },
+    {
+      "epoch": 0.9033520853627941,
+      "grad_norm": 46.0625,
+      "learning_rate": 9.858851475928754e-06,
+      "loss": 16.4468,
+      "step": 49960
+    },
+    {
+      "epoch": 0.903532900431922,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.858823223622094e-06,
+      "loss": 16.6946,
+      "step": 49970
+    },
+    {
+      "epoch": 0.9037137155010498,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.858794971315433e-06,
+      "loss": 16.6168,
+      "step": 49980
+    },
+    {
+      "epoch": 0.9038945305701778,
+      "grad_norm": 43.5,
+      "learning_rate": 9.858766719008774e-06,
+      "loss": 16.8038,
+      "step": 49990
+    },
+    {
+      "epoch": 0.9040753456393056,
+      "grad_norm": 41.90625,
+      "learning_rate": 9.858738466702115e-06,
+      "loss": 16.6773,
+      "step": 50000
+    },
+    {
+      "epoch": 0.9040753456393056,
+      "eval_loss": 2.0931339263916016,
+      "eval_runtime": 229.7481,
+      "eval_samples_per_second": 3160.239,
+      "eval_steps_per_second": 49.38,
+      "step": 50000
+    },
+    {
+      "epoch": 0.9042561607084334,
+      "grad_norm": 42.5,
+      "learning_rate": 9.858710214395455e-06,
+      "loss": 16.8778,
+      "step": 50010
+    },
+    {
+      "epoch": 0.9044369757775613,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.858681962088796e-06,
+      "loss": 16.5687,
+      "step": 50020
+    },
+    {
+      "epoch": 0.9046177908466891,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.858653709782135e-06,
+      "loss": 16.6903,
+      "step": 50030
+    },
+    {
+      "epoch": 0.9047986059158171,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.858625457475477e-06,
+      "loss": 16.3654,
+      "step": 50040
+    },
+    {
+      "epoch": 0.9049794209849449,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.858597205168818e-06,
+      "loss": 16.9448,
+      "step": 50050
+    },
+    {
+      "epoch": 0.9051602360540727,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.858568952862158e-06,
+      "loss": 17.0241,
+      "step": 50060
+    },
+    {
+      "epoch": 0.9053410511232006,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.858540700555497e-06,
+      "loss": 17.1035,
+      "step": 50070
+    },
+    {
+      "epoch": 0.9055218661923284,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.858512448248838e-06,
+      "loss": 17.3552,
+      "step": 50080
+    },
+    {
+      "epoch": 0.9057026812614564,
+      "grad_norm": 44.125,
+      "learning_rate": 9.858484195942179e-06,
+      "loss": 16.1245,
+      "step": 50090
+    },
+    {
+      "epoch": 0.9058834963305842,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.85845594363552e-06,
+      "loss": 16.7964,
+      "step": 50100
+    },
+    {
+      "epoch": 0.906064311399712,
+      "grad_norm": 46.71875,
+      "learning_rate": 9.85842769132886e-06,
+      "loss": 16.774,
+      "step": 50110
+    },
+    {
+      "epoch": 0.9062451264688399,
+      "grad_norm": 45.4375,
+      "learning_rate": 9.858399439022199e-06,
+      "loss": 16.5967,
+      "step": 50120
+    },
+    {
+      "epoch": 0.9064259415379677,
+      "grad_norm": 45.03125,
+      "learning_rate": 9.858371186715541e-06,
+      "loss": 16.5357,
+      "step": 50130
+    },
+    {
+      "epoch": 0.9066067566070957,
+      "grad_norm": 43.90625,
+      "learning_rate": 9.858342934408882e-06,
+      "loss": 16.9263,
+      "step": 50140
+    },
+    {
+      "epoch": 0.9067875716762235,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.85831468210222e-06,
+      "loss": 16.9574,
+      "step": 50150
+    },
+    {
+      "epoch": 0.9069683867453514,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.858286429795561e-06,
+      "loss": 16.55,
+      "step": 50160
+    },
+    {
+      "epoch": 0.9071492018144792,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.858258177488902e-06,
+      "loss": 16.7054,
+      "step": 50170
+    },
+    {
+      "epoch": 0.907330016883607,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.858229925182242e-06,
+      "loss": 16.5315,
+      "step": 50180
+    },
+    {
+      "epoch": 0.907510831952735,
+      "grad_norm": 46.71875,
+      "learning_rate": 9.858201672875583e-06,
+      "loss": 17.2083,
+      "step": 50190
+    },
+    {
+      "epoch": 0.9076916470218628,
+      "grad_norm": 45.96875,
+      "learning_rate": 9.858173420568922e-06,
+      "loss": 16.9714,
+      "step": 50200
+    },
+    {
+      "epoch": 0.9078724620909907,
+      "grad_norm": 45.625,
+      "learning_rate": 9.858145168262263e-06,
+      "loss": 16.5915,
+      "step": 50210
+    },
+    {
+      "epoch": 0.9080532771601185,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.858116915955605e-06,
+      "loss": 16.7036,
+      "step": 50220
+    },
+    {
+      "epoch": 0.9082340922292463,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.858088663648946e-06,
+      "loss": 16.9555,
+      "step": 50230
+    },
+    {
+      "epoch": 0.9084149072983743,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.858060411342284e-06,
+      "loss": 16.5951,
+      "step": 50240
+    },
+    {
+      "epoch": 0.9085957223675021,
+      "grad_norm": 43.15625,
+      "learning_rate": 9.858032159035625e-06,
+      "loss": 16.8463,
+      "step": 50250
+    },
+    {
+      "epoch": 0.90877653743663,
+      "grad_norm": 42.0,
+      "learning_rate": 9.858003906728966e-06,
+      "loss": 16.7482,
+      "step": 50260
+    },
+    {
+      "epoch": 0.9089573525057578,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.857975654422306e-06,
+      "loss": 16.4641,
+      "step": 50270
+    },
+    {
+      "epoch": 0.9091381675748856,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.857947402115647e-06,
+      "loss": 17.0,
+      "step": 50280
+    },
+    {
+      "epoch": 0.9093189826440136,
+      "grad_norm": 42.125,
+      "learning_rate": 9.857919149808986e-06,
+      "loss": 16.7016,
+      "step": 50290
+    },
+    {
+      "epoch": 0.9094997977131414,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.857890897502327e-06,
+      "loss": 16.5641,
+      "step": 50300
+    },
+    {
+      "epoch": 0.9096806127822693,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.857862645195669e-06,
+      "loss": 16.8491,
+      "step": 50310
+    },
+    {
+      "epoch": 0.9098614278513971,
+      "grad_norm": 43.75,
+      "learning_rate": 9.857834392889008e-06,
+      "loss": 16.7857,
+      "step": 50320
+    },
+    {
+      "epoch": 0.910042242920525,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.857806140582348e-06,
+      "loss": 16.7927,
+      "step": 50330
+    },
+    {
+      "epoch": 0.9102230579896529,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.857777888275689e-06,
+      "loss": 17.1758,
+      "step": 50340
+    },
+    {
+      "epoch": 0.9104038730587807,
+      "grad_norm": 41.0,
+      "learning_rate": 9.85774963596903e-06,
+      "loss": 16.2274,
+      "step": 50350
+    },
+    {
+      "epoch": 0.9105846881279086,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.85772138366237e-06,
+      "loss": 16.6145,
+      "step": 50360
+    },
+    {
+      "epoch": 0.9107655031970364,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.85769313135571e-06,
+      "loss": 16.3565,
+      "step": 50370
+    },
+    {
+      "epoch": 0.9109463182661643,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.85766487904905e-06,
+      "loss": 16.6859,
+      "step": 50380
+    },
+    {
+      "epoch": 0.9111271333352922,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.85763662674239e-06,
+      "loss": 16.6597,
+      "step": 50390
+    },
+    {
+      "epoch": 0.91130794840442,
+      "grad_norm": 44.75,
+      "learning_rate": 9.857608374435733e-06,
+      "loss": 17.1254,
+      "step": 50400
+    },
+    {
+      "epoch": 0.9114887634735479,
+      "grad_norm": 44.75,
+      "learning_rate": 9.857580122129072e-06,
+      "loss": 16.8251,
+      "step": 50410
+    },
+    {
+      "epoch": 0.9116695785426757,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.857551869822412e-06,
+      "loss": 16.9839,
+      "step": 50420
+    },
+    {
+      "epoch": 0.9118503936118036,
+      "grad_norm": 43.875,
+      "learning_rate": 9.857523617515753e-06,
+      "loss": 16.8362,
+      "step": 50430
+    },
+    {
+      "epoch": 0.9120312086809315,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.857495365209094e-06,
+      "loss": 16.8484,
+      "step": 50440
+    },
+    {
+      "epoch": 0.9122120237500593,
+      "grad_norm": 47.40625,
+      "learning_rate": 9.857467112902434e-06,
+      "loss": 16.4676,
+      "step": 50450
+    },
+    {
+      "epoch": 0.9123928388191872,
+      "grad_norm": 43.0,
+      "learning_rate": 9.857438860595773e-06,
+      "loss": 16.7908,
+      "step": 50460
+    },
+    {
+      "epoch": 0.912573653888315,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.857410608289114e-06,
+      "loss": 16.5521,
+      "step": 50470
+    },
+    {
+      "epoch": 0.9127544689574429,
+      "grad_norm": 41.875,
+      "learning_rate": 9.857382355982456e-06,
+      "loss": 16.5968,
+      "step": 50480
+    },
+    {
+      "epoch": 0.9129352840265708,
+      "grad_norm": 46.65625,
+      "learning_rate": 9.857354103675795e-06,
+      "loss": 16.4031,
+      "step": 50490
+    },
+    {
+      "epoch": 0.9131160990956987,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.857325851369136e-06,
+      "loss": 17.1271,
+      "step": 50500
+    },
+    {
+      "epoch": 0.9132969141648265,
+      "grad_norm": 43.0,
+      "learning_rate": 9.857297599062476e-06,
+      "loss": 16.6066,
+      "step": 50510
+    },
+    {
+      "epoch": 0.9134777292339543,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.857269346755817e-06,
+      "loss": 16.6022,
+      "step": 50520
+    },
+    {
+      "epoch": 0.9136585443030822,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.857241094449157e-06,
+      "loss": 16.5399,
+      "step": 50530
+    },
+    {
+      "epoch": 0.9138393593722101,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.857212842142498e-06,
+      "loss": 16.75,
+      "step": 50540
+    },
+    {
+      "epoch": 0.914020174441338,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.857184589835837e-06,
+      "loss": 17.1319,
+      "step": 50550
+    },
+    {
+      "epoch": 0.9142009895104658,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.857156337529178e-06,
+      "loss": 16.3081,
+      "step": 50560
+    },
+    {
+      "epoch": 0.9143818045795936,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.85712808522252e-06,
+      "loss": 16.4195,
+      "step": 50570
+    },
+    {
+      "epoch": 0.9145626196487215,
+      "grad_norm": 47.9375,
+      "learning_rate": 9.857099832915859e-06,
+      "loss": 16.6389,
+      "step": 50580
+    },
+    {
+      "epoch": 0.9147434347178494,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.8570715806092e-06,
+      "loss": 16.4462,
+      "step": 50590
+    },
+    {
+      "epoch": 0.9149242497869773,
+      "grad_norm": 46.78125,
+      "learning_rate": 9.85704332830254e-06,
+      "loss": 17.0027,
+      "step": 50600
+    },
+    {
+      "epoch": 0.9151050648561051,
+      "grad_norm": 48.78125,
+      "learning_rate": 9.85701507599588e-06,
+      "loss": 17.1624,
+      "step": 50610
+    },
+    {
