diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.2850923240184784,
-  "best_model_checkpoint": "./cifar100_outputs/checkpoint-26565",
+  "best_metric": 0.2550007104873657,
+  "best_model_checkpoint": "./cifar100_lora_outputs/checkpoint-26565",
   "epoch": 5.0,
   "eval_steps": 500,
   "global_step": 26565,
@@ -10,18649 +10,18649 @@
   "log_history": [
     {
       "epoch": 0.0018821757952192735,
-      "grad_norm": 3.259775161743164,
-      "learning_rate": 2.9988706945228687e-05,
-      "loss": 4.6305,
+      "grad_norm": 1.771733283996582,
+      "learning_rate": 0.0004998117824204781,
+      "loss": 4.6194,
       "step": 10
     },
     {
       "epoch": 0.003764351590438547,
-      "grad_norm": 2.9150309562683105,
-      "learning_rate": 2.997741389045737e-05,
-      "loss": 4.5916,
+      "grad_norm": 1.8498225212097168,
+      "learning_rate": 0.0004996235648409562,
+      "loss": 4.5768,
       "step": 20
     },
     {
       "epoch": 0.00564652738565782,
-      "grad_norm": 2.639889717102051,
-      "learning_rate": 2.9966120835686056e-05,
-      "loss": 4.5977,
+      "grad_norm": 1.8128389120101929,
+      "learning_rate": 0.0004994353472614342,
+      "loss": 4.5582,
       "step": 30
     },
     {
       "epoch": 0.007528703180877094,
-      "grad_norm": 2.729921340942383,
-      "learning_rate": 2.9954827780914738e-05,
-      "loss": 4.5964,
+      "grad_norm": 1.824784755706787,
+      "learning_rate": 0.0004992471296819123,
+      "loss": 4.5086,
       "step": 40
     },
     {
       "epoch": 0.009410878976096368,
-      "grad_norm": 2.3783774375915527,
-      "learning_rate": 2.994353472614342e-05,
-      "loss": 4.6022,
+      "grad_norm": 1.7717084884643555,
+      "learning_rate": 0.0004990589121023904,
+      "loss": 4.5419,
       "step": 50
     },
     {
       "epoch": 0.01129305477131564,
-      "grad_norm": 2.656496524810791,
-      "learning_rate": 2.9932241671372107e-05,
-      "loss": 4.5908,
+      "grad_norm": 1.8626877069473267,
+      "learning_rate": 0.0004988706945228685,
+      "loss": 4.475,
       "step": 60
     },
     {
       "epoch": 0.013175230566534914,
-      "grad_norm": 2.692168712615967,
-      "learning_rate": 2.992094861660079e-05,
-      "loss": 4.5829,
+      "grad_norm": 1.8645660877227783,
+      "learning_rate": 0.0004986824769433465,
+      "loss": 4.4153,
       "step": 70
     },
     {
       "epoch": 0.015057406361754188,
-      "grad_norm": 2.8096225261688232,
-      "learning_rate": 2.9909655561829476e-05,
-      "loss": 4.5645,
+      "grad_norm": 1.8023477792739868,
+      "learning_rate": 0.0004984942593638246,
+      "loss": 4.4258,
       "step": 80
     },
     {
       "epoch": 0.01693958215697346,
-      "grad_norm": 2.7067952156066895,
-      "learning_rate": 2.989836250705816e-05,
-      "loss": 4.5448,
+      "grad_norm": 1.8429126739501953,
+      "learning_rate": 0.0004983060417843027,
+      "loss": 4.3373,
       "step": 90
     },
     {
       "epoch": 0.018821757952192736,
-      "grad_norm": 2.5358433723449707,
-      "learning_rate": 2.9887069452286845e-05,
-      "loss": 4.5471,
+      "grad_norm": 1.9091802835464478,
+      "learning_rate": 0.0004981178242047808,
+      "loss": 4.2982,
       "step": 100
     },
     {
       "epoch": 0.020703933747412008,
-      "grad_norm": 2.4757063388824463,
-      "learning_rate": 2.987577639751553e-05,
-      "loss": 4.5546,
+      "grad_norm": 1.7966872453689575,
+      "learning_rate": 0.0004979296066252589,
+      "loss": 4.2458,
       "step": 110
     },
     {
       "epoch": 0.02258610954263128,
-      "grad_norm": 2.8501641750335693,
-      "learning_rate": 2.9864483342744213e-05,
-      "loss": 4.5471,
+      "grad_norm": 1.9354859590530396,
+      "learning_rate": 0.0004977413890457369,
+      "loss": 4.1789,
       "step": 120
     },
     {
       "epoch": 0.024468285337850556,
-      "grad_norm": 2.506326198577881,
-      "learning_rate": 2.98531902879729e-05,
-      "loss": 4.5289,
+      "grad_norm": 2.0137393474578857,
+      "learning_rate": 0.000497553171466215,
+      "loss": 4.105,
       "step": 130
     },
     {
       "epoch": 0.026350461133069828,
-      "grad_norm": 2.7096266746520996,
-      "learning_rate": 2.9841897233201582e-05,
-      "loss": 4.5213,
+      "grad_norm": 1.9952950477600098,
+      "learning_rate": 0.000497364953886693,
+      "loss": 4.0637,
       "step": 140
     },
     {
       "epoch": 0.028232636928289104,
-      "grad_norm": 2.9105842113494873,
-      "learning_rate": 2.9830604178430265e-05,
-      "loss": 4.5017,
+      "grad_norm": 2.067469835281372,
+      "learning_rate": 0.000497176736307171,
+      "loss": 3.9707,
       "step": 150
     },
     {
       "epoch": 0.030114812723508376,
-      "grad_norm": 2.713897466659546,
-      "learning_rate": 2.981931112365895e-05,
-      "loss": 4.5014,
+      "grad_norm": 2.226876974105835,
+      "learning_rate": 0.0004969885187276491,
+      "loss": 3.958,
       "step": 160
     },
     {
       "epoch": 0.03199698851872765,
-      "grad_norm": 3.520843744277954,
-      "learning_rate": 2.9808018068887634e-05,
-      "loss": 4.482,
+      "grad_norm": 2.162407159805298,
+      "learning_rate": 0.0004968003011481272,
+      "loss": 3.9696,
       "step": 170
     },
     {
       "epoch": 0.03387916431394692,
-      "grad_norm": 3.0594747066497803,
-      "learning_rate": 2.979672501411632e-05,
-      "loss": 4.4706,
+      "grad_norm": 1.9613149166107178,
+      "learning_rate": 0.0004966120835686053,
+      "loss": 3.7365,
       "step": 180
     },
     {
       "epoch": 0.0357613401091662,
-      "grad_norm": 2.884718418121338,
-      "learning_rate": 2.9785431959345006e-05,
-      "loss": 4.4648,
+      "grad_norm": 2.009887933731079,
+      "learning_rate": 0.0004964238659890834,
+      "loss": 3.6623,
       "step": 190
     },
     {
       "epoch": 0.03764351590438547,
-      "grad_norm": 2.8616080284118652,
-      "learning_rate": 2.977413890457369e-05,
-      "loss": 4.4629,
+      "grad_norm": 2.3235840797424316,
+      "learning_rate": 0.0004962356484095614,
+      "loss": 3.6737,
       "step": 200
     },
     {
       "epoch": 0.039525691699604744,
-      "grad_norm": 2.6512672901153564,
-      "learning_rate": 2.9762845849802374e-05,
-      "loss": 4.4555,
+      "grad_norm": 1.996651291847229,
+      "learning_rate": 0.0004960474308300395,
+      "loss": 3.6248,
       "step": 210
     },
     {
       "epoch": 0.041407867494824016,
-      "grad_norm": 3.5559914112091064,
-      "learning_rate": 2.9751552795031054e-05,
-      "loss": 4.4335,
+      "grad_norm": 2.1277413368225098,
+      "learning_rate": 0.0004958592132505176,
+      "loss": 3.5943,
       "step": 220
     },
     {
       "epoch": 0.04329004329004329,
-      "grad_norm": 2.8921027183532715,
-      "learning_rate": 2.974025974025974e-05,
-      "loss": 4.3462,
+      "grad_norm": 2.05141019821167,
+      "learning_rate": 0.0004956709956709957,
+      "loss": 3.2136,
       "step": 230
     },
     {
       "epoch": 0.04517221908526256,
-      "grad_norm": 2.7680187225341797,
-      "learning_rate": 2.9728966685488426e-05,
-      "loss": 4.4017,
+      "grad_norm": 2.045433521270752,
+      "learning_rate": 0.0004954827780914737,
+      "loss": 3.376,
       "step": 240
     },
     {
       "epoch": 0.04705439488048184,
-      "grad_norm": 2.8839056491851807,
-      "learning_rate": 2.971767363071711e-05,
-      "loss": 4.4098,
+      "grad_norm": 2.1396985054016113,
+      "learning_rate": 0.0004952945605119518,
+      "loss": 3.3767,
       "step": 250
     },
     {
       "epoch": 0.04893657067570111,
-      "grad_norm": 2.8372817039489746,
-      "learning_rate": 2.9706380575945795e-05,
-      "loss": 4.3837,
+      "grad_norm": 2.0575475692749023,
+      "learning_rate": 0.0004951063429324299,
+      "loss": 3.2651,
       "step": 260
     },
     {
       "epoch": 0.050818746470920384,
-      "grad_norm": 2.884737968444824,
-      "learning_rate": 2.9695087521174477e-05,
-      "loss": 4.3769,
+      "grad_norm": 2.0891125202178955,
+      "learning_rate": 0.000494918125352908,
+      "loss": 3.2083,
       "step": 270
     },
     {
       "epoch": 0.052700922266139656,
-      "grad_norm": 2.899841070175171,
-      "learning_rate": 2.9683794466403163e-05,
-      "loss": 4.3447,
+      "grad_norm": 2.27366304397583,
+      "learning_rate": 0.000494729907773386,
+      "loss": 2.9784,
       "step": 280
     },
     {
       "epoch": 0.05458309806135893,
-      "grad_norm": 3.1840929985046387,
-      "learning_rate": 2.967250141163185e-05,
-      "loss": 4.3174,
+      "grad_norm": 2.2188055515289307,
+      "learning_rate": 0.0004945416901938641,
+      "loss": 2.981,
       "step": 290
     },
     {
       "epoch": 0.05646527385657821,
-      "grad_norm": 2.9409921169281006,
-      "learning_rate": 2.9661208356860532e-05,
-      "loss": 4.3072,
+      "grad_norm": 2.5624613761901855,
+      "learning_rate": 0.0004943534726143422,
+      "loss": 2.9,
       "step": 300
     },
     {
       "epoch": 0.05834744965179748,
-      "grad_norm": 3.342649459838867,
-      "learning_rate": 2.9649915302089215e-05,
-      "loss": 4.3291,
+      "grad_norm": 2.1516923904418945,
+      "learning_rate": 0.0004941652550348203,
+      "loss": 3.0586,
       "step": 310
     },
     {
       "epoch": 0.06022962544701675,
-      "grad_norm": 2.746373176574707,
-      "learning_rate": 2.9638622247317898e-05,
-      "loss": 4.3288,
+      "grad_norm": 2.2759182453155518,
+      "learning_rate": 0.0004939770374552983,
+      "loss": 3.0083,
       "step": 320
     },
     {
       "epoch": 0.062111801242236024,
-      "grad_norm": 3.7221550941467285,
-      "learning_rate": 2.9627329192546584e-05,
-      "loss": 4.2612,
+      "grad_norm": 2.1301019191741943,
+      "learning_rate": 0.0004937888198757764,
+      "loss": 2.653,
       "step": 330
     },
     {
       "epoch": 0.0639939770374553,
-      "grad_norm": 2.8289361000061035,
-      "learning_rate": 2.961603613777527e-05,
-      "loss": 4.2885,
+      "grad_norm": 2.1771981716156006,
+      "learning_rate": 0.0004936006022962545,
+      "loss": 2.8033,
       "step": 340
     },
     {
       "epoch": 0.06587615283267458,
-      "grad_norm": 3.2310261726379395,
-      "learning_rate": 2.9604743083003952e-05,
-      "loss": 4.2502,
+      "grad_norm": 2.0707926750183105,
+      "learning_rate": 0.0004934123847167326,
+      "loss": 2.6861,
       "step": 350
     },
     {
       "epoch": 0.06775832862789384,
-      "grad_norm": 2.873511791229248,
-      "learning_rate": 2.959345002823264e-05,
-      "loss": 4.2713,
+      "grad_norm": 2.312077283859253,
+      "learning_rate": 0.0004932241671372107,
+      "loss": 2.7215,
       "step": 360
     },
     {
       "epoch": 0.06964050442311312,
-      "grad_norm": 2.8159632682800293,
-      "learning_rate": 2.958215697346132e-05,
-      "loss": 4.237,
+      "grad_norm": 2.0129685401916504,
+      "learning_rate": 0.0004930359495576887,
+      "loss": 2.5832,
       "step": 370
     },
     {
       "epoch": 0.0715226802183324,
-      "grad_norm": 4.2501220703125,
-      "learning_rate": 2.9570863918690007e-05,
-      "loss": 4.2377,
+      "grad_norm": 2.031329393386841,
+      "learning_rate": 0.0004928477319781668,
+      "loss": 2.6758,
       "step": 380
     },
     {
       "epoch": 0.07340485601355166,
-      "grad_norm": 3.2059030532836914,
-      "learning_rate": 2.9559570863918693e-05,
-      "loss": 4.245,
+      "grad_norm": 2.230658769607544,
+      "learning_rate": 0.0004926595143986449,
+      "loss": 2.696,
       "step": 390
     },
     {
       "epoch": 0.07528703180877094,
-      "grad_norm": 3.3538947105407715,
-      "learning_rate": 2.9548277809147373e-05,
-      "loss": 4.2061,
+      "grad_norm": 2.8371505737304688,
+      "learning_rate": 0.0004924712968191229,
+      "loss": 2.5156,
       "step": 400
     },
     {
       "epoch": 0.07716920760399021,
-      "grad_norm": 3.506152391433716,
-      "learning_rate": 2.953698475437606e-05,
-      "loss": 4.1917,
+      "grad_norm": 2.619385004043579,
+      "learning_rate": 0.0004922830792396009,
+      "loss": 2.4818,
       "step": 410
     },
     {
       "epoch": 0.07905138339920949,
-      "grad_norm": 3.823164701461792,
-      "learning_rate": 2.9525691699604745e-05,
-      "loss": 4.1214,
+      "grad_norm": 2.360231637954712,
+      "learning_rate": 0.000492094861660079,
+      "loss": 2.2714,
       "step": 420
     },
     {
       "epoch": 0.08093355919442875,
-      "grad_norm": 3.458685874938965,
-      "learning_rate": 2.9514398644833427e-05,
-      "loss": 4.1866,
+      "grad_norm": 2.3182997703552246,
+      "learning_rate": 0.0004919066440805571,
+      "loss": 2.3886,
       "step": 430
     },
     {
       "epoch": 0.08281573498964803,
-      "grad_norm": 3.0740559101104736,
-      "learning_rate": 2.9503105590062114e-05,
-      "loss": 4.1349,
+      "grad_norm": 2.0524964332580566,
+      "learning_rate": 0.0004917184265010352,
+      "loss": 2.2321,
       "step": 440
     },
     {
       "epoch": 0.08469791078486731,
-      "grad_norm": 4.038434982299805,
-      "learning_rate": 2.9491812535290796e-05,
-      "loss": 4.1513,
+      "grad_norm": 1.905816912651062,
+      "learning_rate": 0.0004915302089215133,
+      "loss": 2.2257,
       "step": 450
     },
     {
       "epoch": 0.08658008658008658,
-      "grad_norm": 3.2392990589141846,
-      "learning_rate": 2.9480519480519482e-05,
-      "loss": 4.0751,
+      "grad_norm": 2.24204421043396,
+      "learning_rate": 0.0004913419913419914,
+      "loss": 2.1486,
       "step": 460
     },
     {
       "epoch": 0.08846226237530586,
-      "grad_norm": 2.7104597091674805,
-      "learning_rate": 2.946922642574817e-05,
-      "loss": 4.1389,
+      "grad_norm": 1.8304845094680786,
+      "learning_rate": 0.0004911537737624695,
+      "loss": 2.3883,