+      "epoch": 0.9152858799252329,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.856986823689221e-06,
+      "loss": 17.0643,
+      "step": 50620
+    },
+    {
+      "epoch": 0.9154666949943608,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.85695857138256e-06,
+      "loss": 16.2225,
+      "step": 50630
+    },
+    {
+      "epoch": 0.9156475100634887,
+      "grad_norm": 45.375,
+      "learning_rate": 9.856930319075901e-06,
+      "loss": 16.9626,
+      "step": 50640
+    },
+    {
+      "epoch": 0.9158283251326166,
+      "grad_norm": 40.28125,
+      "learning_rate": 9.856902066769242e-06,
+      "loss": 17.0447,
+      "step": 50650
+    },
+    {
+      "epoch": 0.9160091402017444,
+      "grad_norm": 48.125,
+      "learning_rate": 9.856873814462584e-06,
+      "loss": 16.5066,
+      "step": 50660
+    },
+    {
+      "epoch": 0.9161899552708723,
+      "grad_norm": 45.6875,
+      "learning_rate": 9.856845562155923e-06,
+      "loss": 16.6187,
+      "step": 50670
+    },
+    {
+      "epoch": 0.9163707703400001,
+      "grad_norm": 46.34375,
+      "learning_rate": 9.856817309849263e-06,
+      "loss": 16.5057,
+      "step": 50680
+    },
+    {
+      "epoch": 0.916551585409128,
+      "grad_norm": 45.8125,
+      "learning_rate": 9.856789057542604e-06,
+      "loss": 16.7443,
+      "step": 50690
+    },
+    {
+      "epoch": 0.9167324004782559,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.856760805235945e-06,
+      "loss": 16.4325,
+      "step": 50700
+    },
+    {
+      "epoch": 0.9169132155473837,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.856732552929285e-06,
+      "loss": 16.9035,
+      "step": 50710
+    },
+    {
+      "epoch": 0.9170940306165116,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.856704300622624e-06,
+      "loss": 16.5906,
+      "step": 50720
+    },
+    {
+      "epoch": 0.9172748456856394,
+      "grad_norm": 46.90625,
+      "learning_rate": 9.856676048315965e-06,
+      "loss": 16.8493,
+      "step": 50730
+    },
+    {
+      "epoch": 0.9174556607547673,
+      "grad_norm": 45.84375,
+      "learning_rate": 9.856647796009305e-06,
+      "loss": 16.6537,
+      "step": 50740
+    },
+    {
+      "epoch": 0.9176364758238952,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.856619543702646e-06,
+      "loss": 16.8398,
+      "step": 50750
+    },
+    {
+      "epoch": 0.917817290893023,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.856591291395987e-06,
+      "loss": 16.7299,
+      "step": 50760
+    },
+    {
+      "epoch": 0.9179981059621509,
+      "grad_norm": 41.625,
+      "learning_rate": 9.856563039089327e-06,
+      "loss": 16.2667,
+      "step": 50770
+    },
+    {
+      "epoch": 0.9181789210312787,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.856534786782668e-06,
+      "loss": 16.7683,
+      "step": 50780
+    },
+    {
+      "epoch": 0.9183597361004066,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.856506534476009e-06,
+      "loss": 16.4613,
+      "step": 50790
+    },
+    {
+      "epoch": 0.9185405511695345,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.856478282169347e-06,
+      "loss": 16.8121,
+      "step": 50800
+    },
+    {
+      "epoch": 0.9187213662386623,
+      "grad_norm": 44.5,
+      "learning_rate": 9.856450029862688e-06,
+      "loss": 16.7824,
+      "step": 50810
+    },
+    {
+      "epoch": 0.9189021813077902,
+      "grad_norm": 46.375,
+      "learning_rate": 9.856421777556029e-06,
+      "loss": 16.6512,
+      "step": 50820
+    },
+    {
+      "epoch": 0.919082996376918,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.856393525249371e-06,
+      "loss": 16.9636,
+      "step": 50830
+    },
+    {
+      "epoch": 0.919263811446046,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.85636527294271e-06,
+      "loss": 16.3377,
+      "step": 50840
+    },
+    {
+      "epoch": 0.9194446265151738,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.85633702063605e-06,
+      "loss": 16.2992,
+      "step": 50850
+    },
+    {
+      "epoch": 0.9196254415843016,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.856308768329391e-06,
+      "loss": 16.8767,
+      "step": 50860
+    },
+    {
+      "epoch": 0.9198062566534295,
+      "grad_norm": 42.8125,
+      "learning_rate": 9.856280516022732e-06,
+      "loss": 17.1975,
+      "step": 50870
+    },
+    {
+      "epoch": 0.9199870717225573,
+      "grad_norm": 46.25,
+      "learning_rate": 9.856252263716072e-06,
+      "loss": 16.8003,
+      "step": 50880
+    },
+    {
+      "epoch": 0.9201678867916853,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.856224011409411e-06,
+      "loss": 16.5021,
+      "step": 50890
+    },
+    {
+      "epoch": 0.9203487018608131,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.856195759102752e-06,
+      "loss": 16.8396,
+      "step": 50900
+    },
+    {
+      "epoch": 0.9205295169299409,
+      "grad_norm": 41.78125,
+      "learning_rate": 9.856167506796093e-06,
+      "loss": 16.3,
+      "step": 50910
+    },
+    {
+      "epoch": 0.9207103319990688,
+      "grad_norm": 47.40625,
+      "learning_rate": 9.856139254489433e-06,
+      "loss": 17.3092,
+      "step": 50920
+    },
+    {
+      "epoch": 0.9208911470681966,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.856111002182774e-06,
+      "loss": 17.0263,
+      "step": 50930
+    },
+    {
+      "epoch": 0.9210719621373246,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.856082749876115e-06,
+      "loss": 16.6091,
+      "step": 50940
+    },
+    {
+      "epoch": 0.9212527772064524,
+      "grad_norm": 46.5,
+      "learning_rate": 9.856054497569455e-06,
+      "loss": 17.2658,
+      "step": 50950
+    },
+    {
+      "epoch": 0.9214335922755802,
+      "grad_norm": 43.1875,
+      "learning_rate": 9.856026245262796e-06,
+      "loss": 16.9387,
+      "step": 50960
+    },
+    {
+      "epoch": 0.9216144073447081,
+      "grad_norm": 42.1875,
+      "learning_rate": 9.855997992956136e-06,
+      "loss": 16.5608,
+      "step": 50970
+    },
+    {
+      "epoch": 0.9217952224138359,
+      "grad_norm": 44.375,
+      "learning_rate": 9.855969740649475e-06,
+      "loss": 16.4755,
+      "step": 50980
+    },
+    {
+      "epoch": 0.9219760374829639,
+      "grad_norm": 44.625,
+      "learning_rate": 9.855941488342816e-06,
+      "loss": 16.774,
+      "step": 50990
+    },
+    {
+      "epoch": 0.9221568525520917,
+      "grad_norm": 46.0,
+      "learning_rate": 9.855913236036157e-06,
+      "loss": 16.2279,
+      "step": 51000
+    },
+    {
+      "epoch": 0.9223376676212196,
+      "grad_norm": 43.25,
+      "learning_rate": 9.855884983729497e-06,
+      "loss": 16.9542,
+      "step": 51010
+    },
+    {
+      "epoch": 0.9225184826903474,
+      "grad_norm": 46.4375,
+      "learning_rate": 9.855856731422838e-06,
+      "loss": 17.0627,
+      "step": 51020
+    },
+    {
+      "epoch": 0.9226992977594752,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.855828479116178e-06,
+      "loss": 16.5087,
+      "step": 51030
+    },
+    {
+      "epoch": 0.9228801128286032,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.855800226809519e-06,
+      "loss": 16.333,
+      "step": 51040
+    },
+    {
+      "epoch": 0.923060927897731,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.85577197450286e-06,
+      "loss": 16.7786,
+      "step": 51050
+    },
+    {
+      "epoch": 0.9232417429668589,
+      "grad_norm": 43.59375,
+      "learning_rate": 9.855743722196199e-06,
+      "loss": 16.6454,
+      "step": 51060
+    },
+    {
+      "epoch": 0.9234225580359867,
+      "grad_norm": 44.71875,
+      "learning_rate": 9.85571546988954e-06,
+      "loss": 16.572,
+      "step": 51070
+    },
+    {
+      "epoch": 0.9236033731051145,
+      "grad_norm": 41.09375,
+      "learning_rate": 9.85568721758288e-06,
+      "loss": 16.4866,
+      "step": 51080
+    },
+    {
+      "epoch": 0.9237841881742425,
+      "grad_norm": 47.84375,
+      "learning_rate": 9.85565896527622e-06,
+      "loss": 16.9147,
+      "step": 51090
+    },
+    {
+      "epoch": 0.9239650032433703,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.855630712969561e-06,
+      "loss": 16.818,
+      "step": 51100
+    },
+    {
+      "epoch": 0.9241458183124982,
+      "grad_norm": 47.375,
+      "learning_rate": 9.855602460662902e-06,
+      "loss": 16.7599,
+      "step": 51110
+    },
+    {
+      "epoch": 0.924326633381626,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.855574208356242e-06,
+      "loss": 16.9495,
+      "step": 51120
+    },
+    {
+      "epoch": 0.9245074484507538,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.855545956049583e-06,
+      "loss": 16.7892,
+      "step": 51130
+    },
+    {
+      "epoch": 0.9246882635198818,
+      "grad_norm": 44.75,
+      "learning_rate": 9.855517703742924e-06,
+      "loss": 16.4863,
+      "step": 51140
+    },
+    {
+      "epoch": 0.9248690785890096,
+      "grad_norm": 46.03125,
+      "learning_rate": 9.855489451436262e-06,
+      "loss": 16.3058,
+      "step": 51150
+    },
+    {
+      "epoch": 0.9250498936581375,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.855461199129603e-06,
+      "loss": 16.1729,
+      "step": 51160
+    },
+    {
+      "epoch": 0.9252307087272653,
+      "grad_norm": 43.625,
+      "learning_rate": 9.855432946822944e-06,
+      "loss": 16.9115,
+      "step": 51170
+    },
+    {
+      "epoch": 0.9254115237963932,
+      "grad_norm": 44.5,
+      "learning_rate": 9.855404694516284e-06,
+      "loss": 16.577,
+      "step": 51180
+    },
+    {
+      "epoch": 0.9255923388655211,
+      "grad_norm": 46.84375,
+      "learning_rate": 9.855376442209625e-06,
+      "loss": 16.3869,
+      "step": 51190
+    },
+    {
+      "epoch": 0.9257731539346489,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.855348189902966e-06,
+      "loss": 16.6404,
+      "step": 51200
+    },
+    {
+      "epoch": 0.9259539690037768,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.855319937596306e-06,
+      "loss": 16.7318,
+      "step": 51210
+    },
+    {
+      "epoch": 0.9261347840729046,
+      "grad_norm": 42.0,
+      "learning_rate": 9.855291685289647e-06,
+      "loss": 16.1691,
+      "step": 51220
+    },
+    {
+      "epoch": 0.9263155991420325,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.855263432982986e-06,
+      "loss": 16.7121,
+      "step": 51230
+    },
+    {
+      "epoch": 0.9264964142111604,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.855235180676326e-06,
+      "loss": 16.653,
+      "step": 51240
+    },
+    {
+      "epoch": 0.9266772292802882,
+      "grad_norm": 44.875,
+      "learning_rate": 9.855206928369667e-06,
+      "loss": 16.7853,
+      "step": 51250
+    },
+    {
+      "epoch": 0.9268580443494161,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.855178676063008e-06,
+      "loss": 16.5499,
+      "step": 51260
+    },
+    {
+      "epoch": 0.9270388594185439,
+      "grad_norm": 41.375,
+      "learning_rate": 9.855150423756348e-06,
+      "loss": 16.4913,
+      "step": 51270
+    },
+    {
+      "epoch": 0.9272196744876718,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.855122171449689e-06,
+      "loss": 16.9152,
+      "step": 51280
+    },
+    {
+      "epoch": 0.9274004895567997,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.85509391914303e-06,