       "step": 470
     },
     {
       "epoch": 0.09034443817052512,
-      "grad_norm": 3.526334762573242,
-      "learning_rate": 2.9457933370976848e-05,
-      "loss": 4.0984,
+      "grad_norm": 2.062603235244751,
+      "learning_rate": 0.0004909655561829475,
+      "loss": 2.2133,
       "step": 480
     },
     {
       "epoch": 0.0922266139657444,
-      "grad_norm": 4.827049732208252,
-      "learning_rate": 2.9446640316205534e-05,
-      "loss": 4.065,
+      "grad_norm": 2.481670618057251,
+      "learning_rate": 0.0004907773386034255,
+      "loss": 1.9732,
       "step": 490
     },
     {
       "epoch": 0.09410878976096368,
-      "grad_norm": 3.1481475830078125,
-      "learning_rate": 2.9435347261434216e-05,
-      "loss": 4.0429,
+      "grad_norm": 1.9979970455169678,
+      "learning_rate": 0.0004905891210239036,
+      "loss": 2.0477,
       "step": 500
     },
     {
       "epoch": 0.09599096555618294,
-      "grad_norm": 3.47776198387146,
-      "learning_rate": 2.9424054206662903e-05,
-      "loss": 4.0432,
+      "grad_norm": 2.0687055587768555,
+      "learning_rate": 0.0004904009034443817,
+      "loss": 1.9344,
       "step": 510
     },
     {
       "epoch": 0.09787314135140222,
-      "grad_norm": 2.9574878215789795,
-      "learning_rate": 2.941276115189159e-05,
-      "loss": 4.0486,
+      "grad_norm": 1.8960387706756592,
+      "learning_rate": 0.0004902126858648598,
+      "loss": 2.1094,
       "step": 520
     },
     {
       "epoch": 0.09975531714662149,
-      "grad_norm": 3.398369073867798,
-      "learning_rate": 2.940146809712027e-05,
-      "loss": 3.9975,
+      "grad_norm": 2.011258125305176,
+      "learning_rate": 0.0004900244682853378,
+      "loss": 2.0836,
       "step": 530
     },
     {
       "epoch": 0.10163749294184077,
-      "grad_norm": 3.2423532009124756,
-      "learning_rate": 2.9390175042348957e-05,
-      "loss": 4.0189,
+      "grad_norm": 2.1397879123687744,
+      "learning_rate": 0.0004898362507058159,
+      "loss": 2.1036,
       "step": 540
     },
     {
       "epoch": 0.10351966873706005,
-      "grad_norm": 4.2174153327941895,
-      "learning_rate": 2.937888198757764e-05,
-      "loss": 3.9679,
+      "grad_norm": 2.2543036937713623,
+      "learning_rate": 0.000489648033126294,
+      "loss": 1.727,
       "step": 550
     },
     {
       "epoch": 0.10540184453227931,
-      "grad_norm": 4.3004045486450195,
-      "learning_rate": 2.9367588932806326e-05,
-      "loss": 4.0162,
+      "grad_norm": 2.0110862255096436,
+      "learning_rate": 0.0004894598155467721,
+      "loss": 2.0504,
       "step": 560
     },
     {
       "epoch": 0.10728402032749859,
-      "grad_norm": 3.0636956691741943,
-      "learning_rate": 2.935629587803501e-05,
-      "loss": 3.9957,
+      "grad_norm": 2.556957483291626,
+      "learning_rate": 0.0004892715979672501,
+      "loss": 2.0245,
       "step": 570
     },
     {
       "epoch": 0.10916619612271786,
-      "grad_norm": 3.1375842094421387,
-      "learning_rate": 2.934500282326369e-05,
-      "loss": 4.0078,
+      "grad_norm": 3.036496639251709,
+      "learning_rate": 0.0004890833803877282,
+      "loss": 1.9936,
       "step": 580
     },
     {
       "epoch": 0.11104837191793714,
-      "grad_norm": 3.439136505126953,
-      "learning_rate": 2.9333709768492378e-05,
-      "loss": 3.9473,
+      "grad_norm": 2.3555924892425537,
+      "learning_rate": 0.0004888951628082063,
+      "loss": 1.8314,
       "step": 590
     },
     {
       "epoch": 0.11293054771315642,
-      "grad_norm": 3.252802610397339,
-      "learning_rate": 2.932241671372106e-05,
-      "loss": 3.9155,
+      "grad_norm": 1.9588710069656372,
+      "learning_rate": 0.0004887069452286844,
+      "loss": 1.8489,
       "step": 600
     },
     {
       "epoch": 0.11481272350837568,
-      "grad_norm": 3.3427743911743164,
-      "learning_rate": 2.9311123658949746e-05,
-      "loss": 3.9003,
+      "grad_norm": 2.1294167041778564,
+      "learning_rate": 0.0004885187276491625,
+      "loss": 2.0076,
       "step": 610
     },
     {
       "epoch": 0.11669489930359496,
-      "grad_norm": 3.1991214752197266,
-      "learning_rate": 2.9299830604178432e-05,
-      "loss": 3.8892,
+      "grad_norm": 2.881962299346924,
+      "learning_rate": 0.0004883305100696405,
+      "loss": 1.6918,
       "step": 620
     },
     {
       "epoch": 0.11857707509881422,
-      "grad_norm": 3.2154698371887207,
-      "learning_rate": 2.9288537549407115e-05,
-      "loss": 3.8717,
+      "grad_norm": 2.8056647777557373,
+      "learning_rate": 0.0004881422924901186,
+      "loss": 1.7188,
       "step": 630
     },
     {
       "epoch": 0.1204592508940335,
-      "grad_norm": 2.7783098220825195,
-      "learning_rate": 2.92772444946358e-05,
-      "loss": 3.8236,
+      "grad_norm": 1.6415379047393799,
+      "learning_rate": 0.0004879540749105967,
+      "loss": 1.5487,
       "step": 640
     },
     {
       "epoch": 0.12234142668925278,
-      "grad_norm": 4.598132610321045,
-      "learning_rate": 2.9265951439864487e-05,
-      "loss": 3.9406,
+      "grad_norm": 4.757891654968262,
+      "learning_rate": 0.00048776585733107476,
+      "loss": 1.8836,
       "step": 650
     },
     {
       "epoch": 0.12422360248447205,
-      "grad_norm": 3.633305549621582,
-      "learning_rate": 2.9254658385093167e-05,
-      "loss": 3.7923,
+      "grad_norm": 3.0680055618286133,
+      "learning_rate": 0.0004875776397515528,
+      "loss": 1.5568,
       "step": 660
     },
     {
       "epoch": 0.12610577827969133,
-      "grad_norm": 3.794126272201538,
-      "learning_rate": 2.9243365330321853e-05,
-      "loss": 3.7865,
+      "grad_norm": 2.2202789783477783,
+      "learning_rate": 0.00048738942217203086,
+      "loss": 1.5855,
       "step": 670
     },
     {
       "epoch": 0.1279879540749106,
-      "grad_norm": 3.1769659519195557,
-      "learning_rate": 2.9232072275550535e-05,
-      "loss": 3.7679,
+      "grad_norm": 4.045197010040283,
+      "learning_rate": 0.00048720120459250894,
+      "loss": 1.6895,
       "step": 680
     },
     {
       "epoch": 0.12987012987012986,
-      "grad_norm": 3.30849289894104,
-      "learning_rate": 2.922077922077922e-05,
-      "loss": 3.8356,
+      "grad_norm": 2.5233097076416016,
+      "learning_rate": 0.000487012987012987,
+      "loss": 1.5461,
       "step": 690
     },
     {
       "epoch": 0.13175230566534915,
-      "grad_norm": 4.426156997680664,
-      "learning_rate": 2.9209486166007908e-05,
-      "loss": 3.7757,
+      "grad_norm": 2.6602494716644287,
+      "learning_rate": 0.0004868247694334651,
+      "loss": 1.6852,
       "step": 700
     },
     {
       "epoch": 0.13363448146056842,
-      "grad_norm": 3.004460096359253,
-      "learning_rate": 2.919819311123659e-05,
-      "loss": 3.6859,
+      "grad_norm": 1.7626466751098633,
+      "learning_rate": 0.00048663655185394317,
+      "loss": 1.4548,
       "step": 710
     },
     {
       "epoch": 0.13551665725578768,
-      "grad_norm": 3.6174192428588867,
-      "learning_rate": 2.9186900056465276e-05,
-      "loss": 3.745,
+      "grad_norm": 2.280395030975342,
+      "learning_rate": 0.00048644833427442124,
+      "loss": 1.5311,
       "step": 720
     },
     {
       "epoch": 0.13739883305100697,
-      "grad_norm": 3.1848292350769043,
-      "learning_rate": 2.917560700169396e-05,
-      "loss": 3.6902,
+      "grad_norm": 1.9135655164718628,
+      "learning_rate": 0.0004862601166948993,
+      "loss": 1.5031,
       "step": 730
     },
     {
       "epoch": 0.13928100884622624,
-      "grad_norm": 3.1406095027923584,
-      "learning_rate": 2.9164313946922645e-05,
-      "loss": 3.7553,
+      "grad_norm": 1.908615231513977,
+      "learning_rate": 0.0004860718991153774,
+      "loss": 1.6354,
       "step": 740
     },
     {
       "epoch": 0.1411631846414455,
-      "grad_norm": 3.086644411087036,
-      "learning_rate": 2.9153020892151328e-05,
-      "loss": 3.736,
+      "grad_norm": 1.6709424257278442,
+      "learning_rate": 0.0004858836815358554,
+      "loss": 1.4328,
       "step": 750
     },
     {
       "epoch": 0.1430453604366648,
-      "grad_norm": 5.149982452392578,
-      "learning_rate": 2.914172783738001e-05,
-      "loss": 3.6852,
+      "grad_norm": 3.3939626216888428,
+      "learning_rate": 0.0004856954639563335,
+      "loss": 1.3476,
       "step": 760
     },
     {
       "epoch": 0.14492753623188406,
-      "grad_norm": 4.200068950653076,
-      "learning_rate": 2.9130434782608696e-05,
-      "loss": 3.6916,
+      "grad_norm": 2.578439712524414,
+      "learning_rate": 0.00048550724637681157,
+      "loss": 1.4497,
       "step": 770
     },
     {
       "epoch": 0.14680971202710333,
-      "grad_norm": 3.5384104251861572,
-      "learning_rate": 2.911914172783738e-05,
-      "loss": 3.6724,
+      "grad_norm": 2.4868099689483643,
+      "learning_rate": 0.0004853190287972897,
+      "loss": 1.5841,
       "step": 780
     },
     {
       "epoch": 0.1486918878223226,
-      "grad_norm": 4.555196285247803,
-      "learning_rate": 2.9107848673066065e-05,
-      "loss": 3.6608,
+      "grad_norm": 2.2631711959838867,
+      "learning_rate": 0.0004851308112177678,
+      "loss": 1.2718,
       "step": 790
     },
     {
       "epoch": 0.1505740636175419,
-      "grad_norm": 4.264336585998535,
-      "learning_rate": 2.909655561829475e-05,
-      "loss": 3.6538,
+      "grad_norm": 1.6401269435882568,
+      "learning_rate": 0.00048494259363824585,
+      "loss": 1.3654,
       "step": 800
     },
     {
       "epoch": 0.15245623941276115,
-      "grad_norm": 4.192948818206787,
-      "learning_rate": 2.9085262563523434e-05,
-      "loss": 3.5788,
+      "grad_norm": 3.559016466140747,
+      "learning_rate": 0.00048475437605872393,
+      "loss": 1.4383,
       "step": 810
     },
     {
       "epoch": 0.15433841520798042,
-      "grad_norm": 4.293764114379883,
-      "learning_rate": 2.907396950875212e-05,
-      "loss": 3.6034,
+      "grad_norm": 2.363433361053467,
+      "learning_rate": 0.000484566158479202,
+      "loss": 1.3603,
       "step": 820
     },
     {
       "epoch": 0.1562205910031997,
-      "grad_norm": 3.505387544631958,
-      "learning_rate": 2.90626764539808e-05,
-      "loss": 3.6135,
+      "grad_norm": 2.9046192169189453,
+      "learning_rate": 0.00048437794089968003,
+      "loss": 1.3902,
       "step": 830
     },
     {
       "epoch": 0.15810276679841898,
-      "grad_norm": 3.4899723529815674,
-      "learning_rate": 2.9051383399209485e-05,
-      "loss": 3.7187,
+      "grad_norm": 3.304086923599243,
+      "learning_rate": 0.0004841897233201581,
+      "loss": 1.6489,
       "step": 840
     },
     {
       "epoch": 0.15998494259363824,
-      "grad_norm": 5.375092029571533,
-      "learning_rate": 2.904009034443817e-05,
-      "loss": 3.6486,
+      "grad_norm": 2.0543618202209473,
+      "learning_rate": 0.0004840015057406362,
+      "loss": 1.4324,
       "step": 850
     },
     {
       "epoch": 0.1618671183888575,
-      "grad_norm": 4.234030246734619,
-      "learning_rate": 2.9028797289666854e-05,
-      "loss": 3.7097,
+      "grad_norm": 3.817758560180664,
+      "learning_rate": 0.00048381328816111426,
+      "loss": 1.497,
       "step": 860
     },
     {
       "epoch": 0.1637492941840768,
-      "grad_norm": 3.9143476486206055,
-      "learning_rate": 2.901750423489554e-05,
-      "loss": 3.4356,
+      "grad_norm": 3.9704749584198,
+      "learning_rate": 0.00048362507058159233,
+      "loss": 1.192,
       "step": 870
     },
     {
       "epoch": 0.16563146997929606,
-      "grad_norm": 3.578145742416382,
-      "learning_rate": 2.9006211180124223e-05,
-      "loss": 3.4969,
+      "grad_norm": 2.5795369148254395,
+      "learning_rate": 0.0004834368530020704,
+      "loss": 1.2547,
       "step": 880
     },
     {
       "epoch": 0.16751364577451533,
-      "grad_norm": 4.751120567321777,
-      "learning_rate": 2.899491812535291e-05,
-      "loss": 3.5853,
+      "grad_norm": 3.441765069961548,
+      "learning_rate": 0.0004832486354225485,
+      "loss": 1.396,
       "step": 890
     },
     {
       "epoch": 0.16939582156973462,
-      "grad_norm": 4.476664066314697,
-      "learning_rate": 2.8983625070581595e-05,
-      "loss": 3.5424,
+      "grad_norm": 2.1846425533294678,
+      "learning_rate": 0.00048306041784302656,
+      "loss": 1.4624,
       "step": 900
     },
     {
       "epoch": 0.1712779973649539,
-      "grad_norm": 3.4918501377105713,
-      "learning_rate": 2.8972332015810278e-05,
-      "loss": 3.5004,
+      "grad_norm": 2.7610526084899902,
+      "learning_rate": 0.00048287220026350464,
+      "loss": 1.249,
       "step": 910
     },
     {
       "epoch": 0.17316017316017315,
-      "grad_norm": 9.847784996032715,
-      "learning_rate": 2.896103896103896e-05,
-      "loss": 3.5159,
+      "grad_norm": 3.498777151107788,
+      "learning_rate": 0.00048268398268398266,
+      "loss": 1.2124,
       "step": 920
     },
     {
       "epoch": 0.17504234895539245,
-      "grad_norm": 3.551976442337036,
-      "learning_rate": 2.8949745906267647e-05,
-      "loss": 3.475,
+      "grad_norm": 2.5937039852142334,
+      "learning_rate": 0.00048249576510446074,
+      "loss": 1.2934,
       "step": 930
     },
     {
       "epoch": 0.1769245247506117,
-      "grad_norm": 4.220102787017822,
-      "learning_rate": 2.893845285149633e-05,
-      "loss": 3.4396,
+      "grad_norm": 3.5550036430358887,
+      "learning_rate": 0.0004823075475249388,
+      "loss": 1.2937,
       "step": 940
     },
     {
       "epoch": 0.17880670054583098,
-      "grad_norm": 6.077229976654053,
-      "learning_rate": 2.8927159796725015e-05,
-      "loss": 3.448,
+      "grad_norm": 2.0531303882598877,
+      "learning_rate": 0.0004821193299454169,
+      "loss": 1.3274,
       "step": 950
     },
     {
       "epoch": 0.18068887634105024,
-      "grad_norm": 4.203821659088135,
-      "learning_rate": 2.8915866741953698e-05,
-      "loss": 3.4436,
+      "grad_norm": 2.3458123207092285,
+      "learning_rate": 0.00048193111236589497,
+      "loss": 1.3146,
       "step": 960
     },
     {
       "epoch": 0.18257105213626953,
-      "grad_norm": 3.7120420932769775,
-      "learning_rate": 2.8904573687182384e-05,
-      "loss": 3.416,
+      "grad_norm": 4.257774829864502,
+      "learning_rate": 0.00048174289478637304,
+      "loss": 1.2663,
       "step": 970
     },
     {
       "epoch": 0.1844532279314888,
-      "grad_norm": 4.179491996765137,
-      "learning_rate": 2.889328063241107e-05,
-      "loss": 3.3503,
+      "grad_norm": 4.102976322174072,
+      "learning_rate": 0.0004815546772068511,
+      "loss": 1.0864,
       "step": 980
     },
     {
       "epoch": 0.18633540372670807,
-      "grad_norm": 3.9055488109588623,
-      "learning_rate": 2.8881987577639753e-05,
-      "loss": 3.4424,
+      "grad_norm": 2.92472505569458,
+      "learning_rate": 0.0004813664596273292,
+      "loss": 1.5308,
       "step": 990
     },
     {
       "epoch": 0.18821757952192736,
-      "grad_norm": 3.6015703678131104,
-      "learning_rate": 2.887069452286844e-05,
-      "loss": 3.3619,
+      "grad_norm": 4.026129722595215,
+      "learning_rate": 0.00048117824204780733,
+      "loss": 1.2942,
       "step": 1000
     },
     {
       "epoch": 0.19009975531714662,
-      "grad_norm": 5.409115314483643,
-      "learning_rate": 2.8859401468097118e-05,
-      "loss": 3.4609,
+      "grad_norm": 3.1226966381073,
+      "learning_rate": 0.00048099002446828535,
+      "loss": 1.3299,
       "step": 1010
     },
     {
       "epoch": 0.1919819311123659,
-      "grad_norm": 3.458120584487915,
-      "learning_rate": 2.8848108413325804e-05,
-      "loss": 3.4654,
+      "grad_norm": 1.2881345748901367,
+      "learning_rate": 0.0004808018068887634,
+      "loss": 1.3436,
       "step": 1020
     },
     {
       "epoch": 0.19386410690758518,
-      "grad_norm": 3.861100673675537,
-      "learning_rate": 2.883681535855449e-05,
-      "loss": 3.3997,
+      "grad_norm": 4.3795366287231445,
+      "learning_rate": 0.0004806135893092415,
+      "loss": 1.3336,
       "step": 1030
     },
     {
       "epoch": 0.19574628270280445,
-      "grad_norm": 8.975924491882324,
-      "learning_rate": 2.8825522303783173e-05,
-      "loss": 3.297,
+      "grad_norm": 6.313399791717529,
+      "learning_rate": 0.0004804253717297196,
+      "loss": 0.9824,
       "step": 1040
     },
     {
       "epoch": 0.1976284584980237,
-      "grad_norm": 9.686234474182129,
-      "learning_rate": 2.881422924901186e-05,
-      "loss": 3.3386,
+      "grad_norm": 6.534666061401367,
+      "learning_rate": 0.00048023715415019766,
+      "loss": 1.3169,
       "step": 1050
     },
     {
       "epoch": 0.19951063429324298,
-      "grad_norm": 6.7616658210754395,
-      "learning_rate": 2.8802936194240542e-05,
-      "loss": 3.3721,
+      "grad_norm": 4.672704219818115,
+      "learning_rate": 0.00048004893657067573,
+      "loss": 1.1771,
       "step": 1060
     },
     {
       "epoch": 0.20139281008846227,
-      "grad_norm": 3.126573324203491,
-      "learning_rate": 2.8791643139469228e-05,
-      "loss": 3.4284,
+      "grad_norm": 1.6210001707077026,
+      "learning_rate": 0.0004798607189911538,
+      "loss": 1.2003,
       "step": 1070
     },
     {
       "epoch": 0.20327498588368154,
-      "grad_norm": 4.771273136138916,
-      "learning_rate": 2.8780350084697914e-05,
-      "loss": 3.2652,
+      "grad_norm": 2.4551827907562256,
+      "learning_rate": 0.0004796725014116319,
+      "loss": 1.3198,
       "step": 1080
     },
     {
       "epoch": 0.2051571616789008,
-      "grad_norm": 3.69577956199646,
-      "learning_rate": 2.8769057029926593e-05,
-      "loss": 3.3577,
+      "grad_norm": 4.158587455749512,
+      "learning_rate": 0.0004794842838321099,
+      "loss": 1.2115,
       "step": 1090
     },
     {
       "epoch": 0.2070393374741201,
-      "grad_norm": 4.15922212600708,
-      "learning_rate": 2.875776397515528e-05,
-      "loss": 3.4093,
+      "grad_norm": 2.0976548194885254,
+      "learning_rate": 0.000479296066252588,
+      "loss": 1.4164,
       "step": 1100
     },
     {
       "epoch": 0.20892151326933936,
-      "grad_norm": 15.523391723632812,
-      "learning_rate": 2.8746470920383962e-05,
-      "loss": 3.2856,
+      "grad_norm": 6.073217391967773,
+      "learning_rate": 0.00047910784867306606,
+      "loss": 1.0083,
       "step": 1110
     },
     {
       "epoch": 0.21080368906455862,
-      "grad_norm": 3.744535207748413,
-      "learning_rate": 2.8735177865612648e-05,
-      "loss": 3.344,
+      "grad_norm": 1.7033714056015015,
+      "learning_rate": 0.00047891963109354414,
+      "loss": 1.1871,
       "step": 1120
     },
     {
       "epoch": 0.2126858648597779,
-      "grad_norm": 4.708155632019043,
-      "learning_rate": 2.8723884810841334e-05,
-      "loss": 3.2039,
+      "grad_norm": 5.1013875007629395,
+      "learning_rate": 0.0004787314135140222,
+      "loss": 0.8595,
       "step": 1130
     },
     {
       "epoch": 0.21456804065499718,
-      "grad_norm": 3.991323232650757,
-      "learning_rate": 2.8712591756070017e-05,
-      "loss": 3.276,
+      "grad_norm": 5.065658092498779,
+      "learning_rate": 0.0004785431959345003,
+      "loss": 1.1723,
       "step": 1140
     },
     {
       "epoch": 0.21645021645021645,
-      "grad_norm": 6.3713154792785645,
-      "learning_rate": 2.8701298701298703e-05,
-      "loss": 3.3719,
+      "grad_norm": 2.3889753818511963,
+      "learning_rate": 0.00047835497835497837,
+      "loss": 1.3787,
       "step": 1150
     },
     {
       "epoch": 0.2183323922454357,
-      "grad_norm": 3.9468255043029785,
-      "learning_rate": 2.869000564652739e-05,
-      "loss": 3.2762,
+      "grad_norm": 2.41831374168396,
+      "learning_rate": 0.00047816676077545644,
+      "loss": 1.1492,
       "step": 1160
     },
     {
       "epoch": 0.220214568040655,
-      "grad_norm": 3.5731687545776367,
-      "learning_rate": 2.8678712591756072e-05,
-      "loss": 3.1112,
+      "grad_norm": 0.9150934815406799,
+      "learning_rate": 0.0004779785431959345,
+      "loss": 0.9076,
       "step": 1170
     },
     {
       "epoch": 0.22209674383587427,
-      "grad_norm": 5.240946292877197,
-      "learning_rate": 2.8667419536984754e-05,
-      "loss": 3.1546,
+      "grad_norm": 2.3671467304229736,
+      "learning_rate": 0.00047779032561641254,
+      "loss": 0.8078,
       "step": 1180
     },
     {
       "epoch": 0.22397891963109354,
-      "grad_norm": 3.877687931060791,
-      "learning_rate": 2.8656126482213437e-05,
-      "loss": 3.2343,
+      "grad_norm": 2.145510673522949,
+      "learning_rate": 0.0004776021080368906,
+      "loss": 1.1798,
       "step": 1190
     },
     {
       "epoch": 0.22586109542631283,
-      "grad_norm": 4.6856584548950195,
-      "learning_rate": 2.8644833427442123e-05,
-      "loss": 3.2184,
+      "grad_norm": 2.422679901123047,
+      "learning_rate": 0.0004774138904573687,
+      "loss": 1.2867,
       "step": 1200
     },
     {
       "epoch": 0.2277432712215321,
-      "grad_norm": 2.8222122192382812,
-      "learning_rate": 2.863354037267081e-05,
-      "loss": 3.1481,
+      "grad_norm": 1.5619584321975708,
+      "learning_rate": 0.00047722567287784677,
+      "loss": 0.9626,
       "step": 1210
     },
     {
       "epoch": 0.22962544701675136,
-      "grad_norm": 2.7809884548187256,
-      "learning_rate": 2.8622247317899492e-05,
-      "loss": 3.21,
+      "grad_norm": 1.2849690914154053,
+      "learning_rate": 0.0004770374552983249,
+      "loss": 1.196,
       "step": 1220
     },
     {
       "epoch": 0.23150762281197063,
-      "grad_norm": 4.449065208435059,
-      "learning_rate": 2.8610954263128178e-05,
-      "loss": 3.1756,
+      "grad_norm": 4.115751266479492,
+      "learning_rate": 0.000476849237718803,
+      "loss": 1.1092,
       "step": 1230
     },
     {
       "epoch": 0.23338979860718992,
-      "grad_norm": 5.27800989151001,
-      "learning_rate": 2.859966120835686e-05,
-      "loss": 3.1003,
+      "grad_norm": 3.9603044986724854,
+      "learning_rate": 0.00047666102013928105,
+      "loss": 1.1435,
       "step": 1240
     },
     {
       "epoch": 0.23527197440240918,
-      "grad_norm": 6.26723051071167,
-      "learning_rate": 2.8588368153585547e-05,
-      "loss": 3.3139,
+      "grad_norm": 3.748976230621338,
+      "learning_rate": 0.00047647280255975913,
+      "loss": 1.3499,
       "step": 1250
     },
     {
       "epoch": 0.23715415019762845,
-      "grad_norm": 10.930907249450684,
-      "learning_rate": 2.8577075098814233e-05,
-      "loss": 2.9757,
+      "grad_norm": 3.044076919555664,
+      "learning_rate": 0.0004762845849802372,
+      "loss": 0.8821,
       "step": 1260
     },
     {
       "epoch": 0.23903632599284774,
-      "grad_norm": 8.510289192199707,
-      "learning_rate": 2.8565782044042912e-05,
-      "loss": 3.2473,
+      "grad_norm": 3.245182514190674,
+      "learning_rate": 0.00047609636740071523,
+      "loss": 1.3718,
       "step": 1270
     },
     {
       "epoch": 0.240918501788067,
-      "grad_norm": 9.550591468811035,
-      "learning_rate": 2.8554488989271598e-05,
-      "loss": 3.0974,
+      "grad_norm": 2.004108428955078,
+      "learning_rate": 0.0004759081498211933,
+      "loss": 1.0094,
       "step": 1280
     },
     {
       "epoch": 0.24280067758328627,
-      "grad_norm": 3.702929735183716,
-      "learning_rate": 2.854319593450028e-05,
-      "loss": 3.0416,
+      "grad_norm": 3.3926773071289062,
+      "learning_rate": 0.0004757199322416714,
+      "loss": 1.1662,
       "step": 1290
     },
     {
       "epoch": 0.24468285337850557,
-      "grad_norm": 3.1774191856384277,
-      "learning_rate": 2.8531902879728967e-05,
-      "loss": 3.2329,
+      "grad_norm": 1.7959283590316772,
+      "learning_rate": 0.00047553171466214946,
+      "loss": 1.2912,
       "step": 1300
     },
     {
       "epoch": 0.24656502917372483,
-      "grad_norm": 3.9982802867889404,
-      "learning_rate": 2.8520609824957653e-05,
-      "loss": 3.145,
+      "grad_norm": 3.153946876525879,
+      "learning_rate": 0.00047534349708262753,
+      "loss": 1.0199,
       "step": 1310
     },
     {
       "epoch": 0.2484472049689441,
-      "grad_norm": 10.488585472106934,
-      "learning_rate": 2.8509316770186336e-05,
-      "loss": 3.1751,
+      "grad_norm": 3.7788596153259277,
+      "learning_rate": 0.0004751552795031056,
+      "loss": 1.4888,
       "step": 1320
     },
     {
       "epoch": 0.2503293807641634,
-      "grad_norm": 4.030777454376221,
-      "learning_rate": 2.8498023715415022e-05,
-      "loss": 3.0312,
+      "grad_norm": 2.140678644180298,
+      "learning_rate": 0.0004749670619235837,
+      "loss": 1.2057,
       "step": 1330
     },
     {
       "epoch": 0.25221155655938265,
-      "grad_norm": 5.116481304168701,
-      "learning_rate": 2.8486730660643705e-05,
-      "loss": 3.0881,
+      "grad_norm": 2.290731430053711,
+      "learning_rate": 0.00047477884434406176,
+      "loss": 1.2266,
       "step": 1340
     },
     {
       "epoch": 0.2540937323546019,
-      "grad_norm": 6.651054382324219,
-      "learning_rate": 2.847543760587239e-05,
-      "loss": 2.9569,
+      "grad_norm": 7.3565545082092285,
+      "learning_rate": 0.00047459062676453984,
+      "loss": 0.9182,
       "step": 1350
     },
     {
       "epoch": 0.2559759081498212,
-      "grad_norm": 7.519700050354004,
-      "learning_rate": 2.8464144551101073e-05,
-      "loss": 3.1106,
+      "grad_norm": 2.279402732849121,
+      "learning_rate": 0.00047440240918501786,
+      "loss": 1.1294,
       "step": 1360
     },
     {
       "epoch": 0.25785808394504045,
-      "grad_norm": 5.690998554229736,
-      "learning_rate": 2.8452851496329756e-05,
-      "loss": 3.176,
+      "grad_norm": 2.3557279109954834,
+      "learning_rate": 0.00047421419160549594,
+      "loss": 1.3733,
       "step": 1370
     },
     {
       "epoch": 0.2597402597402597,
-      "grad_norm": 7.928246021270752,
-      "learning_rate": 2.8441558441558442e-05,
-      "loss": 3.0882,
+      "grad_norm": 2.3519694805145264,
+      "learning_rate": 0.000474025974025974,
+      "loss": 0.7605,
       "step": 1380
     },
     {
       "epoch": 0.26162243553547904,
-      "grad_norm": 4.824883460998535,
-      "learning_rate": 2.8430265386787128e-05,
-      "loss": 3.0282,
+      "grad_norm": 5.51458215713501,
+      "learning_rate": 0.0004738377564464521,
+      "loss": 1.2157,
       "step": 1390
     },
     {
       "epoch": 0.2635046113306983,
-      "grad_norm": 5.6374640464782715,
-      "learning_rate": 2.841897233201581e-05,
-      "loss": 2.8817,
+      "grad_norm": 3.193882942199707,
+      "learning_rate": 0.00047364953886693017,
+      "loss": 0.8343,
       "step": 1400
     },
     {
       "epoch": 0.26538678712591757,
-      "grad_norm": 5.54218053817749,
-      "learning_rate": 2.8407679277244497e-05,
-      "loss": 3.0196,
+      "grad_norm": 2.5685150623321533,
+      "learning_rate": 0.00047346132128740824,
+      "loss": 0.8767,
       "step": 1410
     },
     {
       "epoch": 0.26726896292113683,
-      "grad_norm": 3.4347853660583496,
-      "learning_rate": 2.839638622247318e-05,
-      "loss": 2.9811,
+      "grad_norm": 2.659619092941284,
+      "learning_rate": 0.0004732731037078863,
+      "loss": 1.1711,
       "step": 1420
     },
     {
       "epoch": 0.2691511387163561,
-      "grad_norm": 3.8664019107818604,
-      "learning_rate": 2.8385093167701866e-05,
-      "loss": 2.9376,
+      "grad_norm": 2.564542055130005,
+      "learning_rate": 0.0004730848861283644,
+      "loss": 1.0881,
       "step": 1430
     },
     {
       "epoch": 0.27103331451157536,