+      "loss": 16.3459,
+      "step": 51290
+    },
+    {
+      "epoch": 0.9275813046259275,
+      "grad_norm": 42.75,
+      "learning_rate": 9.85506566683637e-06,
+      "loss": 16.257,
+      "step": 51300
+    },
+    {
+      "epoch": 0.9277621196950554,
+      "grad_norm": 43.0,
+      "learning_rate": 9.85503741452971e-06,
+      "loss": 16.853,
+      "step": 51310
+    },
+    {
+      "epoch": 0.9279429347641832,
+      "grad_norm": 48.90625,
+      "learning_rate": 9.85500916222305e-06,
+      "loss": 16.3804,
+      "step": 51320
+    },
+    {
+      "epoch": 0.9281237498333111,
+      "grad_norm": 40.75,
+      "learning_rate": 9.85498090991639e-06,
+      "loss": 16.1179,
+      "step": 51330
+    },
+    {
+      "epoch": 0.928304564902439,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.854952657609731e-06,
+      "loss": 16.7624,
+      "step": 51340
+    },
+    {
+      "epoch": 0.9284853799715669,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.854924405303072e-06,
+      "loss": 16.7898,
+      "step": 51350
+    },
+    {
+      "epoch": 0.9286661950406947,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.854896152996412e-06,
+      "loss": 16.8738,
+      "step": 51360
+    },
+    {
+      "epoch": 0.9288470101098225,
+      "grad_norm": 43.25,
+      "learning_rate": 9.854867900689753e-06,
+      "loss": 16.4131,
+      "step": 51370
+    },
+    {
+      "epoch": 0.9290278251789504,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.854839648383093e-06,
+      "loss": 16.6141,
+      "step": 51380
+    },
+    {
+      "epoch": 0.9292086402480783,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.854811396076434e-06,
+      "loss": 16.2441,
+      "step": 51390
+    },
+    {
+      "epoch": 0.9293894553172062,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.854783143769775e-06,
+      "loss": 16.764,
+      "step": 51400
+    },
+    {
+      "epoch": 0.929570270386334,
+      "grad_norm": 46.9375,
+      "learning_rate": 9.854754891463114e-06,
+      "loss": 17.3107,
+      "step": 51410
+    },
+    {
+      "epoch": 0.9297510854554618,
+      "grad_norm": 46.21875,
+      "learning_rate": 9.854726639156454e-06,
+      "loss": 16.2477,
+      "step": 51420
+    },
+    {
+      "epoch": 0.9299319005245897,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.854698386849795e-06,
+      "loss": 16.8256,
+      "step": 51430
+    },
+    {
+      "epoch": 0.9301127155937176,
+      "grad_norm": 45.75,
+      "learning_rate": 9.854670134543135e-06,
+      "loss": 16.806,
+      "step": 51440
+    },
+    {
+      "epoch": 0.9302935306628455,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.854641882236476e-06,
+      "loss": 16.7943,
+      "step": 51450
+    },
+    {
+      "epoch": 0.9304743457319733,
+      "grad_norm": 46.4375,
+      "learning_rate": 9.854613629929817e-06,
+      "loss": 16.4803,
+      "step": 51460
+    },
+    {
+      "epoch": 0.9306551608011011,
+      "grad_norm": 41.9375,
+      "learning_rate": 9.854585377623157e-06,
+      "loss": 17.0244,
+      "step": 51470
+    },
+    {
+      "epoch": 0.930835975870229,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.854557125316498e-06,
+      "loss": 16.4517,
+      "step": 51480
+    },
+    {
+      "epoch": 0.9310167909393569,
+      "grad_norm": 44.84375,
+      "learning_rate": 9.854528873009837e-06,
+      "loss": 16.8978,
+      "step": 51490
+    },
+    {
+      "epoch": 0.9311976060084848,
+      "grad_norm": 46.6875,
+      "learning_rate": 9.854500620703178e-06,
+      "loss": 16.7435,
+      "step": 51500
+    },
+    {
+      "epoch": 0.9313784210776126,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.854472368396518e-06,
+      "loss": 16.402,
+      "step": 51510
+    },
+    {
+      "epoch": 0.9315592361467405,
+      "grad_norm": 48.34375,
+      "learning_rate": 9.854444116089859e-06,
+      "loss": 16.7344,
+      "step": 51520
+    },
+    {
+      "epoch": 0.9317400512158683,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.8544158637832e-06,
+      "loss": 16.4329,
+      "step": 51530
+    },
+    {
+      "epoch": 0.9319208662849962,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.85438761147654e-06,
+      "loss": 16.3157,
+      "step": 51540
+    },
+    {
+      "epoch": 0.9321016813541241,
+      "grad_norm": 42.625,
+      "learning_rate": 9.85435935916988e-06,
+      "loss": 16.9422,
+      "step": 51550
+    },
+    {
+      "epoch": 0.9322824964232519,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.854331106863221e-06,
+      "loss": 16.4778,
+      "step": 51560
+    },
+    {
+      "epoch": 0.9324633114923798,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.854302854556562e-06,
+      "loss": 16.7903,
+      "step": 51570
+    },
+    {
+      "epoch": 0.9326441265615076,
+      "grad_norm": 48.03125,
+      "learning_rate": 9.8542746022499e-06,
+      "loss": 16.8277,
+      "step": 51580
+    },
+    {
+      "epoch": 0.9328249416306355,
+      "grad_norm": 44.15625,
+      "learning_rate": 9.854246349943241e-06,
+      "loss": 16.7422,
+      "step": 51590
+    },
+    {
+      "epoch": 0.9330057566997634,
+      "grad_norm": 42.5,
+      "learning_rate": 9.854218097636582e-06,
+      "loss": 16.8482,
+      "step": 51600
+    },
+    {
+      "epoch": 0.9331865717688912,
+      "grad_norm": 45.625,
+      "learning_rate": 9.854189845329923e-06,
+      "loss": 16.5671,
+      "step": 51610
+    },
+    {
+      "epoch": 0.9333673868380191,
+      "grad_norm": 42.625,
+      "learning_rate": 9.854161593023263e-06,
+      "loss": 16.6159,
+      "step": 51620
+    },
+    {
+      "epoch": 0.9335482019071469,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.854133340716604e-06,
+      "loss": 16.8276,
+      "step": 51630
+    },
+    {
+      "epoch": 0.9337290169762748,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.854105088409945e-06,
+      "loss": 16.4779,
+      "step": 51640
+    },
+    {
+      "epoch": 0.9339098320454027,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.854076836103285e-06,
+      "loss": 16.5682,
+      "step": 51650
+    },
+    {
+      "epoch": 0.9340906471145305,
+      "grad_norm": 42.96875,
+      "learning_rate": 9.854048583796624e-06,
+      "loss": 16.5536,
+      "step": 51660
+    },
+    {
+      "epoch": 0.9342714621836584,
+      "grad_norm": 49.0625,
+      "learning_rate": 9.854020331489965e-06,
+      "loss": 16.8541,
+      "step": 51670
+    },
+    {
+      "epoch": 0.9344522772527862,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.853992079183305e-06,
+      "loss": 17.2924,
+      "step": 51680
+    },
+    {
+      "epoch": 0.9346330923219142,
+      "grad_norm": 45.8125,
+      "learning_rate": 9.853963826876646e-06,
+      "loss": 16.4853,
+      "step": 51690
+    },
+    {
+      "epoch": 0.934813907391042,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.853935574569987e-06,
+      "loss": 16.4445,
+      "step": 51700
+    },
+    {
+      "epoch": 0.9349947224601698,
+      "grad_norm": 46.5625,
+      "learning_rate": 9.853907322263327e-06,
+      "loss": 16.7954,
+      "step": 51710
+    },
+    {
+      "epoch": 0.9351755375292977,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.853879069956668e-06,
+      "loss": 16.6579,
+      "step": 51720
+    },
+    {
+      "epoch": 0.9353563525984255,
+      "grad_norm": 46.5,
+      "learning_rate": 9.853850817650008e-06,
+      "loss": 17.1303,
+      "step": 51730
+    },
+    {
+      "epoch": 0.9355371676675535,
+      "grad_norm": 44.75,
+      "learning_rate": 9.853822565343349e-06,
+      "loss": 16.8715,
+      "step": 51740
+    },
+    {
+      "epoch": 0.9357179827366813,
+      "grad_norm": 43.75,
+      "learning_rate": 9.853794313036688e-06,
+      "loss": 16.7436,
+      "step": 51750
+    },
+    {
+      "epoch": 0.9358987978058091,
+      "grad_norm": 45.5,
+      "learning_rate": 9.853766060730029e-06,
+      "loss": 16.4995,
+      "step": 51760
+    },
+    {
+      "epoch": 0.936079612874937,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.85373780842337e-06,
+      "loss": 16.6631,
+      "step": 51770
+    },
+    {
+      "epoch": 0.9362604279440648,
+      "grad_norm": 41.625,
+      "learning_rate": 9.85370955611671e-06,
+      "loss": 16.1694,
+      "step": 51780
+    },
+    {
+      "epoch": 0.9364412430131928,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.85368130381005e-06,
+      "loss": 16.9699,
+      "step": 51790
+    },
+    {
+      "epoch": 0.9366220580823206,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.853653051503391e-06,
+      "loss": 16.5178,
+      "step": 51800
+    },
+    {
+      "epoch": 0.9368028731514484,
+      "grad_norm": 42.875,
+      "learning_rate": 9.853624799196732e-06,
+      "loss": 16.3057,
+      "step": 51810
+    },
+    {
+      "epoch": 0.9369836882205763,
+      "grad_norm": 45.28125,
+      "learning_rate": 9.853596546890072e-06,
+      "loss": 16.699,
+      "step": 51820
+    },
+    {
+      "epoch": 0.9371645032897041,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.853568294583413e-06,
+      "loss": 16.8024,
+      "step": 51830
+    },
+    {
+      "epoch": 0.9373453183588321,
+      "grad_norm": 41.59375,
+      "learning_rate": 9.853540042276752e-06,
+      "loss": 16.6078,
+      "step": 51840
+    },
+    {
+      "epoch": 0.9375261334279599,
+      "grad_norm": 47.21875,
+      "learning_rate": 9.853511789970093e-06,
+      "loss": 16.8264,
+      "step": 51850
+    },
+    {
+      "epoch": 0.9377069484970878,
+      "grad_norm": 44.9375,
+      "learning_rate": 9.853483537663433e-06,
+      "loss": 16.8107,
+      "step": 51860
+    },
+    {
+      "epoch": 0.9378877635662156,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.853455285356774e-06,
+      "loss": 16.7995,
+      "step": 51870
+    },
+    {
+      "epoch": 0.9380685786353434,
+      "grad_norm": 44.875,
+      "learning_rate": 9.853427033050114e-06,
+      "loss": 16.6142,
+      "step": 51880
+    },
+    {
+      "epoch": 0.9382493937044714,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.853398780743455e-06,
+      "loss": 16.7819,
+      "step": 51890
+    },
+    {
+      "epoch": 0.9384302087735992,
+      "grad_norm": 44.125,
+      "learning_rate": 9.853370528436796e-06,
+      "loss": 16.7242,
+      "step": 51900
+    },
+    {
+      "epoch": 0.9386110238427271,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.853342276130136e-06,
+      "loss": 16.6088,
+      "step": 51910
+    },
+    {
+      "epoch": 0.9387918389118549,
+      "grad_norm": 47.1875,
+      "learning_rate": 9.853314023823475e-06,
+      "loss": 16.1789,
+      "step": 51920
+    },
+    {
+      "epoch": 0.9389726539809827,
+      "grad_norm": 43.9375,
+      "learning_rate": 9.853285771516816e-06,
+      "loss": 16.3855,
+      "step": 51930
+    },
+    {
+      "epoch": 0.9391534690501107,
+      "grad_norm": 45.8125,
+      "learning_rate": 9.853257519210156e-06,
+      "loss": 17.0041,
+      "step": 51940
+    },
+    {
+      "epoch": 0.9393342841192385,
+      "grad_norm": 41.3125,
+      "learning_rate": 9.853229266903497e-06,
+      "loss": 16.7801,
+      "step": 51950
+    },
+    {
+      "epoch": 0.9395150991883664,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.853201014596838e-06,
+      "loss": 16.7844,
+      "step": 51960
+    },
+    {
+      "epoch": 0.9396959142574942,