-      "grad_norm": 5.806457996368408,
-      "learning_rate": 2.837380011293055e-05,
-      "loss": 3.1184,
+      "grad_norm": 2.108336925506592,
+      "learning_rate": 0.0004728966685488424,
+      "loss": 1.2762,
       "step": 1440
     },
     {
       "epoch": 0.27291549030679463,
-      "grad_norm": 4.753952980041504,
-      "learning_rate": 2.836250705815923e-05,
-      "loss": 2.9645,
+      "grad_norm": 2.0329091548919678,
+      "learning_rate": 0.00047270845096932055,
+      "loss": 0.9341,
       "step": 1450
     },
     {
       "epoch": 0.27479766610201395,
-      "grad_norm": 4.03913688659668,
-      "learning_rate": 2.8351214003387917e-05,
-      "loss": 2.8864,
+      "grad_norm": 1.2084460258483887,
+      "learning_rate": 0.0004725202333897986,
+      "loss": 0.7881,
       "step": 1460
     },
     {
       "epoch": 0.2766798418972332,
-      "grad_norm": 8.83630657196045,
-      "learning_rate": 2.83399209486166e-05,
-      "loss": 3.0768,
+      "grad_norm": 1.7741458415985107,
+      "learning_rate": 0.0004723320158102767,
+      "loss": 1.1409,
       "step": 1470
     },
     {
       "epoch": 0.2785620176924525,
-      "grad_norm": 7.6859540939331055,
-      "learning_rate": 2.8328627893845286e-05,
-      "loss": 2.915,
+      "grad_norm": 3.576479911804199,
+      "learning_rate": 0.0004721437982307548,
+      "loss": 1.0313,
       "step": 1480
     },
     {
       "epoch": 0.28044419348767174,
-      "grad_norm": 4.7729811668396,
-      "learning_rate": 2.8317334839073972e-05,
-      "loss": 2.9354,
+      "grad_norm": 4.221381664276123,
+      "learning_rate": 0.00047195558065123286,
+      "loss": 1.138,
       "step": 1490
     },
     {
       "epoch": 0.282326369282891,
-      "grad_norm": 6.917285442352295,
-      "learning_rate": 2.8306041784302655e-05,
-      "loss": 2.8416,
+      "grad_norm": 3.0888664722442627,
+      "learning_rate": 0.00047176736307171093,
+      "loss": 1.0203,
       "step": 1500
     },
     {
       "epoch": 0.2842085450781103,
-      "grad_norm": 4.168449878692627,
-      "learning_rate": 2.829474872953134e-05,
-      "loss": 3.0573,
+      "grad_norm": 3.0183537006378174,
+      "learning_rate": 0.000471579145492189,
+      "loss": 1.2671,
       "step": 1510
     },
     {
       "epoch": 0.2860907208733296,
-      "grad_norm": 3.04876708984375,
-      "learning_rate": 2.8283455674760023e-05,
-      "loss": 2.9579,
+      "grad_norm": 0.9898282289505005,
+      "learning_rate": 0.0004713909279126671,
+      "loss": 1.2029,
       "step": 1520
     },
     {
       "epoch": 0.28797289666854886,
-      "grad_norm": 6.365512847900391,
-      "learning_rate": 2.8272162619988706e-05,
-      "loss": 2.9621,
+      "grad_norm": 2.6439521312713623,
+      "learning_rate": 0.0004712027103331451,
+      "loss": 1.047,
       "step": 1530
     },
     {
       "epoch": 0.2898550724637681,
-      "grad_norm": 4.556066989898682,
-      "learning_rate": 2.8260869565217392e-05,
-      "loss": 2.9296,
+      "grad_norm": 2.931302070617676,
+      "learning_rate": 0.0004710144927536232,
+      "loss": 1.4159,
       "step": 1540
     },
     {
       "epoch": 0.2917372482589874,
-      "grad_norm": 2.86582088470459,
-      "learning_rate": 2.8249576510446075e-05,
-      "loss": 2.7293,
+      "grad_norm": 0.5969520807266235,
+      "learning_rate": 0.00047082627517410126,
+      "loss": 0.7143,
       "step": 1550
     },
     {
       "epoch": 0.29361942405420666,
-      "grad_norm": 4.107886791229248,
-      "learning_rate": 2.823828345567476e-05,
-      "loss": 3.0,
+      "grad_norm": 2.982952117919922,
+      "learning_rate": 0.00047063805759457934,
+      "loss": 1.2682,
       "step": 1560
     },
     {
       "epoch": 0.2955015998494259,
-      "grad_norm": 5.17814826965332,
-      "learning_rate": 2.8226990400903444e-05,
-      "loss": 2.9396,
+      "grad_norm": 4.021246433258057,
+      "learning_rate": 0.0004704498400150574,
+      "loss": 1.1403,
       "step": 1570
     },
     {
       "epoch": 0.2973837756446452,
-      "grad_norm": 3.8069369792938232,
-      "learning_rate": 2.821569734613213e-05,
-      "loss": 2.8673,
+      "grad_norm": 1.7243213653564453,
+      "learning_rate": 0.0004702616224355355,
+      "loss": 0.875,
       "step": 1580
     },
     {
       "epoch": 0.2992659514398645,
-      "grad_norm": 4.464517593383789,
-      "learning_rate": 2.8204404291360816e-05,
-      "loss": 2.8787,
+      "grad_norm": 1.9460840225219727,
+      "learning_rate": 0.00047007340485601357,
+      "loss": 1.1448,
       "step": 1590
     },
     {
       "epoch": 0.3011481272350838,
-      "grad_norm": 6.4331536293029785,
-      "learning_rate": 2.81931112365895e-05,
-      "loss": 2.7073,
+      "grad_norm": 3.757749080657959,
+      "learning_rate": 0.00046988518727649164,
+      "loss": 1.0793,
       "step": 1600
     },
     {
       "epoch": 0.30303030303030304,
-      "grad_norm": 5.987295627593994,
-      "learning_rate": 2.8181818181818185e-05,
-      "loss": 2.7962,
+      "grad_norm": 1.5117601156234741,
+      "learning_rate": 0.0004696969696969697,
+      "loss": 1.0534,
       "step": 1610
     },
     {
       "epoch": 0.3049124788255223,
-      "grad_norm": 5.483144760131836,
-      "learning_rate": 2.8170525127046864e-05,
-      "loss": 2.9668,
+      "grad_norm": 5.415975093841553,
+      "learning_rate": 0.00046950875211744774,
+      "loss": 1.2198,
       "step": 1620
     },
     {
       "epoch": 0.30679465462074157,
-      "grad_norm": 3.3265535831451416,
-      "learning_rate": 2.815923207227555e-05,
-      "loss": 2.8663,
+      "grad_norm": 3.129091739654541,
+      "learning_rate": 0.0004693205345379258,
+      "loss": 1.1713,
       "step": 1630
     },
     {
       "epoch": 0.30867683041596083,
-      "grad_norm": 6.472630500793457,
-      "learning_rate": 2.8147939017504236e-05,
-      "loss": 2.8595,
+      "grad_norm": 2.059892177581787,
+      "learning_rate": 0.0004691323169584039,
+      "loss": 1.0626,
       "step": 1640
     },
     {
       "epoch": 0.3105590062111801,
-      "grad_norm": 4.674870491027832,
-      "learning_rate": 2.813664596273292e-05,
-      "loss": 2.9178,
+      "grad_norm": 3.697364091873169,
+      "learning_rate": 0.00046894409937888197,
+      "loss": 1.1652,
       "step": 1650
     },
     {
       "epoch": 0.3124411820063994,
-      "grad_norm": 4.7341628074646,
-      "learning_rate": 2.8125352907961605e-05,
-      "loss": 2.7904,
+      "grad_norm": 3.6531107425689697,
+      "learning_rate": 0.0004687558817993601,
+      "loss": 1.1815,
       "step": 1660
     },
     {
       "epoch": 0.3143233578016187,
-      "grad_norm": 3.6063005924224854,
-      "learning_rate": 2.811405985319029e-05,
-      "loss": 2.8127,
+      "grad_norm": 2.3599324226379395,
+      "learning_rate": 0.0004685676642198382,
+      "loss": 1.1458,
       "step": 1670
     },
     {
       "epoch": 0.31620553359683795,
-      "grad_norm": 10.121143341064453,
-      "learning_rate": 2.8102766798418974e-05,
-      "loss": 2.8588,
+      "grad_norm": 1.7624262571334839,
+      "learning_rate": 0.00046837944664031625,
+      "loss": 1.018,
       "step": 1680
     },
     {
       "epoch": 0.3180877093920572,
-      "grad_norm": 5.036760330200195,
-      "learning_rate": 2.809147374364766e-05,
-      "loss": 2.7812,
+      "grad_norm": 0.9360486268997192,
+      "learning_rate": 0.00046819122906079433,
+      "loss": 1.07,
       "step": 1690
     },
     {
       "epoch": 0.3199698851872765,
-      "grad_norm": 8.621694564819336,
-      "learning_rate": 2.808018068887634e-05,
-      "loss": 2.8555,
+      "grad_norm": 4.804688453674316,
+      "learning_rate": 0.00046800301148127235,
+      "loss": 1.2024,
       "step": 1700
     },
     {
       "epoch": 0.32185206098249575,
-      "grad_norm": 6.541132926940918,
-      "learning_rate": 2.8068887634105025e-05,
-      "loss": 2.794,
+      "grad_norm": 2.2155728340148926,
+      "learning_rate": 0.00046781479390175043,
+      "loss": 1.0527,
       "step": 1710
     },
     {
       "epoch": 0.323734236777715,
-      "grad_norm": 5.8154425621032715,
-      "learning_rate": 2.805759457933371e-05,
-      "loss": 2.6459,
+      "grad_norm": 2.3131532669067383,
+      "learning_rate": 0.0004676265763222285,
+      "loss": 0.8434,
       "step": 1720
     },
     {
       "epoch": 0.32561641257293433,
-      "grad_norm": 3.9110445976257324,
-      "learning_rate": 2.8046301524562394e-05,
-      "loss": 2.7375,
+      "grad_norm": 6.129045486450195,
+      "learning_rate": 0.0004674383587427066,
+      "loss": 1.0509,
       "step": 1730
     },
     {
       "epoch": 0.3274985883681536,
-      "grad_norm": 3.47908878326416,
-      "learning_rate": 2.803500846979108e-05,
-      "loss": 2.8092,
+      "grad_norm": 2.4154810905456543,
+      "learning_rate": 0.00046725014116318466,
+      "loss": 1.0117,
       "step": 1740
     },
     {
       "epoch": 0.32938076416337286,
-      "grad_norm": 4.073140621185303,
-      "learning_rate": 2.8023715415019763e-05,
-      "loss": 2.7205,
+      "grad_norm": 1.0004245042800903,
+      "learning_rate": 0.00046706192358366273,
+      "loss": 1.0944,
       "step": 1750
     },
     {
       "epoch": 0.33126293995859213,
-      "grad_norm": 9.20858097076416,
-      "learning_rate": 2.801242236024845e-05,
-      "loss": 2.6537,
+      "grad_norm": 2.5676519870758057,
+      "learning_rate": 0.0004668737060041408,
+      "loss": 0.9929,
       "step": 1760
     },
     {
       "epoch": 0.3331451157538114,
-      "grad_norm": 3.811323881149292,
-      "learning_rate": 2.8001129305477135e-05,
-      "loss": 2.7925,
+      "grad_norm": 3.247866153717041,
+      "learning_rate": 0.0004666854884246189,
+      "loss": 1.089,
       "step": 1770
     },
     {
       "epoch": 0.33502729154903066,
-      "grad_norm": 3.6044814586639404,
-      "learning_rate": 2.7989836250705817e-05,
-      "loss": 2.7287,
+      "grad_norm": 2.009756088256836,
+      "learning_rate": 0.00046649727084509696,
+      "loss": 1.0313,
       "step": 1780
     },
     {
       "epoch": 0.33690946734425,
-      "grad_norm": 5.667727470397949,
-      "learning_rate": 2.79785431959345e-05,
-      "loss": 2.5265,
+      "grad_norm": 1.9623408317565918,
+      "learning_rate": 0.000466309053265575,
+      "loss": 0.7833,
       "step": 1790
     },
     {
       "epoch": 0.33879164313946925,
-      "grad_norm": 4.497561931610107,
-      "learning_rate": 2.7967250141163183e-05,
-      "loss": 2.5919,
+      "grad_norm": 1.7636711597442627,
+      "learning_rate": 0.00046612083568605306,
+      "loss": 0.7889,
       "step": 1800
     },
     {
       "epoch": 0.3406738189346885,
-      "grad_norm": 6.864144802093506,
-      "learning_rate": 2.795595708639187e-05,
-      "loss": 2.7344,
+      "grad_norm": 3.7700984477996826,
+      "learning_rate": 0.00046593261810653114,
+      "loss": 1.3347,
       "step": 1810
     },
     {
       "epoch": 0.3425559947299078,
-      "grad_norm": 5.193475723266602,
-      "learning_rate": 2.7944664031620555e-05,
-      "loss": 2.7341,
+      "grad_norm": 1.4358450174331665,
+      "learning_rate": 0.0004657444005270092,
+      "loss": 0.9798,
       "step": 1820
     },
     {
       "epoch": 0.34443817052512704,
-      "grad_norm": 10.060043334960938,
-      "learning_rate": 2.7933370976849238e-05,
-      "loss": 2.6031,
+      "grad_norm": 4.207058906555176,
+      "learning_rate": 0.0004655561829474873,
+      "loss": 1.0304,
       "step": 1830
     },
     {
       "epoch": 0.3463203463203463,
-      "grad_norm": 7.921087265014648,
-      "learning_rate": 2.7922077922077924e-05,
-      "loss": 2.6902,
+      "grad_norm": 2.066248655319214,
+      "learning_rate": 0.00046536796536796537,
+      "loss": 0.9523,
       "step": 1840
     },
     {
       "epoch": 0.34820252211556557,
-      "grad_norm": 5.25139856338501,
-      "learning_rate": 2.7910784867306606e-05,
-      "loss": 2.7383,
+      "grad_norm": 2.937966823577881,
+      "learning_rate": 0.00046517974778844344,
+      "loss": 1.2679,
       "step": 1850
     },
     {
       "epoch": 0.3500846979107849,
-      "grad_norm": 9.80843734741211,
-      "learning_rate": 2.7899491812535292e-05,
-      "loss": 2.5868,
+      "grad_norm": 4.462133407592773,
+      "learning_rate": 0.0004649915302089215,
+      "loss": 0.9922,
       "step": 1860
     },
     {
       "epoch": 0.35196687370600416,
-      "grad_norm": 8.038129806518555,
-      "learning_rate": 2.788819875776398e-05,
-      "loss": 2.6574,
+      "grad_norm": 2.355633020401001,
+      "learning_rate": 0.0004648033126293996,
+      "loss": 1.0605,
       "step": 1870
     },
     {
       "epoch": 0.3538490495012234,
-      "grad_norm": 5.235549449920654,
-      "learning_rate": 2.7876905702992658e-05,
-      "loss": 2.623,
+      "grad_norm": 3.410402297973633,
+      "learning_rate": 0.0004646150950498776,
+      "loss": 1.1485,
       "step": 1880
     },
     {
       "epoch": 0.3557312252964427,
-      "grad_norm": 3.0741400718688965,
-      "learning_rate": 2.7865612648221344e-05,
-      "loss": 2.6216,
+      "grad_norm": 4.6109113693237305,
+      "learning_rate": 0.00046442687747035575,
+      "loss": 1.2647,
       "step": 1890
     },
     {
       "epoch": 0.35761340109166195,
-      "grad_norm": 3.884474515914917,
-      "learning_rate": 2.785431959345003e-05,
-      "loss": 2.5854,
+      "grad_norm": 3.604688882827759,
+      "learning_rate": 0.0004642386598908338,
+      "loss": 0.8707,
       "step": 1900
     },
     {
       "epoch": 0.3594955768868812,
-      "grad_norm": 3.7831873893737793,
-      "learning_rate": 2.7843026538678713e-05,
-      "loss": 2.3631,
+      "grad_norm": 5.367961406707764,
+      "learning_rate": 0.0004640504423113119,
+      "loss": 0.9146,
       "step": 1910
     },
     {
       "epoch": 0.3613777526821005,
-      "grad_norm": 6.738157272338867,
-      "learning_rate": 2.78317334839074e-05,
-      "loss": 2.6551,
+      "grad_norm": 2.2205069065093994,
+      "learning_rate": 0.00046386222473179,
+      "loss": 1.1418,
       "step": 1920
     },
     {
       "epoch": 0.3632599284773198,
-      "grad_norm": 5.1181793212890625,
-      "learning_rate": 2.782044042913608e-05,
-      "loss": 2.6403,
+      "grad_norm": 3.098167896270752,
+      "learning_rate": 0.00046367400715226806,
+      "loss": 1.1865,
       "step": 1930
     },
     {
       "epoch": 0.36514210427253907,
-      "grad_norm": 8.148448944091797,
-      "learning_rate": 2.7809147374364768e-05,
-      "loss": 2.4368,
+      "grad_norm": 1.8569923639297485,
+      "learning_rate": 0.00046348578957274613,
+      "loss": 0.7607,
       "step": 1940
     },
     {
       "epoch": 0.36702428006775834,
-      "grad_norm": 14.908823013305664,
-      "learning_rate": 2.7797854319593454e-05,
-      "loss": 2.7317,
+      "grad_norm": 4.62168025970459,
+      "learning_rate": 0.0004632975719932242,
+      "loss": 1.2438,
       "step": 1950
     },
     {
       "epoch": 0.3689064558629776,
-      "grad_norm": 4.497227668762207,
-      "learning_rate": 2.7786561264822133e-05,
-      "loss": 2.5193,
+      "grad_norm": 4.406530380249023,
+      "learning_rate": 0.00046310935441370223,
+      "loss": 0.9185,
       "step": 1960
     },
     {
       "epoch": 0.37078863165819687,
-      "grad_norm": 7.139688491821289,
-      "learning_rate": 2.777526821005082e-05,
-      "loss": 2.6937,
+      "grad_norm": 2.9073426723480225,
+      "learning_rate": 0.0004629211368341803,
+      "loss": 0.9801,
       "step": 1970
     },
     {
       "epoch": 0.37267080745341613,
-      "grad_norm": 11.658743858337402,
-      "learning_rate": 2.77639751552795e-05,
-      "loss": 2.3396,
+      "grad_norm": 7.176252365112305,
+      "learning_rate": 0.0004627329192546584,
+      "loss": 0.7723,
       "step": 1980
     },
     {
       "epoch": 0.3745529832486354,
-      "grad_norm": 11.21719741821289,
-      "learning_rate": 2.7752682100508188e-05,
-      "loss": 2.6433,
+      "grad_norm": 2.731356382369995,
+      "learning_rate": 0.00046254470167513646,
+      "loss": 1.2079,
       "step": 1990
     },
     {
       "epoch": 0.3764351590438547,
-      "grad_norm": 3.650562047958374,
-      "learning_rate": 2.7741389045736874e-05,
-      "loss": 2.5423,
+      "grad_norm": 1.2026039361953735,
+      "learning_rate": 0.00046235648409561454,
+      "loss": 1.0573,
       "step": 2000
     },
     {
       "epoch": 0.378317334839074,
-      "grad_norm": 6.978822231292725,
-      "learning_rate": 2.7730095990965556e-05,
-      "loss": 2.4149,
+      "grad_norm": 3.050309419631958,
+      "learning_rate": 0.0004621682665160926,
+      "loss": 0.9802,
       "step": 2010
     },
     {
       "epoch": 0.38019951063429325,
-      "grad_norm": 12.115926742553711,
-      "learning_rate": 2.7718802936194243e-05,
-      "loss": 2.6864,
+      "grad_norm": 3.7857017517089844,
+      "learning_rate": 0.0004619800489365707,
+      "loss": 1.2937,
       "step": 2020
     },
     {
       "epoch": 0.3820816864295125,
-      "grad_norm": 12.849691390991211,
-      "learning_rate": 2.7707509881422925e-05,
-      "loss": 2.4976,
+      "grad_norm": 3.345871925354004,
+      "learning_rate": 0.00046179183135704877,
+      "loss": 1.0949,
       "step": 2030
     },
     {
       "epoch": 0.3839638622247318,
-      "grad_norm": 7.091000556945801,
-      "learning_rate": 2.769621682665161e-05,
-      "loss": 2.3657,
+      "grad_norm": 3.366966485977173,
+      "learning_rate": 0.00046160361377752684,
+      "loss": 0.9634,
       "step": 2040
     },
     {
       "epoch": 0.38584603801995104,
-      "grad_norm": 7.3910746574401855,
-      "learning_rate": 2.7684923771880294e-05,
-      "loss": 2.4003,
+      "grad_norm": 2.2242603302001953,
+      "learning_rate": 0.00046141539619800486,
+      "loss": 0.8493,
       "step": 2050
     },
     {
       "epoch": 0.38772821381517036,
-      "grad_norm": 4.848266124725342,
-      "learning_rate": 2.7673630717108977e-05,
-      "loss": 2.6143,
+      "grad_norm": 2.916806221008301,
+      "learning_rate": 0.00046122717861848294,
+      "loss": 1.1871,
       "step": 2060
     },
     {
       "epoch": 0.38961038961038963,
-      "grad_norm": 5.94016170501709,
-      "learning_rate": 2.7662337662337663e-05,
-      "loss": 2.4341,
+      "grad_norm": 4.017305850982666,
+      "learning_rate": 0.000461038961038961,
+      "loss": 0.925,
       "step": 2070
     },
     {
       "epoch": 0.3914925654056089,
-      "grad_norm": 8.351221084594727,
-      "learning_rate": 2.7651044607566345e-05,
-      "loss": 2.4935,
+      "grad_norm": 3.971277952194214,
+      "learning_rate": 0.0004608507434594391,
+      "loss": 0.962,
       "step": 2080
     },
     {
       "epoch": 0.39337474120082816,
-      "grad_norm": 8.837783813476562,
-      "learning_rate": 2.763975155279503e-05,
-      "loss": 2.3477,
+      "grad_norm": 4.612211227416992,
+      "learning_rate": 0.00046066252587991717,
+      "loss": 0.8731,
       "step": 2090
     },
     {
       "epoch": 0.3952569169960474,
-      "grad_norm": 9.10621166229248,
-      "learning_rate": 2.7628458498023718e-05,
-      "loss": 2.4406,
+      "grad_norm": 2.1569032669067383,
+      "learning_rate": 0.00046047430830039525,
+      "loss": 0.839,
       "step": 2100
     },
     {
       "epoch": 0.3971390927912667,
-      "grad_norm": 9.467292785644531,
-      "learning_rate": 2.76171654432524e-05,
-      "loss": 2.6356,
+      "grad_norm": 2.8226122856140137,
+      "learning_rate": 0.0004602860907208734,
+      "loss": 1.1247,
       "step": 2110
     },
     {
       "epoch": 0.39902126858648596,
-      "grad_norm": 5.230591297149658,
-      "learning_rate": 2.7605872388481086e-05,
-      "loss": 2.4345,
+      "grad_norm": 5.517162322998047,
+      "learning_rate": 0.00046009787314135145,
+      "loss": 1.1889,
       "step": 2120
     },
     {
       "epoch": 0.4009034443817053,
-      "grad_norm": 5.1924591064453125,
-      "learning_rate": 2.759457933370977e-05,
-      "loss": 2.5366,
+      "grad_norm": 1.8455901145935059,
+      "learning_rate": 0.00045990965556182953,
+      "loss": 0.9728,
       "step": 2130
     },
     {
       "epoch": 0.40278562017692454,
-      "grad_norm": 6.529165744781494,
-      "learning_rate": 2.7583286278938452e-05,
-      "loss": 2.3564,
+      "grad_norm": 3.769131660461426,
+      "learning_rate": 0.00045972143798230755,
+      "loss": 0.7964,
       "step": 2140
     },
     {
       "epoch": 0.4046677959721438,
-      "grad_norm": 7.249635219573975,
-      "learning_rate": 2.7571993224167138e-05,
-      "loss": 2.2548,
+      "grad_norm": 2.248361825942993,
+      "learning_rate": 0.00045953322040278563,
+      "loss": 0.6952,
       "step": 2150
     },
     {
       "epoch": 0.40654997176736307,
-      "grad_norm": 9.388668060302734,
-      "learning_rate": 2.756070016939582e-05,
-      "loss": 2.6006,
+      "grad_norm": 3.3281447887420654,
+      "learning_rate": 0.0004593450028232637,
+      "loss": 1.1356,
       "step": 2160
     },
     {
       "epoch": 0.40843214756258234,
-      "grad_norm": 3.7543675899505615,
-      "learning_rate": 2.7549407114624507e-05,
-      "loss": 2.3548,
+      "grad_norm": 1.2701328992843628,
+      "learning_rate": 0.0004591567852437418,
+      "loss": 0.9366,
       "step": 2170
     },
     {
       "epoch": 0.4103143233578016,
-      "grad_norm": 7.2959442138671875,
-      "learning_rate": 2.7538114059853193e-05,
-      "loss": 2.4884,
+      "grad_norm": 2.7324378490448,
+      "learning_rate": 0.00045896856766421986,
+      "loss": 0.9716,
       "step": 2180
     },
     {
       "epoch": 0.41219649915302087,
-      "grad_norm": 4.733274936676025,
-      "learning_rate": 2.7526821005081875e-05,
-      "loss": 2.5703,
+      "grad_norm": 1.8330858945846558,
+      "learning_rate": 0.00045878035008469793,
+      "loss": 1.2041,
       "step": 2190
     },
     {
       "epoch": 0.4140786749482402,
-      "grad_norm": 14.75526237487793,
-      "learning_rate": 2.751552795031056e-05,
-      "loss": 2.4666,
+      "grad_norm": 4.173506736755371,
+      "learning_rate": 0.000458592132505176,
+      "loss": 0.8957,
       "step": 2200
     },
     {
       "epoch": 0.41596085074345945,
-      "grad_norm": 5.791693210601807,
-      "learning_rate": 2.7504234895539244e-05,
-      "loss": 2.4131,
+      "grad_norm": 2.277362108230591,
+      "learning_rate": 0.0004584039149256541,
+      "loss": 0.9527,
       "step": 2210
     },
     {
       "epoch": 0.4178430265386787,
-      "grad_norm": 5.518586158752441,
-      "learning_rate": 2.749294184076793e-05,
-      "loss": 2.2924,
+      "grad_norm": 2.5186853408813477,
+      "learning_rate": 0.00045821569734613216,
+      "loss": 0.94,
       "step": 2220
     },
     {
       "epoch": 0.419725202333898,
-      "grad_norm": 8.407965660095215,
-      "learning_rate": 2.7481648785996613e-05,
-      "loss": 2.2358,
+      "grad_norm": 3.54046893119812,
+      "learning_rate": 0.0004580274797666102,
+      "loss": 0.722,
       "step": 2230
     },
     {
       "epoch": 0.42160737812911725,
-      "grad_norm": 6.686824798583984,
-      "learning_rate": 2.7470355731225296e-05,
-      "loss": 2.4671,
+      "grad_norm": 4.817910671234131,
+      "learning_rate": 0.00045783926218708826,
+      "loss": 1.0113,
       "step": 2240
     },
     {
       "epoch": 0.4234895539243365,
-      "grad_norm": 5.854467391967773,
-      "learning_rate": 2.745906267645398e-05,
-      "loss": 2.3917,
+      "grad_norm": 2.397691011428833,
+      "learning_rate": 0.00045765104460756634,
+      "loss": 0.9322,
       "step": 2250
     },
     {
       "epoch": 0.4253717297195558,
-      "grad_norm": 9.136099815368652,
-      "learning_rate": 2.7447769621682664e-05,
-      "loss": 2.2994,
+      "grad_norm": 2.3579189777374268,
+      "learning_rate": 0.0004574628270280444,
+      "loss": 1.1121,
       "step": 2260
     },
     {
       "epoch": 0.4272539055147751,
-      "grad_norm": 4.834329128265381,
-      "learning_rate": 2.743647656691135e-05,
-      "loss": 2.1033,
+      "grad_norm": 6.9814453125,
+      "learning_rate": 0.0004572746094485225,
+      "loss": 0.6762,
       "step": 2270
     },
     {
       "epoch": 0.42913608130999437,
-      "grad_norm": 5.209042072296143,
-      "learning_rate": 2.7425183512140036e-05,
-      "loss": 2.4016,
+      "grad_norm": 4.888842582702637,
+      "learning_rate": 0.00045708639186900057,
+      "loss": 0.8394,
       "step": 2280
     },
     {
       "epoch": 0.43101825710521363,
-      "grad_norm": 11.149210929870605,
-      "learning_rate": 2.741389045736872e-05,
-      "loss": 2.3709,
+      "grad_norm": 4.279430389404297,
+      "learning_rate": 0.00045689817428947864,
+      "loss": 0.9319,
       "step": 2290
     },
     {
       "epoch": 0.4329004329004329,
-      "grad_norm": 12.44275188446045,
-      "learning_rate": 2.7402597402597405e-05,
-      "loss": 2.3464,
+      "grad_norm": 7.005417823791504,
+      "learning_rate": 0.0004567099567099567,
+      "loss": 0.8899,
       "step": 2300
     },
     {
       "epoch": 0.43478260869565216,
-      "grad_norm": 12.738330841064453,
-      "learning_rate": 2.7391304347826085e-05,
-      "loss": 2.1641,
+      "grad_norm": 3.8382112979888916,
+      "learning_rate": 0.00045652173913043474,
+      "loss": 0.8132,
       "step": 2310
     },
     {
       "epoch": 0.4366647844908714,
-      "grad_norm": 8.074936866760254,
-      "learning_rate": 2.738001129305477e-05,
-      "loss": 2.1561,
+      "grad_norm": 4.163649559020996,
+      "learning_rate": 0.0004563335215509128,
+      "loss": 0.7774,
       "step": 2320
     },
     {
       "epoch": 0.43854696028609075,
-      "grad_norm": 5.971907615661621,
-      "learning_rate": 2.7368718238283457e-05,
-      "loss": 2.2271,
+      "grad_norm": 2.5022313594818115,
+      "learning_rate": 0.00045614530397139095,
+      "loss": 0.9465,
       "step": 2330
     },
     {
       "epoch": 0.44042913608131,
-      "grad_norm": 3.5811924934387207,
-      "learning_rate": 2.735742518351214e-05,
-      "loss": 2.2939,
+      "grad_norm": 2.3390891551971436,
+      "learning_rate": 0.000455957086391869,
+      "loss": 1.1819,
       "step": 2340
     },
     {
       "epoch": 0.4423113118765293,
-      "grad_norm": 7.316919326782227,
-      "learning_rate": 2.7346132128740825e-05,
-      "loss": 2.2452,
+      "grad_norm": 4.921488285064697,
+      "learning_rate": 0.0004557688688123471,
+      "loss": 1.0202,
       "step": 2350
     },
     {
       "epoch": 0.44419348767174854,
-      "grad_norm": 12.880569458007812,
-      "learning_rate": 2.7334839073969508e-05,
-      "loss": 2.2113,
+      "grad_norm": 3.4981935024261475,
+      "learning_rate": 0.0004555806512328252,
+      "loss": 0.8734,
       "step": 2360
     },
     {
       "epoch": 0.4460756634669678,
-      "grad_norm": 7.582205295562744,
-      "learning_rate": 2.7323546019198194e-05,
-      "loss": 2.3516,
+      "grad_norm": 2.6764838695526123,
+      "learning_rate": 0.00045539243365330326,
+      "loss": 0.8693,
       "step": 2370
     },
     {
       "epoch": 0.4479578392621871,
-      "grad_norm": 5.996733665466309,
-      "learning_rate": 2.731225296442688e-05,
-      "loss": 2.17,
+      "grad_norm": 2.6041977405548096,
+      "learning_rate": 0.00045520421607378133,
+      "loss": 0.773,
       "step": 2380
     },
     {
       "epoch": 0.44984001505740634,
-      "grad_norm": 4.608865737915039,
-      "learning_rate": 2.7300959909655563e-05,
-      "loss": 2.3382,
+      "grad_norm": 2.146116018295288,
+      "learning_rate": 0.0004550159984942594,
+      "loss": 0.981,
       "step": 2390
     },
     {
       "epoch": 0.45172219085262566,
-      "grad_norm": 8.942387580871582,
-      "learning_rate": 2.7289666854884246e-05,