+      "grad_norm": 42.125,
+      "learning_rate": 9.853172762290178e-06,
+      "loss": 16.6697,
+      "step": 51970
+    },
+    {
+      "epoch": 0.939876729326622,
+      "grad_norm": 42.625,
+      "learning_rate": 9.853144509983519e-06,
+      "loss": 16.4901,
+      "step": 51980
+    },
+    {
+      "epoch": 0.94005754439575,
+      "grad_norm": 44.375,
+      "learning_rate": 9.85311625767686e-06,
+      "loss": 16.4376,
+      "step": 51990
+    },
+    {
+      "epoch": 0.9402383594648778,
+      "grad_norm": 41.625,
+      "learning_rate": 9.8530880053702e-06,
+      "loss": 16.743,
+      "step": 52000
+    },
+    {
+      "epoch": 0.9404191745340057,
+      "grad_norm": 45.125,
+      "learning_rate": 9.853059753063539e-06,
+      "loss": 16.7142,
+      "step": 52010
+    },
+    {
+      "epoch": 0.9405999896031335,
+      "grad_norm": 47.90625,
+      "learning_rate": 9.85303150075688e-06,
+      "loss": 16.6547,
+      "step": 52020
+    },
+    {
+      "epoch": 0.9407808046722614,
+      "grad_norm": 44.53125,
+      "learning_rate": 9.85300324845022e-06,
+      "loss": 17.2,
+      "step": 52030
+    },
+    {
+      "epoch": 0.9409616197413893,
+      "grad_norm": 46.03125,
+      "learning_rate": 9.852974996143561e-06,
+      "loss": 16.5735,
+      "step": 52040
+    },
+    {
+      "epoch": 0.9411424348105171,
+      "grad_norm": 46.0625,
+      "learning_rate": 9.852946743836902e-06,
+      "loss": 16.9435,
+      "step": 52050
+    },
+    {
+      "epoch": 0.941323249879645,
+      "grad_norm": 45.84375,
+      "learning_rate": 9.852918491530242e-06,
+      "loss": 16.5768,
+      "step": 52060
+    },
+    {
+      "epoch": 0.9415040649487728,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.852890239223583e-06,
+      "loss": 16.8057,
+      "step": 52070
+    },
+    {
+      "epoch": 0.9416848800179007,
+      "grad_norm": 44.96875,
+      "learning_rate": 9.852861986916923e-06,
+      "loss": 16.56,
+      "step": 52080
+    },
+    {
+      "epoch": 0.9418656950870286,
+      "grad_norm": 43.84375,
+      "learning_rate": 9.852833734610262e-06,
+      "loss": 17.1698,
+      "step": 52090
+    },
+    {
+      "epoch": 0.9420465101561564,
+      "grad_norm": 45.8125,
+      "learning_rate": 9.852805482303603e-06,
+      "loss": 16.6837,
+      "step": 52100
+    },
+    {
+      "epoch": 0.9422273252252843,
+      "grad_norm": 45.78125,
+      "learning_rate": 9.852777229996944e-06,
+      "loss": 16.8276,
+      "step": 52110
+    },
+    {
+      "epoch": 0.9424081402944121,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.852748977690284e-06,
+      "loss": 16.3848,
+      "step": 52120
+    },
+    {
+      "epoch": 0.94258895536354,
+      "grad_norm": 43.28125,
+      "learning_rate": 9.852720725383625e-06,
+      "loss": 16.8509,
+      "step": 52130
+    },
+    {
+      "epoch": 0.9427697704326679,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.852692473076965e-06,
+      "loss": 16.6207,
+      "step": 52140
+    },
+    {
+      "epoch": 0.9429505855017957,
+      "grad_norm": 44.53125,
+      "learning_rate": 9.852664220770306e-06,
+      "loss": 16.405,
+      "step": 52150
+    },
+    {
+      "epoch": 0.9431314005709236,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.852635968463647e-06,
+      "loss": 16.8411,
+      "step": 52160
+    },
+    {
+      "epoch": 0.9433122156400514,
+      "grad_norm": 45.96875,
+      "learning_rate": 9.852607716156987e-06,
+      "loss": 16.647,
+      "step": 52170
+    },
+    {
+      "epoch": 0.9434930307091793,
+      "grad_norm": 47.15625,
+      "learning_rate": 9.852579463850326e-06,
+      "loss": 16.826,
+      "step": 52180
+    },
+    {
+      "epoch": 0.9436738457783072,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.852551211543667e-06,
+      "loss": 16.9083,
+      "step": 52190
+    },
+    {
+      "epoch": 0.9438546608474351,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.852522959237008e-06,
+      "loss": 17.141,
+      "step": 52200
+    },
+    {
+      "epoch": 0.9440354759165629,
+      "grad_norm": 42.875,
+      "learning_rate": 9.852494706930348e-06,
+      "loss": 16.7258,
+      "step": 52210
+    },
+    {
+      "epoch": 0.9442162909856907,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.852466454623689e-06,
+      "loss": 16.6493,
+      "step": 52220
+    },
+    {
+      "epoch": 0.9443971060548186,
+      "grad_norm": 43.375,
+      "learning_rate": 9.85243820231703e-06,
+      "loss": 16.6304,
+      "step": 52230
+    },
+    {
+      "epoch": 0.9445779211239465,
+      "grad_norm": 45.25,
+      "learning_rate": 9.85240995001037e-06,
+      "loss": 16.5853,
+      "step": 52240
+    },
+    {
+      "epoch": 0.9447587361930744,
+      "grad_norm": 43.625,
+      "learning_rate": 9.85238169770371e-06,
+      "loss": 17.031,
+      "step": 52250
+    },
+    {
+      "epoch": 0.9449395512622022,
+      "grad_norm": 45.625,
+      "learning_rate": 9.852353445397051e-06,
+      "loss": 16.8603,
+      "step": 52260
+    },
+    {
+      "epoch": 0.94512036633133,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.85232519309039e-06,
+      "loss": 16.9095,
+      "step": 52270
+    },
+    {
+      "epoch": 0.9453011814004579,
+      "grad_norm": 43.53125,
+      "learning_rate": 9.85229694078373e-06,
+      "loss": 16.8246,
+      "step": 52280
+    },
+    {
+      "epoch": 0.9454819964695858,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.852268688477071e-06,
+      "loss": 16.8132,
+      "step": 52290
+    },
+    {
+      "epoch": 0.9456628115387137,
+      "grad_norm": 48.875,
+      "learning_rate": 9.852240436170412e-06,
+      "loss": 16.8849,
+      "step": 52300
+    },
+    {
+      "epoch": 0.9458436266078415,
+      "grad_norm": 44.75,
+      "learning_rate": 9.852212183863753e-06,
+      "loss": 16.3753,
+      "step": 52310
+    },
+    {
+      "epoch": 0.9460244416769693,
+      "grad_norm": 42.59375,
+      "learning_rate": 9.852183931557093e-06,
+      "loss": 16.4101,
+      "step": 52320
+    },
+    {
+      "epoch": 0.9462052567460972,
+      "grad_norm": 47.6875,
+      "learning_rate": 9.852155679250434e-06,
+      "loss": 16.9043,
+      "step": 52330
+    },
+    {
+      "epoch": 0.946386071815225,
+      "grad_norm": 45.78125,
+      "learning_rate": 9.852127426943775e-06,
+      "loss": 16.5976,
+      "step": 52340
+    },
+    {
+      "epoch": 0.946566886884353,
+      "grad_norm": 45.0625,
+      "learning_rate": 9.852099174637113e-06,
+      "loss": 16.648,
+      "step": 52350
+    },
+    {
+      "epoch": 0.9467477019534808,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.852070922330454e-06,
+      "loss": 16.923,
+      "step": 52360
+    },
+    {
+      "epoch": 0.9469285170226087,
+      "grad_norm": 49.625,
+      "learning_rate": 9.852042670023795e-06,
+      "loss": 17.094,
+      "step": 52370
+    },
+    {
+      "epoch": 0.9471093320917365,
+      "grad_norm": 45.75,
+      "learning_rate": 9.852014417717135e-06,
+      "loss": 17.2729,
+      "step": 52380
+    },
+    {
+      "epoch": 0.9472901471608643,
+      "grad_norm": 44.53125,
+      "learning_rate": 9.851986165410476e-06,
+      "loss": 16.6978,
+      "step": 52390
+    },
+    {
+      "epoch": 0.9474709622299923,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.851957913103815e-06,
+      "loss": 16.9399,
+      "step": 52400
+    },
+    {
+      "epoch": 0.9476517772991201,
+      "grad_norm": 44.8125,
+      "learning_rate": 9.851929660797157e-06,
+      "loss": 17.1842,
+      "step": 52410
+    },
+    {
+      "epoch": 0.947832592368248,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.851901408490498e-06,
+      "loss": 17.0201,
+      "step": 52420
+    },
+    {
+      "epoch": 0.9480134074373758,
+      "grad_norm": 46.3125,
+      "learning_rate": 9.851873156183838e-06,
+      "loss": 16.8392,
+      "step": 52430
+    },
+    {
+      "epoch": 0.9481942225065036,
+      "grad_norm": 44.6875,
+      "learning_rate": 9.851844903877177e-06,
+      "loss": 16.222,
+      "step": 52440
+    },
+    {
+      "epoch": 0.9483750375756316,
+      "grad_norm": 46.53125,
+      "learning_rate": 9.851816651570518e-06,
+      "loss": 16.7396,
+      "step": 52450
+    },
+    {
+      "epoch": 0.9485558526447594,
+      "grad_norm": 44.375,
+      "learning_rate": 9.851788399263859e-06,
+      "loss": 16.944,
+      "step": 52460
+    },
+    {
+      "epoch": 0.9487366677138873,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.8517601469572e-06,
+      "loss": 16.3753,
+      "step": 52470
+    },
+    {
+      "epoch": 0.9489174827830151,
+      "grad_norm": 43.625,
+      "learning_rate": 9.85173189465054e-06,
+      "loss": 16.7334,
+      "step": 52480
+    },
+    {
+      "epoch": 0.949098297852143,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.851703642343879e-06,
+      "loss": 16.5398,
+      "step": 52490
+    },
+    {
+      "epoch": 0.9492791129212709,
+      "grad_norm": 45.28125,
+      "learning_rate": 9.851675390037221e-06,
+      "loss": 16.5976,
+      "step": 52500
+    },
+    {
+      "epoch": 0.9494599279903987,
+      "grad_norm": 42.5,
+      "learning_rate": 9.851647137730562e-06,
+      "loss": 16.487,
+      "step": 52510
+    },
+    {
+      "epoch": 0.9496407430595266,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.8516188854239e-06,
+      "loss": 17.0914,
+      "step": 52520
+    },
+    {
+      "epoch": 0.9498215581286544,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.851590633117241e-06,
+      "loss": 16.472,
+      "step": 52530
+    },
+    {
+      "epoch": 0.9500023731977824,
+      "grad_norm": 46.90625,
+      "learning_rate": 9.851562380810582e-06,
+      "loss": 16.7584,
+      "step": 52540
+    },
+    {
+      "epoch": 0.9501831882669102,
+      "grad_norm": 44.75,
+      "learning_rate": 9.851534128503923e-06,
+      "loss": 16.8644,
+      "step": 52550
+    },
+    {
+      "epoch": 0.950364003336038,
+      "grad_norm": 43.875,
+      "learning_rate": 9.851505876197263e-06,
+      "loss": 16.5356,
+      "step": 52560
+    },
+    {
+      "epoch": 0.9505448184051659,
+      "grad_norm": 43.875,
+      "learning_rate": 9.851477623890604e-06,
+      "loss": 16.2732,
+      "step": 52570
+    },
+    {
+      "epoch": 0.9507256334742937,
+      "grad_norm": 44.84375,
+      "learning_rate": 9.851449371583944e-06,
+      "loss": 16.608,
+      "step": 52580
+    },
+    {
+      "epoch": 0.9509064485434217,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.851421119277285e-06,
+      "loss": 16.5287,
+      "step": 52590
+    },
+    {
+      "epoch": 0.9510872636125495,
+      "grad_norm": 45.71875,
+      "learning_rate": 9.851392866970626e-06,
+      "loss": 17.0554,
+      "step": 52600
+    },
+    {
+      "epoch": 0.9512680786816773,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.851364614663965e-06,
+      "loss": 16.7669,
+      "step": 52610
+    },
+    {
+      "epoch": 0.9514488937508052,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.851336362357305e-06,
+      "loss": 16.934,
+      "step": 52620
+    },
+    {
+      "epoch": 0.951629708819933,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.851308110050646e-06,
+      "loss": 15.9467,
+      "step": 52630
+    },
+    {
+      "epoch": 0.951810523889061,
+      "grad_norm": 45.28125,