-      "loss": 2.4042,
+      "grad_norm": 1.5882443189620972,
+      "learning_rate": 0.00045482778091473743,
+      "loss": 1.2898,
       "step": 2400
     },
     {
       "epoch": 0.4536043666478449,
-      "grad_norm": 5.910847187042236,
-      "learning_rate": 2.7278373800112932e-05,
-      "loss": 2.2939,
+      "grad_norm": 2.2083537578582764,
+      "learning_rate": 0.0004546395633352155,
+      "loss": 1.0654,
       "step": 2410
     },
     {
       "epoch": 0.4554865424430642,
-      "grad_norm": 7.828692436218262,
-      "learning_rate": 2.7267080745341614e-05,
-      "loss": 2.244,
+      "grad_norm": 3.9478726387023926,
+      "learning_rate": 0.0004544513457556936,
+      "loss": 0.8267,
       "step": 2420
     },
     {
       "epoch": 0.45736871823828346,
-      "grad_norm": 6.50481653213501,
-      "learning_rate": 2.72557876905703e-05,
-      "loss": 2.2278,
+      "grad_norm": 3.5692625045776367,
+      "learning_rate": 0.00045426312817617166,
+      "loss": 0.8909,
       "step": 2430
     },
     {
       "epoch": 0.4592508940335027,
-      "grad_norm": 6.783820629119873,
-      "learning_rate": 2.7244494635798983e-05,
-      "loss": 2.2561,
+      "grad_norm": 4.099732398986816,
+      "learning_rate": 0.00045407491059664974,
+      "loss": 0.9475,
       "step": 2440
     },
     {
       "epoch": 0.461133069828722,
-      "grad_norm": 5.222927570343018,
-      "learning_rate": 2.723320158102767e-05,
-      "loss": 2.1948,
+      "grad_norm": 2.3055837154388428,
+      "learning_rate": 0.0004538866930171278,
+      "loss": 0.9427,
       "step": 2450
     },
     {
       "epoch": 0.46301524562394125,
-      "grad_norm": 8.41578197479248,
-      "learning_rate": 2.7221908526256355e-05,
-      "loss": 2.1365,
+      "grad_norm": 1.5885486602783203,
+      "learning_rate": 0.0004536984754376059,
+      "loss": 0.8121,
       "step": 2460
     },
     {
       "epoch": 0.4648974214191606,
-      "grad_norm": 7.914490222930908,
-      "learning_rate": 2.7210615471485038e-05,
-      "loss": 2.0912,
+      "grad_norm": 2.3060200214385986,
+      "learning_rate": 0.00045351025785808397,
+      "loss": 0.8262,
       "step": 2470
     },
     {
       "epoch": 0.46677959721437984,
-      "grad_norm": 10.265892028808594,
-      "learning_rate": 2.7199322416713724e-05,
-      "loss": 2.2027,
+      "grad_norm": 2.7296106815338135,
+      "learning_rate": 0.00045332204027856204,
+      "loss": 1.0472,
       "step": 2480
     },
     {
       "epoch": 0.4686617730095991,
-      "grad_norm": 5.603802680969238,
-      "learning_rate": 2.7188029361942403e-05,
-      "loss": 2.1609,
+      "grad_norm": 1.5217969417572021,
+      "learning_rate": 0.00045313382269904006,
+      "loss": 0.8783,
       "step": 2490
     },
     {
       "epoch": 0.47054394880481837,
-      "grad_norm": 11.390222549438477,
-      "learning_rate": 2.717673630717109e-05,
-      "loss": 2.1755,
+      "grad_norm": 2.4178285598754883,
+      "learning_rate": 0.00045294560511951814,
+      "loss": 1.0481,
       "step": 2500
     },
     {
       "epoch": 0.47242612460003763,
-      "grad_norm": 4.981803894042969,
-      "learning_rate": 2.7165443252399776e-05,
-      "loss": 1.873,
+      "grad_norm": 2.3695430755615234,
+      "learning_rate": 0.0004527573875399962,
+      "loss": 0.4676,
       "step": 2510
     },
     {
       "epoch": 0.4743083003952569,
-      "grad_norm": 6.110774040222168,
-      "learning_rate": 2.7154150197628458e-05,
-      "loss": 2.0585,
+      "grad_norm": 1.467848777770996,
+      "learning_rate": 0.0004525691699604743,
+      "loss": 0.7778,
       "step": 2520
     },
     {
       "epoch": 0.47619047619047616,
-      "grad_norm": 6.286103248596191,
-      "learning_rate": 2.7142857142857144e-05,
-      "loss": 1.9113,
+      "grad_norm": 2.409031629562378,
+      "learning_rate": 0.00045238095238095237,
+      "loss": 0.7023,
       "step": 2530
     },
     {
       "epoch": 0.4780726519856955,
-      "grad_norm": 5.199717044830322,
-      "learning_rate": 2.7131564088085827e-05,
-      "loss": 2.0643,
+      "grad_norm": 1.9384913444519043,
+      "learning_rate": 0.00045219273480143045,
+      "loss": 0.8401,
       "step": 2540
     },
     {
       "epoch": 0.47995482778091475,
-      "grad_norm": 10.684744834899902,
-      "learning_rate": 2.7120271033314513e-05,
-      "loss": 2.1255,
+      "grad_norm": 1.967634916305542,
+      "learning_rate": 0.0004520045172219086,
+      "loss": 0.7964,
       "step": 2550
     },
     {
       "epoch": 0.481837003576134,
-      "grad_norm": 6.335547924041748,
-      "learning_rate": 2.71089779785432e-05,
-      "loss": 2.0588,
+      "grad_norm": 2.215298652648926,
+      "learning_rate": 0.00045181629964238665,
+      "loss": 0.8712,
       "step": 2560
     },
     {
       "epoch": 0.4837191793713533,
-      "grad_norm": 15.644701957702637,
-      "learning_rate": 2.709768492377188e-05,
-      "loss": 2.2138,
+      "grad_norm": 5.0799078941345215,
+      "learning_rate": 0.0004516280820628647,
+      "loss": 1.0858,
       "step": 2570
     },
     {
       "epoch": 0.48560135516657255,
-      "grad_norm": 8.629146575927734,
-      "learning_rate": 2.7086391869000565e-05,
-      "loss": 1.8363,
+      "grad_norm": 4.027543067932129,
+      "learning_rate": 0.00045143986448334275,
+      "loss": 0.5627,
       "step": 2580
     },
     {
       "epoch": 0.4874835309617918,
-      "grad_norm": 6.585087299346924,
-      "learning_rate": 2.7075098814229247e-05,
-      "loss": 2.096,
+      "grad_norm": 2.23659348487854,
+      "learning_rate": 0.00045125164690382083,
+      "loss": 0.9524,
       "step": 2590
     },
     {
       "epoch": 0.48936570675701113,
-      "grad_norm": 4.812081336975098,
-      "learning_rate": 2.7063805759457933e-05,
-      "loss": 2.0322,
+      "grad_norm": 2.323295831680298,
+      "learning_rate": 0.0004510634293242989,
+      "loss": 0.915,
       "step": 2600
     },
     {
       "epoch": 0.4912478825522304,
-      "grad_norm": 3.4937989711761475,
-      "learning_rate": 2.705251270468662e-05,
-      "loss": 2.2819,
+      "grad_norm": 1.380283236503601,
+      "learning_rate": 0.000450875211744777,
+      "loss": 0.9743,
       "step": 2610
     },
     {
       "epoch": 0.49313005834744966,
-      "grad_norm": 13.015551567077637,
-      "learning_rate": 2.7041219649915302e-05,
-      "loss": 2.0931,
+      "grad_norm": 2.988337516784668,
+      "learning_rate": 0.00045068699416525506,
+      "loss": 0.8426,
       "step": 2620
     },
     {
       "epoch": 0.4950122341426689,
-      "grad_norm": 26.068567276000977,
-      "learning_rate": 2.7029926595143988e-05,
-      "loss": 2.1267,
+      "grad_norm": 4.08306884765625,
+      "learning_rate": 0.00045049877658573313,
+      "loss": 0.8518,
       "step": 2630
     },
     {
       "epoch": 0.4968944099378882,
-      "grad_norm": 6.577176094055176,
-      "learning_rate": 2.701863354037267e-05,
-      "loss": 1.9937,
+      "grad_norm": 2.828056812286377,
+      "learning_rate": 0.0004503105590062112,
+      "loss": 0.8582,
       "step": 2640
     },
     {
       "epoch": 0.49877658573310746,
-      "grad_norm": 8.388608932495117,
-      "learning_rate": 2.7007340485601357e-05,
-      "loss": 2.1152,
+      "grad_norm": 1.375273585319519,
+      "learning_rate": 0.0004501223414266893,
+      "loss": 0.8976,
       "step": 2650
     },
     {
       "epoch": 0.5006587615283268,
-      "grad_norm": 5.307988166809082,
-      "learning_rate": 2.699604743083004e-05,
-      "loss": 2.0221,
+      "grad_norm": 4.221978187561035,
+      "learning_rate": 0.0004499341238471673,
+      "loss": 0.9568,
       "step": 2660
     },
     {
       "epoch": 0.502540937323546,
-      "grad_norm": 7.120198726654053,
-      "learning_rate": 2.6984754376058722e-05,
-      "loss": 2.0487,
+      "grad_norm": 2.1360394954681396,
+      "learning_rate": 0.0004497459062676454,
+      "loss": 0.8552,
       "step": 2670
     },
     {
       "epoch": 0.5044231131187653,
-      "grad_norm": 5.206243991851807,
-      "learning_rate": 2.697346132128741e-05,
-      "loss": 2.112,
+      "grad_norm": 3.500844955444336,
+      "learning_rate": 0.00044955768868812346,
+      "loss": 0.9702,
       "step": 2680
     },
     {
       "epoch": 0.5063052889139845,
-      "grad_norm": 12.726581573486328,
-      "learning_rate": 2.6962168266516094e-05,
-      "loss": 2.1619,
+      "grad_norm": 3.776733636856079,
+      "learning_rate": 0.00044936947110860154,
+      "loss": 1.0864,
       "step": 2690
     },
     {
       "epoch": 0.5081874647092038,
-      "grad_norm": 5.604727745056152,
-      "learning_rate": 2.6950875211744777e-05,
-      "loss": 1.8609,
+      "grad_norm": 3.072545289993286,
+      "learning_rate": 0.0004491812535290796,
+      "loss": 0.8421,
       "step": 2700
     },
     {
       "epoch": 0.5100696405044232,
-      "grad_norm": 4.6965012550354,
-      "learning_rate": 2.6939582156973463e-05,
-      "loss": 2.095,
+      "grad_norm": 0.5951893925666809,
+      "learning_rate": 0.0004489930359495577,
+      "loss": 0.9968,
       "step": 2710
     },
     {
       "epoch": 0.5119518162996424,
-      "grad_norm": 6.7473602294921875,
-      "learning_rate": 2.6928289102202146e-05,
-      "loss": 2.0439,
+      "grad_norm": 4.747662544250488,
+      "learning_rate": 0.00044880481837003577,
+      "loss": 0.9355,
       "step": 2720
     },
     {
       "epoch": 0.5138339920948617,
-      "grad_norm": 9.199115753173828,
-      "learning_rate": 2.6916996047430832e-05,
-      "loss": 2.1664,
+      "grad_norm": 3.932286024093628,
+      "learning_rate": 0.00044861660079051384,
+      "loss": 1.321,
       "step": 2730
     },
     {
       "epoch": 0.5157161678900809,
-      "grad_norm": 8.569664001464844,
-      "learning_rate": 2.6905702992659518e-05,
-      "loss": 1.9945,
+      "grad_norm": 2.5157997608184814,
+      "learning_rate": 0.0004484283832109919,
+      "loss": 0.8607,
       "step": 2740
     },
     {
       "epoch": 0.5175983436853002,
-      "grad_norm": 13.89288330078125,
-      "learning_rate": 2.6894409937888197e-05,
-      "loss": 1.9975,
+      "grad_norm": 1.6060855388641357,
+      "learning_rate": 0.00044824016563146994,
+      "loss": 0.9437,
       "step": 2750
     },
     {
       "epoch": 0.5194805194805194,
-      "grad_norm": 3.2873876094818115,
-      "learning_rate": 2.6883116883116883e-05,
-      "loss": 2.0202,
+      "grad_norm": 0.8732848763465881,
+      "learning_rate": 0.000448051948051948,
+      "loss": 0.9885,
       "step": 2760
     },
     {
       "epoch": 0.5213626952757388,
-      "grad_norm": 3.6722116470336914,
-      "learning_rate": 2.6871823828345566e-05,
-      "loss": 2.1103,
+      "grad_norm": 1.563890814781189,
+      "learning_rate": 0.00044786373047242615,
+      "loss": 1.1645,
       "step": 2770
     },
     {
       "epoch": 0.5232448710709581,
-      "grad_norm": 6.789108753204346,
-      "learning_rate": 2.6860530773574252e-05,
-      "loss": 2.0607,
+      "grad_norm": 2.6220126152038574,
+      "learning_rate": 0.0004476755128929042,
+      "loss": 0.9488,
       "step": 2780
     },
     {
       "epoch": 0.5251270468661773,
-      "grad_norm": 9.947601318359375,
-      "learning_rate": 2.6849237718802938e-05,
-      "loss": 1.8943,
+      "grad_norm": 2.038132429122925,
+      "learning_rate": 0.0004474872953133823,
+      "loss": 0.7027,
       "step": 2790
     },
     {
       "epoch": 0.5270092226613966,
-      "grad_norm": 29.515361785888672,
-      "learning_rate": 2.683794466403162e-05,
-      "loss": 1.9724,
+      "grad_norm": 4.242816925048828,
+      "learning_rate": 0.0004472990777338604,
+      "loss": 0.927,
       "step": 2800
     },
     {
       "epoch": 0.5288913984566158,
-      "grad_norm": 14.963418960571289,
-      "learning_rate": 2.6826651609260307e-05,
-      "loss": 2.0618,
+      "grad_norm": 3.240506410598755,
+      "learning_rate": 0.00044711086015433846,
+      "loss": 1.091,
       "step": 2810
     },
     {
       "epoch": 0.5307735742518351,
-      "grad_norm": 16.72760772705078,
-      "learning_rate": 2.681535855448899e-05,
-      "loss": 2.1478,
+      "grad_norm": 5.185576438903809,
+      "learning_rate": 0.00044692264257481653,
+      "loss": 0.9626,
       "step": 2820
     },
     {
       "epoch": 0.5326557500470543,
-      "grad_norm": 11.509770393371582,
-      "learning_rate": 2.6804065499717676e-05,
-      "loss": 2.0953,
+      "grad_norm": 3.4086499214172363,
+      "learning_rate": 0.0004467344249952946,
+      "loss": 1.359,
       "step": 2830
     },
     {
       "epoch": 0.5345379258422737,
-      "grad_norm": 13.363036155700684,
-      "learning_rate": 2.679277244494636e-05,
-      "loss": 2.0589,
+      "grad_norm": 1.9781001806259155,
+      "learning_rate": 0.00044654620741577263,
+      "loss": 0.959,
       "step": 2840
     },
     {
       "epoch": 0.536420101637493,
-      "grad_norm": 4.475217342376709,
-      "learning_rate": 2.678147939017504e-05,
-      "loss": 1.8262,
+      "grad_norm": 1.5130640268325806,
+      "learning_rate": 0.0004463579898362507,
+      "loss": 0.7976,
       "step": 2850
     },
     {
       "epoch": 0.5383022774327122,
-      "grad_norm": 3.634226083755493,
-      "learning_rate": 2.6770186335403727e-05,
-      "loss": 1.7898,
+      "grad_norm": 1.7196760177612305,
+      "learning_rate": 0.0004461697722567288,
+      "loss": 0.7976,
       "step": 2860
     },
     {
       "epoch": 0.5401844532279315,
-      "grad_norm": 22.259286880493164,
-      "learning_rate": 2.675889328063241e-05,
-      "loss": 1.8528,
+      "grad_norm": 4.232097625732422,
+      "learning_rate": 0.00044598155467720686,
+      "loss": 0.87,
       "step": 2870
     },
     {
       "epoch": 0.5420666290231507,
-      "grad_norm": 11.183906555175781,
-      "learning_rate": 2.6747600225861096e-05,
-      "loss": 1.7605,