+      "learning_rate": 9.851279857743986e-06,
+      "loss": 16.2832,
+      "step": 52640
+    },
+    {
+      "epoch": 0.9519913389581888,
+      "grad_norm": 45.375,
+      "learning_rate": 9.851251605437327e-06,
+      "loss": 16.6556,
+      "step": 52650
+    },
+    {
+      "epoch": 0.9521721540273166,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.851223353130666e-06,
+      "loss": 16.8088,
+      "step": 52660
+    },
+    {
+      "epoch": 0.9523529690964445,
+      "grad_norm": 46.03125,
+      "learning_rate": 9.851195100824008e-06,
+      "loss": 16.5832,
+      "step": 52670
+    },
+    {
+      "epoch": 0.9525337841655723,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.851166848517349e-06,
+      "loss": 17.0086,
+      "step": 52680
+    },
+    {
+      "epoch": 0.9527145992347003,
+      "grad_norm": 45.71875,
+      "learning_rate": 9.851138596210688e-06,
+      "loss": 16.6411,
+      "step": 52690
+    },
+    {
+      "epoch": 0.9528954143038281,
+      "grad_norm": 46.34375,
+      "learning_rate": 9.851110343904028e-06,
+      "loss": 16.5188,
+      "step": 52700
+    },
+    {
+      "epoch": 0.953076229372956,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.851082091597369e-06,
+      "loss": 16.6827,
+      "step": 52710
+    },
+    {
+      "epoch": 0.9532570444420838,
+      "grad_norm": 46.4375,
+      "learning_rate": 9.85105383929071e-06,
+      "loss": 16.5534,
+      "step": 52720
+    },
+    {
+      "epoch": 0.9534378595112116,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.85102558698405e-06,
+      "loss": 16.7903,
+      "step": 52730
+    },
+    {
+      "epoch": 0.9536186745803396,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.850997334677391e-06,
+      "loss": 16.7192,
+      "step": 52740
+    },
+    {
+      "epoch": 0.9537994896494674,
+      "grad_norm": 45.84375,
+      "learning_rate": 9.85096908237073e-06,
+      "loss": 16.6573,
+      "step": 52750
+    },
+    {
+      "epoch": 0.9539803047185953,
+      "grad_norm": 45.625,
+      "learning_rate": 9.850940830064072e-06,
+      "loss": 16.2963,
+      "step": 52760
+    },
+    {
+      "epoch": 0.9541611197877231,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.850912577757413e-06,
+      "loss": 16.2455,
+      "step": 52770
+    },
+    {
+      "epoch": 0.9543419348568509,
+      "grad_norm": 48.15625,
+      "learning_rate": 9.850884325450752e-06,
+      "loss": 16.774,
+      "step": 52780
+    },
+    {
+      "epoch": 0.9545227499259789,
+      "grad_norm": 42.40625,
+      "learning_rate": 9.850856073144092e-06,
+      "loss": 16.7267,
+      "step": 52790
+    },
+    {
+      "epoch": 0.9547035649951067,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.850827820837433e-06,
+      "loss": 16.56,
+      "step": 52800
+    },
+    {
+      "epoch": 0.9548843800642346,
+      "grad_norm": 41.96875,
+      "learning_rate": 9.850799568530774e-06,
+      "loss": 17.191,
+      "step": 52810
+    },
+    {
+      "epoch": 0.9550651951333624,
+      "grad_norm": 43.75,
+      "learning_rate": 9.850771316224114e-06,
+      "loss": 16.2504,
+      "step": 52820
+    },
+    {
+      "epoch": 0.9552460102024902,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.850743063917453e-06,
+      "loss": 16.4462,
+      "step": 52830
+    },
+    {
+      "epoch": 0.9554268252716182,
+      "grad_norm": 44.25,
+      "learning_rate": 9.850714811610794e-06,
+      "loss": 16.653,
+      "step": 52840
+    },
+    {
+      "epoch": 0.955607640340746,
+      "grad_norm": 41.15625,
+      "learning_rate": 9.850686559304136e-06,
+      "loss": 16.4501,
+      "step": 52850
+    },
+    {
+      "epoch": 0.9557884554098739,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.850658306997477e-06,
+      "loss": 16.4311,
+      "step": 52860
+    },
+    {
+      "epoch": 0.9559692704790017,
+      "grad_norm": 42.5,
+      "learning_rate": 9.850630054690816e-06,
+      "loss": 16.4035,
+      "step": 52870
+    },
+    {
+      "epoch": 0.9561500855481296,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.850601802384156e-06,
+      "loss": 16.8654,
+      "step": 52880
+    },
+    {
+      "epoch": 0.9563309006172575,
+      "grad_norm": 43.125,
+      "learning_rate": 9.850573550077497e-06,
+      "loss": 16.8872,
+      "step": 52890
+    },
+    {
+      "epoch": 0.9565117156863853,
+      "grad_norm": 45.59375,
+      "learning_rate": 9.850545297770838e-06,
+      "loss": 17.3379,
+      "step": 52900
+    },
+    {
+      "epoch": 0.9566925307555132,
+      "grad_norm": 47.5,
+      "learning_rate": 9.850517045464178e-06,
+      "loss": 16.7555,
+      "step": 52910
+    },
+    {
+      "epoch": 0.956873345824641,
+      "grad_norm": 47.375,
+      "learning_rate": 9.850488793157517e-06,
+      "loss": 16.522,
+      "step": 52920
+    },
+    {
+      "epoch": 0.9570541608937689,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.85046054085086e-06,
+      "loss": 16.7725,
+      "step": 52930
+    },
+    {
+      "epoch": 0.9572349759628968,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.8504322885442e-06,
+      "loss": 16.493,
+      "step": 52940
+    },
+    {
+      "epoch": 0.9574157910320246,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.850404036237539e-06,
+      "loss": 16.8442,
+      "step": 52950
+    },
+    {
+      "epoch": 0.9575966061011525,
+      "grad_norm": 47.34375,
+      "learning_rate": 9.85037578393088e-06,
+      "loss": 16.7463,
+      "step": 52960
+    },
+    {
+      "epoch": 0.9577774211702803,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.85034753162422e-06,
+      "loss": 16.7918,
+      "step": 52970
+    },
+    {
+      "epoch": 0.9579582362394082,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.85031927931756e-06,
+      "loss": 16.6675,
+      "step": 52980
+    },
+    {
+      "epoch": 0.958139051308536,
+      "grad_norm": 40.71875,
+      "learning_rate": 9.850291027010901e-06,
+      "loss": 16.662,
+      "step": 52990
+    },
+    {
+      "epoch": 0.9583198663776639,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.85026277470424e-06,
+      "loss": 16.813,
+      "step": 53000
+    },
+    {
+      "epoch": 0.9585006814467918,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.850234522397581e-06,
+      "loss": 16.8846,
+      "step": 53010
+    },
+    {
+      "epoch": 0.9586814965159196,
+      "grad_norm": 45.625,
+      "learning_rate": 9.850206270090923e-06,
+      "loss": 17.0815,
+      "step": 53020
+    },
+    {
+      "epoch": 0.9588623115850475,
+      "grad_norm": 47.40625,
+      "learning_rate": 9.850178017784264e-06,
+      "loss": 16.2415,
+      "step": 53030
+    },
+    {
+      "epoch": 0.9590431266541753,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.850149765477603e-06,
+      "loss": 16.5826,
+      "step": 53040
+    },
+    {
+      "epoch": 0.9592239417233033,
+      "grad_norm": 45.4375,
+      "learning_rate": 9.850121513170943e-06,
+      "loss": 16.5556,
+      "step": 53050
+    },
+    {
+      "epoch": 0.9594047567924311,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.850093260864284e-06,
+      "loss": 16.7386,
+      "step": 53060
+    },
+    {
+      "epoch": 0.9595855718615589,
+      "grad_norm": 41.21875,
+      "learning_rate": 9.850065008557625e-06,
+      "loss": 16.5449,
+      "step": 53070
+    },
+    {
+      "epoch": 0.9597663869306868,
+      "grad_norm": 51.65625,
+      "learning_rate": 9.850036756250965e-06,
+      "loss": 16.669,
+      "step": 53080
+    },
+    {
+      "epoch": 0.9599472019998146,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.850008503944304e-06,
+      "loss": 16.748,
+      "step": 53090
+    },
+    {
+      "epoch": 0.9601280170689426,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.849980251637645e-06,
+      "loss": 17.1109,
+      "step": 53100
+    },
+    {
+      "epoch": 0.9603088321380704,
+      "grad_norm": 40.03125,
+      "learning_rate": 9.849951999330987e-06,
+      "loss": 16.6088,
+      "step": 53110
+    },
+    {
+      "epoch": 0.9604896472071982,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.849923747024326e-06,
+      "loss": 16.7283,
+      "step": 53120
+    },
+    {
+      "epoch": 0.9606704622763261,
+      "grad_norm": 46.03125,
+      "learning_rate": 9.849895494717667e-06,
+      "loss": 16.4558,
+      "step": 53130
+    },
+    {
+      "epoch": 0.960851277345454,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.849867242411007e-06,
+      "loss": 16.6179,
+      "step": 53140
+    },
+    {
+      "epoch": 0.9610320924145819,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.849838990104348e-06,
+      "loss": 16.864,
+      "step": 53150
+    },
+    {
+      "epoch": 0.9612129074837097,
+      "grad_norm": 43.71875,
+      "learning_rate": 9.849810737797689e-06,
+      "loss": 16.7878,
+      "step": 53160
+    },
+    {
+      "epoch": 0.9613937225528375,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.84978248549103e-06,
+      "loss": 16.676,
+      "step": 53170
+    },
+    {
+      "epoch": 0.9615745376219654,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.849754233184368e-06,
+      "loss": 16.5465,
+      "step": 53180
+    },
+    {
+      "epoch": 0.9617553526910932,
+      "grad_norm": 45.0,
+      "learning_rate": 9.849725980877709e-06,
+      "loss": 16.7175,
+      "step": 53190
+    },
+    {
+      "epoch": 0.9619361677602212,
+      "grad_norm": 44.1875,
+      "learning_rate": 9.849697728571051e-06,
+      "loss": 16.9195,
+      "step": 53200
+    },
+    {
+      "epoch": 0.962116982829349,
+      "grad_norm": 46.0625,
+      "learning_rate": 9.84966947626439e-06,
+      "loss": 17.0238,
+      "step": 53210
+    },
+    {
+      "epoch": 0.9622977978984769,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.84964122395773e-06,
+      "loss": 16.6907,
+      "step": 53220
+    },
+    {
+      "epoch": 0.9624786129676047,
+      "grad_norm": 48.65625,
+      "learning_rate": 9.849612971651071e-06,
+      "loss": 16.7542,
+      "step": 53230
+    },
+    {
+      "epoch": 0.9626594280367325,
+      "grad_norm": 45.90625,
+      "learning_rate": 9.849584719344412e-06,
+      "loss": 16.4365,
+      "step": 53240
+    },
+    {
+      "epoch": 0.9628402431058605,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.849556467037753e-06,
+      "loss": 16.5799,
+      "step": 53250
+    },
+    {
+      "epoch": 0.9630210581749883,
+      "grad_norm": 42.71875,
+      "learning_rate": 9.849528214731091e-06,
+      "loss": 17.1928,
+      "step": 53260
+    },
+    {
+      "epoch": 0.9632018732441162,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.849499962424432e-06,
+      "loss": 16.5996,
+      "step": 53270
+    },
+    {
+      "epoch": 0.963382688313244,
+      "grad_norm": 44.875,
+      "learning_rate": 9.849471710117774e-06,
+      "loss": 16.595,
+      "step": 53280
+    },
+    {
+      "epoch": 0.9635635033823718,
+      "grad_norm": 42.9375,
+      "learning_rate": 9.849443457811115e-06,
+      "loss": 16.4589,
+      "step": 53290
+    },
+    {
+      "epoch": 0.9637443184514998,
+      "grad_norm": 45.0,
+      "learning_rate": 9.849415205504454e-06,
+      "loss": 16.8769,
+      "step": 53300
+    },
+    {
+      "epoch": 0.9639251335206276,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.849386953197795e-06,
+      "loss": 16.4602,
+      "step": 53310