+      "grad_norm": 3.785689115524292,
+      "learning_rate": 0.00044579333709768494,
+      "loss": 0.8258,
       "step": 2880
     },
     {
       "epoch": 0.54394880481837,
-      "grad_norm": 6.606888294219971,
-      "learning_rate": 2.6736307171089782e-05,
-      "loss": 1.8818,
+      "grad_norm": 1.9887821674346924,
+      "learning_rate": 0.000445605119518163,
+      "loss": 0.7209,
       "step": 2890
     },
     {
       "epoch": 0.5458309806135893,
-      "grad_norm": 8.689189910888672,
-      "learning_rate": 2.6725014116318465e-05,
-      "loss": 1.9096,
+      "grad_norm": 1.6758228540420532,
+      "learning_rate": 0.0004454169019386411,
+      "loss": 0.8262,
       "step": 2900
     },
     {
       "epoch": 0.5477131564088086,
-      "grad_norm": 6.653372287750244,
-      "learning_rate": 2.671372106154715e-05,
-      "loss": 1.7785,
+      "grad_norm": 2.393141031265259,
+      "learning_rate": 0.00044522868435911917,
+      "loss": 0.8338,
       "step": 2910
     },
     {
       "epoch": 0.5495953322040279,
-      "grad_norm": 11.017134666442871,
-      "learning_rate": 2.6702428006775834e-05,
-      "loss": 1.8423,
+      "grad_norm": 6.172318458557129,
+      "learning_rate": 0.0004450404667795972,
+      "loss": 0.8609,
       "step": 2920
     },
     {
       "epoch": 0.5514775079992471,
-      "grad_norm": 6.855719089508057,
-      "learning_rate": 2.6691134952004516e-05,
-      "loss": 1.8001,
+      "grad_norm": 0.41666731238365173,
+      "learning_rate": 0.00044485224920007526,
+      "loss": 0.86,
       "step": 2930
     },
     {
       "epoch": 0.5533596837944664,
-      "grad_norm": 5.900759220123291,
-      "learning_rate": 2.6679841897233202e-05,
-      "loss": 1.987,
+      "grad_norm": 3.7566018104553223,
+      "learning_rate": 0.00044466403162055334,
+      "loss": 0.9788,
       "step": 2940
     },
     {
       "epoch": 0.5552418595896856,
-      "grad_norm": 5.474452495574951,
-      "learning_rate": 2.6668548842461885e-05,
-      "loss": 1.9144,
+      "grad_norm": 0.9079445004463196,
+      "learning_rate": 0.0004444758140410314,
+      "loss": 0.8222,
       "step": 2950
     },
     {
       "epoch": 0.557124035384905,
-      "grad_norm": 13.115139961242676,
-      "learning_rate": 2.665725578769057e-05,
-      "loss": 2.0743,
+      "grad_norm": 2.0461573600769043,
+      "learning_rate": 0.0004442875964615095,
+      "loss": 1.1583,
       "step": 2960
     },
     {
       "epoch": 0.5590062111801242,
-      "grad_norm": 8.339238166809082,
-      "learning_rate": 2.6645962732919257e-05,
-      "loss": 1.9056,
+      "grad_norm": 3.3327903747558594,
+      "learning_rate": 0.00044409937888198757,
+      "loss": 0.9933,
       "step": 2970
     },
     {
       "epoch": 0.5608883869753435,
-      "grad_norm": 12.62999153137207,
-      "learning_rate": 2.663466967814794e-05,
-      "loss": 1.8391,
+      "grad_norm": 2.7679216861724854,
+      "learning_rate": 0.00044391116130246565,
+      "loss": 0.8464,
       "step": 2980
     },
     {
       "epoch": 0.5627705627705628,
-      "grad_norm": 5.597290515899658,
-      "learning_rate": 2.6623376623376626e-05,
-      "loss": 1.9349,
+      "grad_norm": 1.8985081911087036,
+      "learning_rate": 0.0004437229437229438,
+      "loss": 0.9887,
       "step": 2990
     },
     {
       "epoch": 0.564652738565782,
-      "grad_norm": 2.5785648822784424,
-      "learning_rate": 2.661208356860531e-05,
-      "loss": 1.9463,
+      "grad_norm": 1.1193677186965942,
+      "learning_rate": 0.00044353472614342185,
+      "loss": 0.9821,
       "step": 3000
     },
     {
       "epoch": 0.5665349143610013,
-      "grad_norm": 13.423247337341309,
-      "learning_rate": 2.660079051383399e-05,
-      "loss": 1.7291,
+      "grad_norm": 2.58882999420166,
+      "learning_rate": 0.0004433465085638999,
+      "loss": 0.7463,
       "step": 3010
     },
     {
       "epoch": 0.5684170901562206,
-      "grad_norm": 7.054490089416504,
-      "learning_rate": 2.6589497459062677e-05,
-      "loss": 1.8435,
+      "grad_norm": 4.277840614318848,
+      "learning_rate": 0.00044315829098437795,
+      "loss": 0.7602,
       "step": 3020
     },
     {
       "epoch": 0.5702992659514399,
-      "grad_norm": 7.4367523193359375,
-      "learning_rate": 2.657820440429136e-05,
-      "loss": 1.7105,
+      "grad_norm": 0.9204025268554688,
+      "learning_rate": 0.00044297007340485603,
+      "loss": 0.6596,
       "step": 3030
     },
     {
       "epoch": 0.5721814417466592,
-      "grad_norm": 9.216171264648438,
-      "learning_rate": 2.6566911349520046e-05,
-      "loss": 1.9434,
+      "grad_norm": 3.1999411582946777,
+      "learning_rate": 0.0004427818558253341,
+      "loss": 0.8291,
       "step": 3040
     },
     {
       "epoch": 0.5740636175418784,
-      "grad_norm": 7.6077165603637695,
-      "learning_rate": 2.655561829474873e-05,
-      "loss": 1.8798,
+      "grad_norm": 1.4632461071014404,
+      "learning_rate": 0.0004425936382458122,
+      "loss": 1.1211,
       "step": 3050
     },
     {
       "epoch": 0.5759457933370977,
-      "grad_norm": 11.647812843322754,
-      "learning_rate": 2.6544325239977415e-05,
-      "loss": 1.7522,
+      "grad_norm": 2.777308225631714,
+      "learning_rate": 0.00044240542066629026,
+      "loss": 0.7434,
       "step": 3060
     },
     {
       "epoch": 0.5778279691323169,
-      "grad_norm": 5.945573806762695,
-      "learning_rate": 2.65330321852061e-05,
-      "loss": 2.0416,
+      "grad_norm": 2.7749745845794678,
+      "learning_rate": 0.00044221720308676833,
+      "loss": 1.0088,
       "step": 3070
     },
     {
       "epoch": 0.5797101449275363,
-      "grad_norm": 24.724990844726562,
-      "learning_rate": 2.6521739130434784e-05,
-      "loss": 1.8017,
+      "grad_norm": 1.9405794143676758,
+      "learning_rate": 0.0004420289855072464,
+      "loss": 0.9694,
       "step": 3080
     },
     {
       "epoch": 0.5815923207227555,
-      "grad_norm": 4.811154842376709,
-      "learning_rate": 2.651044607566347e-05,
-      "loss": 1.6813,
+      "grad_norm": 1.6995877027511597,
+      "learning_rate": 0.0004418407679277245,
+      "loss": 0.5558,
       "step": 3090
     },
     {
       "epoch": 0.5834744965179748,
-      "grad_norm": 11.191183090209961,
-      "learning_rate": 2.649915302089215e-05,
-      "loss": 1.7622,
+      "grad_norm": 2.6965272426605225,
+      "learning_rate": 0.0004416525503482025,
+      "loss": 0.8034,
       "step": 3100
     },
     {
       "epoch": 0.5853566723131941,
-      "grad_norm": 12.16427230834961,
-      "learning_rate": 2.6487859966120835e-05,
-      "loss": 1.9663,
+      "grad_norm": 1.4357980489730835,
+      "learning_rate": 0.0004414643327686806,
+      "loss": 1.1173,
       "step": 3110
     },
     {
       "epoch": 0.5872388481084133,
-      "grad_norm": 8.902604103088379,
-      "learning_rate": 2.647656691134952e-05,
-      "loss": 1.7131,
+      "grad_norm": 4.319276332855225,
+      "learning_rate": 0.00044127611518915866,
+      "loss": 0.9903,
       "step": 3120
     },
     {
       "epoch": 0.5891210239036326,
-      "grad_norm": 7.6382904052734375,
-      "learning_rate": 2.6465273856578204e-05,
-      "loss": 1.7472,
+      "grad_norm": 2.23298978805542,
+      "learning_rate": 0.00044108789760963674,
+      "loss": 0.7759,
       "step": 3130
     },
     {
       "epoch": 0.5910031996988518,
-      "grad_norm": 23.864648818969727,
-      "learning_rate": 2.645398080180689e-05,
-      "loss": 1.679,
+      "grad_norm": 3.0950074195861816,
+      "learning_rate": 0.0004408996800301148,
+      "loss": 0.7602,
       "step": 3140
     },
     {
       "epoch": 0.5928853754940712,
-      "grad_norm": 18.402881622314453,
-      "learning_rate": 2.6442687747035573e-05,
-      "loss": 2.1442,
+      "grad_norm": 5.252658843994141,
+      "learning_rate": 0.0004407114624505929,
+      "loss": 1.4432,
       "step": 3150
     },
     {
       "epoch": 0.5947675512892904,
-      "grad_norm": 6.911153793334961,
-      "learning_rate": 2.643139469226426e-05,
-      "loss": 1.6756,
+      "grad_norm": 3.7704455852508545,
+      "learning_rate": 0.00044052324487107097,
+      "loss": 0.7093,
       "step": 3160
     },
     {
       "epoch": 0.5966497270845097,
-      "grad_norm": 12.472163200378418,
-      "learning_rate": 2.6420101637492945e-05,
-      "loss": 1.9381,
+      "grad_norm": 3.0606801509857178,
+      "learning_rate": 0.00044033502729154904,
+      "loss": 1.3438,
       "step": 3170
     },
     {
       "epoch": 0.598531902879729,
-      "grad_norm": 4.999609470367432,
-      "learning_rate": 2.6408808582721624e-05,
-      "loss": 1.5681,
+      "grad_norm": 1.9222668409347534,
+      "learning_rate": 0.00044014680971202707,
+      "loss": 0.8256,
       "step": 3180
     },
     {
       "epoch": 0.6004140786749482,
-      "grad_norm": 5.08629846572876,
-      "learning_rate": 2.639751552795031e-05,
-      "loss": 1.8375,
+      "grad_norm": 1.1805700063705444,
+      "learning_rate": 0.00043995859213250514,
+      "loss": 1.0023,
       "step": 3190
     },
     {
       "epoch": 0.6022962544701675,
-      "grad_norm": 5.805210590362549,
-      "learning_rate": 2.6386222473178996e-05,
-      "loss": 2.0463,
+      "grad_norm": 4.068279266357422,
+      "learning_rate": 0.0004397703745529832,
+      "loss": 1.2515,
       "step": 3200
     },
     {
       "epoch": 0.6041784302653868,
-      "grad_norm": 7.146366119384766,
-      "learning_rate": 2.637492941840768e-05,
-      "loss": 1.5877,
+      "grad_norm": 1.760879397392273,
+      "learning_rate": 0.0004395821569734613,
+      "loss": 0.6779,
       "step": 3210
     },
     {
       "epoch": 0.6060606060606061,
-      "grad_norm": 4.034036159515381,
-      "learning_rate": 2.6363636363636365e-05,
-      "loss": 1.835,
+      "grad_norm": 0.343962162733078,
+      "learning_rate": 0.0004393939393939394,
+      "loss": 0.8352,
       "step": 3220
     },
     {
       "epoch": 0.6079427818558253,
-      "grad_norm": 22.046175003051758,
-      "learning_rate": 2.6352343308865048e-05,
-      "loss": 1.702,
+      "grad_norm": 4.934606075286865,
+      "learning_rate": 0.0004392057218144175,
+      "loss": 0.7756,
       "step": 3230
     },
     {
       "epoch": 0.6098249576510446,
-      "grad_norm": 20.46267318725586,
-      "learning_rate": 2.6341050254093734e-05,
-      "loss": 1.8594,
+      "grad_norm": 5.34637975692749,
+      "learning_rate": 0.0004390175042348956,
+      "loss": 0.9066,
       "step": 3240
     },
     {
       "epoch": 0.6117071334462639,
-      "grad_norm": 6.352889537811279,
-      "learning_rate": 2.632975719932242e-05,
-      "loss": 1.78,
+      "grad_norm": 1.3073912858963013,
+      "learning_rate": 0.00043882928665537366,
+      "loss": 0.8333,
       "step": 3250
     },
     {
       "epoch": 0.6135893092414831,
-      "grad_norm": 5.597209453582764,
-      "learning_rate": 2.6318464144551103e-05,
-      "loss": 1.9062,
+      "grad_norm": 1.1687594652175903,
+      "learning_rate": 0.00043864106907585173,
+      "loss": 0.9611,
       "step": 3260
     },
     {
       "epoch": 0.6154714850367025,
-      "grad_norm": 8.404434204101562,
-      "learning_rate": 2.6307171089779785e-05,
-      "loss": 1.7729,
+      "grad_norm": 1.1420823335647583,
+      "learning_rate": 0.00043845285149632975,
+      "loss": 0.9158,
       "step": 3270
     },
     {
       "epoch": 0.6173536608319217,
-      "grad_norm": 9.512775421142578,
-      "learning_rate": 2.6295878035008468e-05,
-      "loss": 1.6567,
+      "grad_norm": 2.647937536239624,
+      "learning_rate": 0.00043826463391680783,
+      "loss": 0.7782,
       "step": 3280
     },
     {
       "epoch": 0.619235836627141,
-      "grad_norm": 11.096758842468262,
-      "learning_rate": 2.6284584980237154e-05,
-      "loss": 1.5892,
+      "grad_norm": 4.739561557769775,
+      "learning_rate": 0.0004380764163372859,
+      "loss": 0.709,
       "step": 3290
     },
     {
       "epoch": 0.6211180124223602,
-      "grad_norm": 4.303831100463867,
-      "learning_rate": 2.627329192546584e-05,
-      "loss": 1.7157,
+      "grad_norm": 1.5627340078353882,
+      "learning_rate": 0.000437888198757764,
+      "loss": 0.8886,
       "step": 3300
     },
     {
       "epoch": 0.6230001882175795,
-      "grad_norm": 3.8192107677459717,
-      "learning_rate": 2.6261998870694523e-05,
-      "loss": 1.7188,
+      "grad_norm": 3.9366836547851562,
+      "learning_rate": 0.00043769998117824206,
+      "loss": 0.9539,
       "step": 3310
     },
     {
       "epoch": 0.6248823640127988,
-      "grad_norm": 10.297846794128418,
-      "learning_rate": 2.625070581592321e-05,
-      "loss": 1.7473,
+      "grad_norm": 3.5125503540039062,
+      "learning_rate": 0.00043751176359872014,
+      "loss": 0.9994,
       "step": 3320
     },
     {
       "epoch": 0.626764539808018,
-      "grad_norm": 8.126511573791504,
-      "learning_rate": 2.623941276115189e-05,
-      "loss": 1.5694,
+      "grad_norm": 3.3123700618743896,
+      "learning_rate": 0.0004373235460191982,
+      "loss": 0.7918,
       "step": 3330
     },
     {
       "epoch": 0.6286467156032374,
-      "grad_norm": 4.767712116241455,
-      "learning_rate": 2.6228119706380578e-05,
-      "loss": 1.7124,
+      "grad_norm": 2.980029821395874,
+      "learning_rate": 0.0004371353284396763,
+      "loss": 0.9837,
       "step": 3340
     },
     {
       "epoch": 0.6305288913984566,
-      "grad_norm": 4.750823020935059,
-      "learning_rate": 2.6216826651609264e-05,
-      "loss": 1.8946,
+      "grad_norm": 1.5912530422210693,
+      "learning_rate": 0.00043694711086015437,
+      "loss": 1.2457,
       "step": 3350
     },
     {
       "epoch": 0.6324110671936759,
-      "grad_norm": 9.973583221435547,
-      "learning_rate": 2.6205533596837943e-05,
-      "loss": 1.6632,
+      "grad_norm": 1.0829591751098633,