+    },
+    {
+      "epoch": 0.9641059485897555,
+      "grad_norm": 39.125,
+      "learning_rate": 9.849358700891135e-06,
+      "loss": 17.0599,
+      "step": 53320
+    },
+    {
+      "epoch": 0.9642867636588833,
+      "grad_norm": 46.09375,
+      "learning_rate": 9.849330448584476e-06,
+      "loss": 16.9556,
+      "step": 53330
+    },
+    {
+      "epoch": 0.9644675787280111,
+      "grad_norm": 45.875,
+      "learning_rate": 9.849302196277816e-06,
+      "loss": 16.5521,
+      "step": 53340
+    },
+    {
+      "epoch": 0.9646483937971391,
+      "grad_norm": 46.15625,
+      "learning_rate": 9.849273943971155e-06,
+      "loss": 16.6604,
+      "step": 53350
+    },
+    {
+      "epoch": 0.9648292088662669,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.849245691664496e-06,
+      "loss": 16.7637,
+      "step": 53360
+    },
+    {
+      "epoch": 0.9650100239353948,
+      "grad_norm": 42.5,
+      "learning_rate": 9.849217439357838e-06,
+      "loss": 16.5801,
+      "step": 53370
+    },
+    {
+      "epoch": 0.9651908390045226,
+      "grad_norm": 45.625,
+      "learning_rate": 9.849189187051177e-06,
+      "loss": 16.5078,
+      "step": 53380
+    },
+    {
+      "epoch": 0.9653716540736504,
+      "grad_norm": 46.28125,
+      "learning_rate": 9.849160934744518e-06,
+      "loss": 16.5973,
+      "step": 53390
+    },
+    {
+      "epoch": 0.9655524691427784,
+      "grad_norm": 47.65625,
+      "learning_rate": 9.849132682437858e-06,
+      "loss": 17.1258,
+      "step": 53400
+    },
+    {
+      "epoch": 0.9657332842119062,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.849104430131199e-06,
+      "loss": 16.6324,
+      "step": 53410
+    },
+    {
+      "epoch": 0.9659140992810341,
+      "grad_norm": 42.90625,
+      "learning_rate": 9.84907617782454e-06,
+      "loss": 16.9065,
+      "step": 53420
+    },
+    {
+      "epoch": 0.9660949143501619,
+      "grad_norm": 44.46875,
+      "learning_rate": 9.849047925517879e-06,
+      "loss": 16.5174,
+      "step": 53430
+    },
+    {
+      "epoch": 0.9662757294192899,
+      "grad_norm": 45.03125,
+      "learning_rate": 9.84901967321122e-06,
+      "loss": 16.9315,
+      "step": 53440
+    },
+    {
+      "epoch": 0.9664565444884177,
+      "grad_norm": 49.90625,
+      "learning_rate": 9.84899142090456e-06,
+      "loss": 16.6896,
+      "step": 53450
+    },
+    {
+      "epoch": 0.9666373595575455,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.848963168597902e-06,
+      "loss": 16.5972,
+      "step": 53460
+    },
+    {
+      "epoch": 0.9668181746266734,
+      "grad_norm": 45.21875,
+      "learning_rate": 9.848934916291241e-06,
+      "loss": 16.7322,
+      "step": 53470
+    },
+    {
+      "epoch": 0.9669989896958012,
+      "grad_norm": 45.25,
+      "learning_rate": 9.848906663984582e-06,
+      "loss": 16.9706,
+      "step": 53480
+    },
+    {
+      "epoch": 0.9671798047649292,
+      "grad_norm": 46.5625,
+      "learning_rate": 9.848878411677922e-06,
+      "loss": 16.8125,
+      "step": 53490
+    },
+    {
+      "epoch": 0.967360619834057,
+      "grad_norm": 45.75,
+      "learning_rate": 9.848850159371263e-06,
+      "loss": 16.3223,
+      "step": 53500
+    },
+    {
+      "epoch": 0.9675414349031848,
+      "grad_norm": 40.0,
+      "learning_rate": 9.848821907064604e-06,
+      "loss": 16.235,
+      "step": 53510
+    },
+    {
+      "epoch": 0.9677222499723127,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.848793654757943e-06,
+      "loss": 16.5035,
+      "step": 53520
+    },
+    {
+      "epoch": 0.9679030650414405,
+      "grad_norm": 46.15625,
+      "learning_rate": 9.848765402451283e-06,
+      "loss": 16.6695,
+      "step": 53530
+    },
+    {
+      "epoch": 0.9680838801105685,
+      "grad_norm": 45.46875,
+      "learning_rate": 9.848737150144624e-06,
+      "loss": 16.551,
+      "step": 53540
+    },
+    {
+      "epoch": 0.9682646951796963,
+      "grad_norm": 47.21875,
+      "learning_rate": 9.848708897837964e-06,
+      "loss": 16.9247,
+      "step": 53550
+    },
+    {
+      "epoch": 0.9684455102488241,
+      "grad_norm": 44.34375,
+      "learning_rate": 9.848680645531305e-06,
+      "loss": 16.7716,
+      "step": 53560
+    },
+    {
+      "epoch": 0.968626325317952,
+      "grad_norm": 42.125,
+      "learning_rate": 9.848652393224646e-06,
+      "loss": 17.0277,
+      "step": 53570
+    },
+    {
+      "epoch": 0.9688071403870798,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.848624140917986e-06,
+      "loss": 16.7511,
+      "step": 53580
+    },
+    {
+      "epoch": 0.9689879554562078,
+      "grad_norm": 46.75,
+      "learning_rate": 9.848595888611327e-06,
+      "loss": 16.8811,
+      "step": 53590
+    },
+    {
+      "epoch": 0.9691687705253356,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.848567636304668e-06,
+      "loss": 16.6817,
+      "step": 53600
+    },
+    {
+      "epoch": 0.9693495855944635,
+      "grad_norm": 46.46875,
+      "learning_rate": 9.848539383998006e-06,
+      "loss": 16.6221,
+      "step": 53610
+    },
+    {
+      "epoch": 0.9695304006635913,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.848511131691347e-06,
+      "loss": 17.0512,
+      "step": 53620
+    },
+    {
+      "epoch": 0.9697112157327191,
+      "grad_norm": 40.8125,
+      "learning_rate": 9.84848287938469e-06,
+      "loss": 16.2973,
+      "step": 53630
+    },
+    {
+      "epoch": 0.969892030801847,
+      "grad_norm": 44.71875,
+      "learning_rate": 9.848454627078028e-06,
+      "loss": 16.8274,
+      "step": 53640
+    },
+    {
+      "epoch": 0.9700728458709749,
+      "grad_norm": 46.03125,
+      "learning_rate": 9.848426374771369e-06,
+      "loss": 16.3493,
+      "step": 53650
+    },
+    {
+      "epoch": 0.9702536609401028,
+      "grad_norm": 45.1875,
+      "learning_rate": 9.84839812246471e-06,
+      "loss": 16.2608,
+      "step": 53660
+    },
+    {
+      "epoch": 0.9704344760092306,
+      "grad_norm": 41.5625,
+      "learning_rate": 9.84836987015805e-06,
+      "loss": 16.3783,
+      "step": 53670
+    },
+    {
+      "epoch": 0.9706152910783584,
+      "grad_norm": 44.75,
+      "learning_rate": 9.84834161785139e-06,
+      "loss": 16.6116,
+      "step": 53680
+    },
+    {
+      "epoch": 0.9707961061474863,
+      "grad_norm": 43.5,
+      "learning_rate": 9.84831336554473e-06,
+      "loss": 16.6586,
+      "step": 53690
+    },
+    {
+      "epoch": 0.9709769212166142,
+      "grad_norm": 44.71875,
+      "learning_rate": 9.84828511323807e-06,
+      "loss": 16.7844,
+      "step": 53700
+    },
+    {
+      "epoch": 0.9711577362857421,
+      "grad_norm": 45.34375,
+      "learning_rate": 9.848256860931411e-06,
+      "loss": 16.8105,
+      "step": 53710
+    },
+    {
+      "epoch": 0.9713385513548699,
+      "grad_norm": 45.25,
+      "learning_rate": 9.848228608624753e-06,
+      "loss": 16.7695,
+      "step": 53720
+    },
+    {
+      "epoch": 0.9715193664239977,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.848200356318092e-06,
+      "loss": 16.7423,
+      "step": 53730
+    },
+    {
+      "epoch": 0.9717001814931256,
+      "grad_norm": 48.625,
+      "learning_rate": 9.848172104011433e-06,
+      "loss": 16.6176,
+      "step": 53740
+    },
+    {
+      "epoch": 0.9718809965622535,
+      "grad_norm": 47.3125,
+      "learning_rate": 9.848143851704773e-06,
+      "loss": 16.8514,
+      "step": 53750
+    },
+    {
+      "epoch": 0.9720618116313814,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.848115599398114e-06,
+      "loss": 16.7589,
+      "step": 53760
+    },
+    {
+      "epoch": 0.9722426267005092,
+      "grad_norm": 47.03125,
+      "learning_rate": 9.848087347091455e-06,
+      "loss": 16.797,
+      "step": 53770
+    },
+    {
+      "epoch": 0.9724234417696371,
+      "grad_norm": 45.34375,
+      "learning_rate": 9.848059094784794e-06,
+      "loss": 16.6639,
+      "step": 53780
+    },
+    {
+      "epoch": 0.972604256838765,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.848030842478134e-06,
+      "loss": 16.494,
+      "step": 53790
+    },
+    {
+      "epoch": 0.9727850719078928,
+      "grad_norm": 44.75,
+      "learning_rate": 9.848002590171475e-06,
+      "loss": 16.8343,
+      "step": 53800
+    },
+    {
+      "epoch": 0.9729658869770207,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.847974337864816e-06,
+      "loss": 16.5076,
+      "step": 53810
+    },
+    {
+      "epoch": 0.9731467020461485,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.847946085558156e-06,
+      "loss": 16.5628,
+      "step": 53820
+    },
+    {
+      "epoch": 0.9733275171152764,
+      "grad_norm": 42.625,
+      "learning_rate": 9.847917833251497e-06,
+      "loss": 16.3621,
+      "step": 53830
+    },
+    {
+      "epoch": 0.9735083321844042,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.847889580944837e-06,
+      "loss": 16.68,
+      "step": 53840
+    },
+    {
+      "epoch": 0.9736891472535321,
+      "grad_norm": 42.75,
+      "learning_rate": 9.847861328638178e-06,
+      "loss": 16.4268,
+      "step": 53850
+    },
+    {
+      "epoch": 0.97386996232266,
+      "grad_norm": 46.90625,
+      "learning_rate": 9.847833076331517e-06,
+      "loss": 16.3298,
+      "step": 53860
+    },
+    {
+      "epoch": 0.9740507773917878,
+      "grad_norm": 46.625,
+      "learning_rate": 9.847804824024858e-06,
+      "loss": 16.6696,
+      "step": 53870
+    },
+    {
+      "epoch": 0.9742315924609157,
+      "grad_norm": 46.3125,
+      "learning_rate": 9.847776571718198e-06,
+      "loss": 16.8559,
+      "step": 53880
+    },
+    {
+      "epoch": 0.9744124075300435,
+      "grad_norm": 40.90625,
+      "learning_rate": 9.847748319411539e-06,
+      "loss": 16.8465,
+      "step": 53890
+    },
+    {
+      "epoch": 0.9745932225991714,
+      "grad_norm": 45.03125,
+      "learning_rate": 9.84772006710488e-06,
+      "loss": 16.295,
+      "step": 53900
+    },
+    {
+      "epoch": 0.9747740376682993,
+      "grad_norm": 44.125,
+      "learning_rate": 9.84769181479822e-06,
+      "loss": 16.9551,
+      "step": 53910
+    },
+    {
+      "epoch": 0.9749548527374271,
+      "grad_norm": 44.4375,
+      "learning_rate": 9.84766356249156e-06,
+      "loss": 16.4401,
+      "step": 53920
+    },
+    {
+      "epoch": 0.975135667806555,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.847635310184901e-06,
+      "loss": 16.5902,
+      "step": 53930
+    },
+    {
+      "epoch": 0.9753164828756828,
+      "grad_norm": 45.71875,
+      "learning_rate": 9.847607057878242e-06,
+      "loss": 16.6108,
+      "step": 53940
+    },
+    {
+      "epoch": 0.9754972979448108,
+      "grad_norm": 43.875,
+      "learning_rate": 9.84757880557158e-06,
+      "loss": 16.9275,
+      "step": 53950
+    },
+    {
+      "epoch": 0.9756781130139386,
+      "grad_norm": 45.0,
+      "learning_rate": 9.847550553264921e-06,
+      "loss": 16.6905,
+      "step": 53960
+    },
+    {
+      "epoch": 0.9758589280830664,
+      "grad_norm": 43.75,
+      "learning_rate": 9.847522300958262e-06,
+      "loss": 16.5468,
+      "step": 53970
+    },
+    {
+      "epoch": 0.9760397431521943,
+      "grad_norm": 44.0625,
+      "learning_rate": 9.847494048651603e-06,
+      "loss": 16.9278,
+      "step": 53980
+    },
+    {
+      "epoch": 0.9762205582213221,
+      "grad_norm": 43.8125,
+      "learning_rate": 9.847465796344943e-06,
+      "loss": 16.379,
+      "step": 53990
+    },
+    {
+      "epoch": 0.9764013732904501,
+      "grad_norm": 43.40625,
+      "learning_rate": 9.847437544038284e-06,
+      "loss": 16.4727,
+      "step": 54000
+    },
+    {
+      "epoch": 0.9765821883595779,
+      "grad_norm": 45.3125,
+      "learning_rate": 9.847409291731625e-06,
+      "loss": 16.854,
+      "step": 54010
+    },
+    {
+      "epoch": 0.9767630034287057,
+      "grad_norm": 43.46875,
+      "learning_rate": 9.847381039424965e-06,
+      "loss": 16.6393,
+      "step": 54020
+    },
+    {
+      "epoch": 0.9769438184978336,
+      "grad_norm": 45.65625,
+      "learning_rate": 9.847352787118306e-06,
+      "loss": 16.786,
+      "step": 54030
+    },
+    {
+      "epoch": 0.9771246335669614,
+      "grad_norm": 43.09375,
+      "learning_rate": 9.847324534811645e-06,
+      "loss": 16.5585,
+      "step": 54040
+    },
+    {
+      "epoch": 0.9773054486360894,
+      "grad_norm": 42.46875,
+      "learning_rate": 9.847296282504985e-06,
+      "loss": 16.6101,
+      "step": 54050
+    },
+    {
+      "epoch": 0.9774862637052172,
+      "grad_norm": 45.53125,
+      "learning_rate": 9.847268030198326e-06,
+      "loss": 16.7073,
+      "step": 54060
+    },
+    {
+      "epoch": 0.977667078774345,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.847239777891667e-06,
+      "loss": 16.6391,
+      "step": 54070
+    },
+    {
+      "epoch": 0.9778478938434729,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.847211525585007e-06,
+      "loss": 16.8252,
+      "step": 54080
+    },
+    {
+      "epoch": 0.9780287089126007,
+      "grad_norm": 44.375,
+      "learning_rate": 9.847183273278348e-06,
+      "loss": 16.2046,
+      "step": 54090
+    },
+    {
+      "epoch": 0.9782095239817287,
+      "grad_norm": 45.4375,
+      "learning_rate": 9.847155020971688e-06,
+      "loss": 16.8368,
+      "step": 54100
+    },
+    {
+      "epoch": 0.9783903390508565,
+      "grad_norm": 41.84375,
+      "learning_rate": 9.847126768665029e-06,
+      "loss": 16.9655,
+      "step": 54110
+    },
+    {
+      "epoch": 0.9785711541199844,
+      "grad_norm": 42.34375,
+      "learning_rate": 9.847098516358368e-06,
+      "loss": 16.7057,
+      "step": 54120
+    },
+    {
+      "epoch": 0.9787519691891122,
+      "grad_norm": 47.28125,
+      "learning_rate": 9.847070264051709e-06,
+      "loss": 16.7744,
+      "step": 54130
+    },
+    {
+      "epoch": 0.97893278425824,
+      "grad_norm": 46.1875,
+      "learning_rate": 9.84704201174505e-06,
+      "loss": 16.8817,
+      "step": 54140
+    },
+    {
+      "epoch": 0.979113599327368,
+      "grad_norm": 48.625,
+      "learning_rate": 9.84701375943839e-06,
+      "loss": 16.5609,
+      "step": 54150
+    },
+    {
+      "epoch": 0.9792944143964958,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.84698550713173e-06,
+      "loss": 16.8131,
+      "step": 54160
+    },
+    {
+      "epoch": 0.9794752294656237,
+      "grad_norm": 42.09375,
+      "learning_rate": 9.846957254825071e-06,
+      "loss": 16.4902,
+      "step": 54170
+    },
+    {
+      "epoch": 0.9796560445347515,
+      "grad_norm": 43.0625,
+      "learning_rate": 9.846929002518412e-06,
+      "loss": 16.5643,
+      "step": 54180
+    },
+    {
+      "epoch": 0.9798368596038793,
+      "grad_norm": 46.84375,
+      "learning_rate": 9.846900750211752e-06,
+      "loss": 16.5155,
+      "step": 54190
+    },
+    {
+      "epoch": 0.9800176746730073,
+      "grad_norm": 45.34375,
+      "learning_rate": 9.846872497905093e-06,
+      "loss": 16.5785,
+      "step": 54200
+    },
+    {
+      "epoch": 0.9801984897421351,
+      "grad_norm": 46.4375,
+      "learning_rate": 9.846844245598432e-06,
+      "loss": 16.6293,
+      "step": 54210
+    },
+    {
+      "epoch": 0.980379304811263,
+      "grad_norm": 42.21875,
+      "learning_rate": 9.846815993291773e-06,
+      "loss": 16.4009,
+      "step": 54220
+    },
+    {
+      "epoch": 0.9805601198803908,
+      "grad_norm": 43.21875,
+      "learning_rate": 9.846787740985113e-06,
+      "loss": 16.6647,
+      "step": 54230
+    },
+    {
+      "epoch": 0.9807409349495186,
+      "grad_norm": 45.5,
+      "learning_rate": 9.846759488678454e-06,
+      "loss": 16.7975,
+      "step": 54240
+    },
+    {
+      "epoch": 0.9809217500186466,
+      "grad_norm": 43.875,
+      "learning_rate": 9.846731236371794e-06,
+      "loss": 16.7165,
+      "step": 54250
+    },
+    {
+      "epoch": 0.9811025650877744,
+      "grad_norm": 45.5625,
+      "learning_rate": 9.846702984065135e-06,
+      "loss": 16.5906,
+      "step": 54260
+    },
+    {
+      "epoch": 0.9812833801569023,
+      "grad_norm": 48.40625,
+      "learning_rate": 9.846674731758476e-06,
+      "loss": 16.5228,
+      "step": 54270
+    },
+    {
+      "epoch": 0.9814641952260301,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.846646479451816e-06,
+      "loss": 16.3795,
+      "step": 54280
+    },
+    {
+      "epoch": 0.981645010295158,
+      "grad_norm": 46.78125,
+      "learning_rate": 9.846618227145155e-06,
+      "loss": 17.0372,
+      "step": 54290
+    },
+    {
+      "epoch": 0.9818258253642859,
+      "grad_norm": 48.375,
+      "learning_rate": 9.846589974838496e-06,
+      "loss": 16.7747,
+      "step": 54300
+    },
+    {
+      "epoch": 0.9820066404334137,
+      "grad_norm": 43.125,
+      "learning_rate": 9.846561722531836e-06,
+      "loss": 16.8236,
+      "step": 54310
+    },
+    {
+      "epoch": 0.9821874555025416,
+      "grad_norm": 46.25,
+      "learning_rate": 9.846533470225177e-06,
+      "loss": 16.6047,
+      "step": 54320
+    },
+    {
+      "epoch": 0.9823682705716694,
+      "grad_norm": 47.9375,
+      "learning_rate": 9.846505217918518e-06,
+      "loss": 16.911,
+      "step": 54330
+    },
+    {
+      "epoch": 0.9825490856407973,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.846476965611858e-06,
+      "loss": 16.969,
+      "step": 54340
+    },
+    {
+      "epoch": 0.9827299007099252,
+      "grad_norm": 41.71875,
+      "learning_rate": 9.846448713305199e-06,
+      "loss": 16.8471,
+      "step": 54350
+    },
+    {
+      "epoch": 0.982910715779053,
+      "grad_norm": 45.5,
+      "learning_rate": 9.84642046099854e-06,
+      "loss": 16.3527,
+      "step": 54360
+    },
+    {
+      "epoch": 0.9830915308481809,
+      "grad_norm": 47.65625,
+      "learning_rate": 9.84639220869188e-06,
+      "loss": 16.3417,
+      "step": 54370
+    },
+    {
+      "epoch": 0.9832723459173087,
+      "grad_norm": 43.6875,
+      "learning_rate": 9.846363956385219e-06,
+      "loss": 16.4029,
+      "step": 54380
+    },
+    {
+      "epoch": 0.9834531609864366,
+      "grad_norm": 44.09375,
+      "learning_rate": 9.84633570407856e-06,
+      "loss": 16.9436,
+      "step": 54390
+    },
+    {
+      "epoch": 0.9836339760555645,
+      "grad_norm": 42.4375,
+      "learning_rate": 9.8463074517719e-06,
+      "loss": 16.6759,
+      "step": 54400
+    },
+    {
+      "epoch": 0.9838147911246923,
+      "grad_norm": 45.375,
+      "learning_rate": 9.846279199465241e-06,
+      "loss": 16.7103,
+      "step": 54410
+    },
+    {
+      "epoch": 0.9839956061938202,
+      "grad_norm": 42.5625,
+      "learning_rate": 9.846250947158582e-06,
+      "loss": 16.3442,
+      "step": 54420
+    },
+    {
+      "epoch": 0.984176421262948,
+      "grad_norm": 46.0625,
+      "learning_rate": 9.846222694851922e-06,
+      "loss": 16.9653,
+      "step": 54430
+    },
+    {
+      "epoch": 0.984357236332076,
+      "grad_norm": 47.03125,
+      "learning_rate": 9.846194442545263e-06,
+      "loss": 16.7699,
+      "step": 54440
+    },
+    {
+      "epoch": 0.9845380514012038,
+      "grad_norm": 46.34375,
+      "learning_rate": 9.846166190238603e-06,
+      "loss": 16.7369,
+      "step": 54450
+    },
+    {
+      "epoch": 0.9847188664703317,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.846137937931944e-06,
+      "loss": 17.2226,
+      "step": 54460
+    },
+    {
+      "epoch": 0.9848996815394595,
+      "grad_norm": 45.03125,
+      "learning_rate": 9.846109685625283e-06,
+      "loss": 16.9318,
+      "step": 54470
+    },
+    {
+      "epoch": 0.9850804966085873,
+      "grad_norm": 45.15625,
+      "learning_rate": 9.846081433318624e-06,
+      "loss": 16.7337,
+      "step": 54480
+    },
+    {
+      "epoch": 0.9852613116777152,
+      "grad_norm": 45.6875,
+      "learning_rate": 9.846053181011964e-06,
+      "loss": 17.0956,
+      "step": 54490
+    },
+    {
+      "epoch": 0.9854421267468431,
+      "grad_norm": 42.84375,
+      "learning_rate": 9.846024928705305e-06,
+      "loss": 16.4707,
+      "step": 54500
+    },
+    {
+      "epoch": 0.985622941815971,
+      "grad_norm": 41.4375,
+      "learning_rate": 9.845996676398646e-06,
+      "loss": 16.2423,
+      "step": 54510
+    },
+    {
+      "epoch": 0.9858037568850988,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.845968424091986e-06,
+      "loss": 16.9379,
+      "step": 54520
+    },
+    {
+      "epoch": 0.9859845719542266,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.845940171785327e-06,
+      "loss": 16.7772,
+      "step": 54530
+    },
+    {
+      "epoch": 0.9861653870233545,
+      "grad_norm": 43.125,
+      "learning_rate": 9.845911919478667e-06,
+      "loss": 16.7428,
+      "step": 54540
+    },
+    {
+      "epoch": 0.9863462020924824,
+      "grad_norm": 45.59375,
+      "learning_rate": 9.845883667172006e-06,
+      "loss": 16.4671,
+      "step": 54550
+    },
+    {
+      "epoch": 0.9865270171616103,
+      "grad_norm": 45.28125,
+      "learning_rate": 9.845855414865347e-06,
+      "loss": 16.8481,
+      "step": 54560
+    },
+    {
+      "epoch": 0.9867078322307381,
+      "grad_norm": 40.65625,
+      "learning_rate": 9.845827162558688e-06,
+      "loss": 16.5104,
+      "step": 54570
+    },
+    {
+      "epoch": 0.9868886472998659,
+      "grad_norm": 44.21875,
+      "learning_rate": 9.845798910252028e-06,
+      "loss": 16.8001,
+      "step": 54580
+    },
+    {
+      "epoch": 0.9870694623689938,
+      "grad_norm": 43.78125,
+      "learning_rate": 9.845770657945369e-06,
+      "loss": 16.502,
+      "step": 54590
+    },
+    {
+      "epoch": 0.9872502774381217,
+      "grad_norm": 44.59375,
+      "learning_rate": 9.84574240563871e-06,
+      "loss": 16.652,
+      "step": 54600
+    },
+    {
+      "epoch": 0.9874310925072496,
+      "grad_norm": 45.875,
+      "learning_rate": 9.84571415333205e-06,
+      "loss": 16.5849,
+      "step": 54610
+    },
+    {
+      "epoch": 0.9876119075763774,
+      "grad_norm": 46.46875,
+      "learning_rate": 9.84568590102539e-06,
+      "loss": 16.6386,
+      "step": 54620
+    },
+    {
+      "epoch": 0.9877927226455053,
+      "grad_norm": 48.0,
+      "learning_rate": 9.845657648718731e-06,
+      "loss": 16.7037,
+      "step": 54630
+    },
+    {
+      "epoch": 0.9879735377146331,
+      "grad_norm": 46.625,
+      "learning_rate": 9.84562939641207e-06,
+      "loss": 16.9638,
+      "step": 54640
+    },
+    {
+      "epoch": 0.988154352783761,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.845601144105411e-06,
+      "loss": 16.5736,
+      "step": 54650
+    },
+    {
+      "epoch": 0.9883351678528889,
+      "grad_norm": 45.28125,
+      "learning_rate": 9.845572891798751e-06,
+      "loss": 16.9063,
+      "step": 54660
+    },
+    {
+      "epoch": 0.9885159829220167,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.845544639492092e-06,
+      "loss": 16.9911,
+      "step": 54670
+    },
+    {
+      "epoch": 0.9886967979911446,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.845516387185433e-06,
+      "loss": 16.7827,
+      "step": 54680
+    },
+    {
+      "epoch": 0.9888776130602724,
+      "grad_norm": 46.53125,
+      "learning_rate": 9.845488134878773e-06,
+      "loss": 16.8007,
+      "step": 54690
+    },
+    {
+      "epoch": 0.9890584281294003,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.845459882572114e-06,
+      "loss": 16.2794,
+      "step": 54700
+    },
+    {
+      "epoch": 0.9892392431985282,
+      "grad_norm": 45.59375,
+      "learning_rate": 9.845431630265455e-06,
+      "loss": 16.7441,
+      "step": 54710
+    },
+    {
+      "epoch": 0.989420058267656,
+      "grad_norm": 43.25,
+      "learning_rate": 9.845403377958794e-06,
+      "loss": 16.602,
+      "step": 54720
+    },
+    {
+      "epoch": 0.9896008733367839,
+      "grad_norm": 46.59375,
+      "learning_rate": 9.845375125652134e-06,
+      "loss": 16.721,
+      "step": 54730
+    },
+    {
+      "epoch": 0.9897816884059117,
+      "grad_norm": 46.40625,
+      "learning_rate": 9.845346873345475e-06,
+      "loss": 17.2184,
+      "step": 54740
+    },
+    {
+      "epoch": 0.9899625034750396,
+      "grad_norm": 41.8125,
+      "learning_rate": 9.845318621038815e-06,
+      "loss": 16.793,
+      "step": 54750
+    },
+    {
+      "epoch": 0.9901433185441675,
+      "grad_norm": 43.5625,
+      "learning_rate": 9.845290368732156e-06,
+      "loss": 16.7583,
+      "step": 54760
+    },
+    {
+      "epoch": 0.9903241336132953,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.845262116425497e-06,
+      "loss": 16.8049,
+      "step": 54770
+    },
+    {
+      "epoch": 0.9905049486824232,
+      "grad_norm": 42.28125,
+      "learning_rate": 9.845233864118837e-06,
+      "loss": 16.8066,
+      "step": 54780
+    },
+    {
+      "epoch": 0.990685763751551,
+      "grad_norm": 42.3125,
+      "learning_rate": 9.845205611812178e-06,
+      "loss": 16.7789,
+      "step": 54790
+    },
+    {
+      "epoch": 0.990866578820679,
+      "grad_norm": 46.125,
+      "learning_rate": 9.845177359505518e-06,
+      "loss": 16.5531,
+      "step": 54800
+    },
+    {
+      "epoch": 0.9910473938898068,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.845149107198857e-06,
+      "loss": 16.586,
+      "step": 54810
+    },
+    {
+      "epoch": 0.9912282089589346,
+      "grad_norm": 44.5625,
+      "learning_rate": 9.845120854892198e-06,
+      "loss": 16.7009,
+      "step": 54820
+    },
+    {
+      "epoch": 0.9914090240280625,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.845092602585539e-06,
+      "loss": 16.6305,
+      "step": 54830
+    },
+    {
+      "epoch": 0.9915898390971903,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.84506435027888e-06,
+      "loss": 16.8697,
+      "step": 54840
+    },
+    {
+      "epoch": 0.9917706541663183,
+      "grad_norm": 45.25,
+      "learning_rate": 9.84503609797222e-06,
+      "loss": 16.5857,
+      "step": 54850
+    },
+    {
+      "epoch": 0.9919514692354461,
+      "grad_norm": 45.9375,
+      "learning_rate": 9.84500784566556e-06,
+      "loss": 16.4074,
+      "step": 54860
+    },
+    {
+      "epoch": 0.9921322843045739,
+      "grad_norm": 48.34375,
+      "learning_rate": 9.844979593358901e-06,
+      "loss": 16.8042,
+      "step": 54870
+    },
+    {
+      "epoch": 0.9923130993737018,
+      "grad_norm": 43.4375,
+      "learning_rate": 9.844951341052242e-06,
+      "loss": 16.6853,
+      "step": 54880
+    },
+    {
+      "epoch": 0.9924939144428296,
+      "grad_norm": 46.5625,
+      "learning_rate": 9.844923088745582e-06,
+      "loss": 16.6639,
+      "step": 54890
+    },
+    {
+      "epoch": 0.9926747295119576,
+      "grad_norm": 47.875,
+      "learning_rate": 9.844894836438921e-06,
+      "loss": 16.7631,
+      "step": 54900
+    },
+    {
+      "epoch": 0.9928555445810854,
+      "grad_norm": 43.34375,
+      "learning_rate": 9.844866584132262e-06,
+      "loss": 17.0811,
+      "step": 54910
+    },
+    {
+      "epoch": 0.9930363596502132,
+      "grad_norm": 47.59375,
+      "learning_rate": 9.844838331825603e-06,
+      "loss": 17.37,
+      "step": 54920
+    },
+    {
+      "epoch": 0.9932171747193411,
+      "grad_norm": 46.125,
+      "learning_rate": 9.844810079518943e-06,
+      "loss": 16.7417,
+      "step": 54930
+    },
+    {
+      "epoch": 0.9933979897884689,
+      "grad_norm": 47.75,
+      "learning_rate": 9.844781827212284e-06,
+      "loss": 16.3878,
+      "step": 54940
+    },
+    {
+      "epoch": 0.9935788048575969,
+      "grad_norm": 42.65625,
+      "learning_rate": 9.844753574905624e-06,
+      "loss": 16.5961,
+      "step": 54950
+    },
+    {
+      "epoch": 0.9937596199267247,
+      "grad_norm": 45.4375,
+      "learning_rate": 9.844725322598965e-06,
+      "loss": 16.6482,
+      "step": 54960
+    },
+    {
+      "epoch": 0.9939404349958526,
+      "grad_norm": 45.8125,
+      "learning_rate": 9.844697070292306e-06,
+      "loss": 16.5548,
+      "step": 54970
+    },
+    {
+      "epoch": 0.9941212500649804,
+      "grad_norm": 45.6875,
+      "learning_rate": 9.844668817985645e-06,
+      "loss": 16.468,
+      "step": 54980
+    },
+    {
+      "epoch": 0.9943020651341082,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.844640565678985e-06,
+      "loss": 16.5594,
+      "step": 54990
+    },
+    {
+      "epoch": 0.9944828802032362,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.844612313372326e-06,
+      "loss": 16.8028,
+      "step": 55000
+    },
+    {
+      "epoch": 0.9944828802032362,
+      "eval_loss": 2.082907199859619,
+      "eval_runtime": 228.736,
+      "eval_samples_per_second": 3174.223,
+      "eval_steps_per_second": 49.599,
+      "step": 55000
+    },
+    {
+      "epoch": 0.994663695272364,
+      "grad_norm": 44.0,
+      "learning_rate": 9.844584061065666e-06,
+      "loss": 16.6952,
+      "step": 55010
+    },
+    {
+      "epoch": 0.9948445103414919,
+      "grad_norm": 44.90625,
+      "learning_rate": 9.844555808759007e-06,
+      "loss": 16.4976,
+      "step": 55020
+    },
+    {
+      "epoch": 0.9950253254106197,
+      "grad_norm": 45.875,
+      "learning_rate": 9.844527556452346e-06,
+      "loss": 16.4644,
+      "step": 55030
+    },
+    {
+      "epoch": 0.9952061404797475,
+      "grad_norm": 44.0,
+      "learning_rate": 9.844499304145688e-06,
+      "loss": 16.7778,
+      "step": 55040
+    },
+    {
+      "epoch": 0.9953869555488755,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.844471051839029e-06,
+      "loss": 17.0716,
+      "step": 55050
+    },
+    {
+      "epoch": 0.9955677706180033,
+      "grad_norm": 53.21875,
+      "learning_rate": 9.84444279953237e-06,
+      "loss": 17.1855,
+      "step": 55060
+    },
+    {
+      "epoch": 0.9957485856871312,
+      "grad_norm": 45.09375,
+      "learning_rate": 9.844414547225709e-06,
+      "loss": 16.8956,
+      "step": 55070
+    },
+    {
+      "epoch": 0.995929400756259,
+      "grad_norm": 45.03125,
+      "learning_rate": 9.844386294919049e-06,
+      "loss": 16.755,
+      "step": 55080
+    },
+    {
+      "epoch": 0.9961102158253868,
+      "grad_norm": 46.8125,
+      "learning_rate": 9.84435804261239e-06,
+      "loss": 16.4342,
+      "step": 55090
+    },
+    {
+      "epoch": 0.9962910308945148,
+      "grad_norm": 43.59375,
+      "learning_rate": 9.84432979030573e-06,
+      "loss": 16.5041,
+      "step": 55100
+    },
+    {
+      "epoch": 0.9964718459636426,
+      "grad_norm": 44.65625,
+      "learning_rate": 9.844301537999071e-06,
+      "loss": 16.9493,
+      "step": 55110
+    },
+    {
+      "epoch": 0.9966526610327705,
+      "grad_norm": 41.65625,
+      "learning_rate": 9.844273285692412e-06,
+      "loss": 16.5818,
+      "step": 55120
+    },
+    {
+      "epoch": 0.9968334761018983,
+      "grad_norm": 44.625,
+      "learning_rate": 9.844245033385752e-06,
+      "loss": 16.5041,
+      "step": 55130
+    },
+    {
+      "epoch": 0.9970142911710262,
+      "grad_norm": 43.3125,
+      "learning_rate": 9.844216781079093e-06,
+      "loss": 16.6332,
+      "step": 55140
+    },
+    {
+      "epoch": 0.9971951062401541,
+      "grad_norm": 43.96875,
+      "learning_rate": 9.844188528772432e-06,
+      "loss": 16.9803,
+      "step": 55150
+    },
+    {
+      "epoch": 0.9973759213092819,
+      "grad_norm": 46.53125,
+      "learning_rate": 9.844160276465772e-06,
+      "loss": 16.5326,
+      "step": 55160
+    },
+    {
+      "epoch": 0.9975567363784098,
+      "grad_norm": 47.46875,
+      "learning_rate": 9.844132024159113e-06,
+      "loss": 16.9527,
+      "step": 55170
+    },
+    {
+      "epoch": 0.9977375514475376,
+      "grad_norm": 49.3125,
+      "learning_rate": 9.844103771852454e-06,
+      "loss": 16.533,
+      "step": 55180
+    },
+    {
+      "epoch": 0.9979183665166655,
+      "grad_norm": 46.78125,
+      "learning_rate": 9.844075519545794e-06,
+      "loss": 16.6507,
+      "step": 55190
+    },
+    {
+      "epoch": 0.9980991815857934,
+      "grad_norm": 45.40625,
+      "learning_rate": 9.844047267239133e-06,
+      "loss": 16.4906,
+      "step": 55200
+    },
+    {
+      "epoch": 0.9982799966549212,
+      "grad_norm": 44.3125,
+      "learning_rate": 9.844019014932476e-06,
+      "loss": 16.5156,
+      "step": 55210
+    },
+    {
+      "epoch": 0.9984608117240491,
+      "grad_norm": 47.5625,
+      "learning_rate": 9.843990762625816e-06,
+      "loss": 16.7121,
+      "step": 55220
+    },
+    {
+      "epoch": 0.9986416267931769,
+      "grad_norm": 44.28125,
+      "learning_rate": 9.843962510319157e-06,
+      "loss": 16.9591,
+      "step": 55230
+    },
+    {
+      "epoch": 0.9988224418623048,
+      "grad_norm": 44.78125,
+      "learning_rate": 9.843934258012496e-06,
+      "loss": 16.4461,
+      "step": 55240
+    },
+    {
+      "epoch": 0.9990032569314327,
+      "grad_norm": 44.40625,
+      "learning_rate": 9.843906005705836e-06,
+      "loss": 16.741,
+      "step": 55250
+    },
+    {
+      "epoch": 0.9991840720005605,
+      "grad_norm": 44.03125,
+      "learning_rate": 9.843877753399177e-06,
+      "loss": 16.965,
+      "step": 55260
+    },
+    {
+      "epoch": 0.9993648870696884,
+      "grad_norm": 41.5,
+      "learning_rate": 9.843849501092518e-06,
+      "loss": 16.5569,
+      "step": 55270
+    },
+    {
+      "epoch": 0.9995457021388162,
+      "grad_norm": 47.46875,
+      "learning_rate": 9.843821248785858e-06,
+      "loss": 16.7305,
+      "step": 55280
+    },
+    {
+      "epoch": 0.9997265172079441,
+      "grad_norm": 42.53125,
+      "learning_rate": 9.843792996479197e-06,
+      "loss": 16.5323,
+      "step": 55290
+    },
+    {
+      "epoch": 0.999907332277072,
+      "grad_norm": 43.65625,
+      "learning_rate": 9.84376474417254e-06,
+      "loss": 16.207,
+      "step": 55300
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 55305,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.653613618876383e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}