+      "learning_rate": 0.0004367588932806324,
+      "loss": 0.8409,
       "step": 3360
     },
     {
       "epoch": 0.6342932429888951,
-      "grad_norm": 10.300981521606445,
-      "learning_rate": 2.619424054206663e-05,
-      "loss": 1.8623,
+      "grad_norm": 1.4647136926651,
+      "learning_rate": 0.00043657067570111046,
+      "loss": 1.1027,
       "step": 3370
     },
     {
       "epoch": 0.6361754187841144,
-      "grad_norm": 17.23389434814453,
-      "learning_rate": 2.6182947487295312e-05,
-      "loss": 1.8112,
+      "grad_norm": 4.678377151489258,
+      "learning_rate": 0.00043638245812158854,
+      "loss": 0.9029,
       "step": 3380
     },
     {
       "epoch": 0.6380575945793338,
-      "grad_norm": 8.108388900756836,
-      "learning_rate": 2.6171654432523998e-05,
-      "loss": 1.6344,
+      "grad_norm": 2.8882977962493896,
+      "learning_rate": 0.0004361942405420666,
+      "loss": 1.0712,
       "step": 3390
     },
     {
       "epoch": 0.639939770374553,
-      "grad_norm": 3.373798131942749,
-      "learning_rate": 2.6160361377752684e-05,
-      "loss": 1.5514,
+      "grad_norm": 3.299114227294922,
+      "learning_rate": 0.0004360060229625447,
+      "loss": 0.846,
       "step": 3400
     },
     {
       "epoch": 0.6418219461697723,
-      "grad_norm": 8.216525077819824,
-      "learning_rate": 2.6149068322981367e-05,
-      "loss": 1.727,
+      "grad_norm": 2.3301033973693848,
+      "learning_rate": 0.00043581780538302277,
+      "loss": 0.8764,
       "step": 3410
     },
     {
       "epoch": 0.6437041219649915,
-      "grad_norm": 9.52584457397461,
-      "learning_rate": 2.6137775268210053e-05,
-      "loss": 1.8575,
+      "grad_norm": 3.6711699962615967,
+      "learning_rate": 0.00043562958780350085,
+      "loss": 1.0153,
       "step": 3420
     },
     {
       "epoch": 0.6455862977602108,
-      "grad_norm": 7.7391581535339355,
-      "learning_rate": 2.612648221343874e-05,
-      "loss": 1.7685,
+      "grad_norm": 3.8490591049194336,
+      "learning_rate": 0.000435441370223979,
+      "loss": 0.9266,
       "step": 3430
     },
     {
       "epoch": 0.64746847355543,
-      "grad_norm": 8.019952774047852,
-      "learning_rate": 2.611518915866742e-05,
-      "loss": 1.6028,
+      "grad_norm": 2.865903377532959,
+      "learning_rate": 0.00043525315264445705,
+      "loss": 0.8813,
       "step": 3440
     },
     {
       "epoch": 0.6493506493506493,
-      "grad_norm": 11.445884704589844,
-      "learning_rate": 2.6103896103896104e-05,
-      "loss": 1.9301,
+      "grad_norm": 3.837740659713745,
+      "learning_rate": 0.0004350649350649351,
+      "loss": 1.2178,
       "step": 3450
     },
     {
       "epoch": 0.6512328251458687,
-      "grad_norm": 8.298888206481934,
-      "learning_rate": 2.6092603049124787e-05,
-      "loss": 1.5639,
+      "grad_norm": 1.263821005821228,
+      "learning_rate": 0.00043487671748541315,
+      "loss": 0.7914,
       "step": 3460
     },
     {
       "epoch": 0.6531150009410879,
-      "grad_norm": 6.707033157348633,
-      "learning_rate": 2.6081309994353473e-05,
-      "loss": 1.9865,
+      "grad_norm": 2.0916056632995605,
+      "learning_rate": 0.00043468849990589123,
+      "loss": 1.3344,
       "step": 3470
     },
     {
       "epoch": 0.6549971767363072,
-      "grad_norm": 13.604408264160156,
-      "learning_rate": 2.607001693958216e-05,
-      "loss": 1.8029,
+      "grad_norm": 2.437861919403076,
+      "learning_rate": 0.0004345002823263693,
+      "loss": 1.138,
       "step": 3480
     },
     {
       "epoch": 0.6568793525315264,
-      "grad_norm": 13.318224906921387,
-      "learning_rate": 2.605872388481084e-05,
-      "loss": 1.6995,
+      "grad_norm": 0.3909652829170227,
+      "learning_rate": 0.0004343120647468474,
+      "loss": 0.8935,
       "step": 3490
     },
     {
       "epoch": 0.6587615283267457,
-      "grad_norm": 10.210710525512695,
-      "learning_rate": 2.6047430830039528e-05,
-      "loss": 1.4445,
+      "grad_norm": 3.115875244140625,
+      "learning_rate": 0.00043412384716732546,
+      "loss": 0.684,
       "step": 3500
     },
     {
       "epoch": 0.6606437041219649,
-      "grad_norm": 23.614696502685547,
-      "learning_rate": 2.603613777526821e-05,
-      "loss": 1.7893,
+      "grad_norm": 6.081343173980713,
+      "learning_rate": 0.00043393562958780353,
+      "loss": 0.9985,
       "step": 3510
     },
     {
       "epoch": 0.6625258799171843,
-      "grad_norm": 6.86085844039917,
-      "learning_rate": 2.6024844720496896e-05,
-      "loss": 1.5184,
+      "grad_norm": 1.3088817596435547,
+      "learning_rate": 0.0004337474120082816,
+      "loss": 0.8571,
       "step": 3520
     },
     {
       "epoch": 0.6644080557124036,
-      "grad_norm": 7.119161128997803,
-      "learning_rate": 2.601355166572558e-05,
-      "loss": 1.6189,
+      "grad_norm": 1.4548195600509644,
+      "learning_rate": 0.00043355919442875963,
+      "loss": 0.9322,
       "step": 3530
     },
     {
       "epoch": 0.6662902315076228,
-      "grad_norm": 7.726982593536377,
-      "learning_rate": 2.6002258610954262e-05,
-      "loss": 1.7796,
+      "grad_norm": 2.6708908081054688,
+      "learning_rate": 0.0004333709768492377,
+      "loss": 1.035,
       "step": 3540
     },
     {
       "epoch": 0.6681724073028421,
-      "grad_norm": 7.957070350646973,
-      "learning_rate": 2.5990965556182948e-05,
-      "loss": 1.5165,
+      "grad_norm": 2.1637444496154785,
+      "learning_rate": 0.0004331827592697158,
+      "loss": 0.5862,
       "step": 3550
     },
     {
       "epoch": 0.6700545830980613,
-      "grad_norm": 11.317330360412598,
-      "learning_rate": 2.597967250141163e-05,
-      "loss": 1.835,
+      "grad_norm": 4.600494861602783,
+      "learning_rate": 0.00043299454169019386,
+      "loss": 0.9426,
       "step": 3560
     },
     {
       "epoch": 0.6719367588932806,
-      "grad_norm": 7.407431602478027,
-      "learning_rate": 2.5968379446640317e-05,
-      "loss": 1.3834,
+      "grad_norm": 2.8519279956817627,
+      "learning_rate": 0.00043280632411067194,
+      "loss": 0.5583,
       "step": 3570
     },
     {
       "epoch": 0.6738189346885,
-      "grad_norm": 9.898704528808594,
-      "learning_rate": 2.5957086391869003e-05,
-      "loss": 1.432,
+      "grad_norm": 1.3281288146972656,
+      "learning_rate": 0.00043261810653115,
+      "loss": 0.7769,
       "step": 3580
     },
     {
       "epoch": 0.6757011104837192,
-      "grad_norm": 5.612309455871582,
-      "learning_rate": 2.5945793337097685e-05,
-      "loss": 1.7256,
+      "grad_norm": 2.5639514923095703,
+      "learning_rate": 0.0004324298889516281,
+      "loss": 1.0062,
       "step": 3590
     },
     {
       "epoch": 0.6775832862789385,
-      "grad_norm": 5.5622663497924805,
-      "learning_rate": 2.593450028232637e-05,
-      "loss": 1.5599,
+      "grad_norm": 2.5893678665161133,
+      "learning_rate": 0.00043224167137210617,
+      "loss": 0.9067,
       "step": 3600
     },
     {
       "epoch": 0.6794654620741577,
-      "grad_norm": 5.885371685028076,
-      "learning_rate": 2.5923207227555054e-05,
-      "loss": 1.7386,
+      "grad_norm": 2.5090417861938477,
+      "learning_rate": 0.00043205345379258424,
+      "loss": 0.905,
       "step": 3610
     },
     {
       "epoch": 0.681347637869377,
-      "grad_norm": 3.2068424224853516,
-      "learning_rate": 2.5911914172783737e-05,
-      "loss": 1.5411,
+      "grad_norm": 3.526054859161377,
+      "learning_rate": 0.00043186523621306227,
+      "loss": 0.8226,
       "step": 3620
     },
     {
       "epoch": 0.6832298136645962,
-      "grad_norm": 3.0026233196258545,
-      "learning_rate": 2.5900621118012423e-05,
-      "loss": 1.7239,
+      "grad_norm": 13.659317016601562,
+      "learning_rate": 0.00043167701863354034,
+      "loss": 0.9691,
       "step": 3630
     },
     {
       "epoch": 0.6851119894598156,
-      "grad_norm": 23.438610076904297,
-      "learning_rate": 2.5889328063241106e-05,
-      "loss": 1.7582,
+      "grad_norm": 3.2574679851531982,
+      "learning_rate": 0.0004314888010540184,
+      "loss": 1.0528,
       "step": 3640
     },
     {
       "epoch": 0.6869941652550349,
-      "grad_norm": 3.962045907974243,
-      "learning_rate": 2.5878035008469792e-05,
-      "loss": 1.4579,
+      "grad_norm": 1.4542733430862427,
+      "learning_rate": 0.0004313005834744965,
+      "loss": 0.8987,
       "step": 3650
     },
     {
       "epoch": 0.6888763410502541,
-      "grad_norm": 3.3566129207611084,
-      "learning_rate": 2.5866741953698478e-05,
-      "loss": 1.4641,
+      "grad_norm": 1.2248775959014893,
+      "learning_rate": 0.0004311123658949746,
+      "loss": 0.6569,
       "step": 3660
     },
     {
       "epoch": 0.6907585168454734,
-      "grad_norm": 8.942924499511719,
-      "learning_rate": 2.585544889892716e-05,
-      "loss": 1.5699,
+      "grad_norm": 6.2517008781433105,
+      "learning_rate": 0.0004309241483154527,
+      "loss": 0.8825,
       "step": 3670
     },
     {
       "epoch": 0.6926406926406926,
-      "grad_norm": 12.78019905090332,
-      "learning_rate": 2.5844155844155847e-05,
-      "loss": 1.6055,
+      "grad_norm": 2.5166094303131104,
+      "learning_rate": 0.0004307359307359308,
+      "loss": 0.7376,
       "step": 3680
     },
     {
       "epoch": 0.6945228684359119,
-      "grad_norm": 14.28182315826416,
-      "learning_rate": 2.583286278938453e-05,
-      "loss": 1.6354,
+      "grad_norm": 5.175745964050293,
+      "learning_rate": 0.00043054771315640886,
+      "loss": 0.8224,
       "step": 3690
     },
     {
       "epoch": 0.6964050442311311,
-      "grad_norm": 10.26236629486084,
-      "learning_rate": 2.5821569734613215e-05,
-      "loss": 1.5518,
+      "grad_norm": 0.9270199537277222,
+      "learning_rate": 0.00043035949557688693,
+      "loss": 0.8223,
       "step": 3700
     },
     {
       "epoch": 0.6982872200263505,
-      "grad_norm": 6.233335494995117,
-      "learning_rate": 2.5810276679841898e-05,
-      "loss": 1.4589,
+      "grad_norm": 1.036481261253357,
+      "learning_rate": 0.00043017127799736495,
+      "loss": 0.6669,
       "step": 3710
     },
     {
       "epoch": 0.7001693958215698,
-      "grad_norm": 17.028034210205078,
-      "learning_rate": 2.579898362507058e-05,
-      "loss": 1.6329,
+      "grad_norm": 4.925989627838135,
+      "learning_rate": 0.00042998306041784303,
+      "loss": 0.6371,
       "step": 3720
     },
     {
       "epoch": 0.702051571616789,
-      "grad_norm": 9.508376121520996,
-      "learning_rate": 2.5787690570299267e-05,
-      "loss": 1.4574,
+      "grad_norm": 4.520897388458252,
+      "learning_rate": 0.0004297948428383211,
+      "loss": 0.7395,
       "step": 3730
     },
     {
       "epoch": 0.7039337474120083,
-      "grad_norm": 10.823552131652832,
-      "learning_rate": 2.577639751552795e-05,
-      "loss": 1.7267,
+      "grad_norm": 3.2131059169769287,
+      "learning_rate": 0.0004296066252587992,
+      "loss": 1.3384,
       "step": 3740
     },
     {
       "epoch": 0.7058159232072275,
-      "grad_norm": 19.552865982055664,
-      "learning_rate": 2.5765104460756636e-05,
-      "loss": 1.7066,
+      "grad_norm": 2.445014476776123,
+      "learning_rate": 0.00042941840767927726,
+      "loss": 0.7885,
       "step": 3750
     },
     {
       "epoch": 0.7076980990024468,
-      "grad_norm": 13.044384002685547,
-      "learning_rate": 2.575381140598532e-05,
-      "loss": 1.5771,
+      "grad_norm": 3.6331963539123535,
+      "learning_rate": 0.00042923019009975534,
+      "loss": 0.9128,
       "step": 3760
     },
     {
       "epoch": 0.709580274797666,
-      "grad_norm": 7.333944797515869,
-      "learning_rate": 2.5742518351214004e-05,
-      "loss": 1.4535,
+      "grad_norm": 3.285827398300171,
+      "learning_rate": 0.0004290419725202334,
+      "loss": 0.8912,
       "step": 3770
     },
     {
       "epoch": 0.7114624505928854,
-      "grad_norm": 12.022722244262695,
-      "learning_rate": 2.573122529644269e-05,
-      "loss": 1.6377,
+      "grad_norm": 9.035295486450195,
+      "learning_rate": 0.0004288537549407115,
+      "loss": 1.109,
       "step": 3780
     },
     {
       "epoch": 0.7133446263881047,
-      "grad_norm": 6.42018461227417,
-      "learning_rate": 2.571993224167137e-05,
-      "loss": 1.4887,
+      "grad_norm": 3.6709072589874268,
+      "learning_rate": 0.0004286655373611895,
+      "loss": 0.6925,
       "step": 3790
     },
     {
       "epoch": 0.7152268021833239,
-      "grad_norm": 2.690678596496582,
-      "learning_rate": 2.5708639186900056e-05,
-      "loss": 1.7047,
+      "grad_norm": 0.12020440399646759,
+      "learning_rate": 0.0004284773197816676,
+      "loss": 1.1243,
       "step": 3800
     },
     {
       "epoch": 0.7171089779785432,
-      "grad_norm": 25.072540283203125,
-      "learning_rate": 2.5697346132128742e-05,
-      "loss": 1.6334,
+      "grad_norm": 4.302594184875488,
+      "learning_rate": 0.00042828910220214566,
+      "loss": 0.9626,
       "step": 3810
     },
     {
       "epoch": 0.7189911537737624,
-      "grad_norm": 19.594491958618164,
-      "learning_rate": 2.5686053077357425e-05,
-      "loss": 1.3771,
+      "grad_norm": 4.240476608276367,
+      "learning_rate": 0.00042810088462262374,
+      "loss": 0.8101,
       "step": 3820
     },
     {
       "epoch": 0.7208733295689818,
-      "grad_norm": 32.11001205444336,
-      "learning_rate": 2.567476002258611e-05,
-      "loss": 1.5945,
+      "grad_norm": 2.9120934009552,
+      "learning_rate": 0.0004279126670431018,
+      "loss": 0.9986,
       "step": 3830
     },
     {
       "epoch": 0.722755505364201,
-      "grad_norm": 5.285523891448975,
-      "learning_rate": 2.5663466967814793e-05,
-      "loss": 1.4196,
+      "grad_norm": 3.199678659439087,
+      "learning_rate": 0.0004277244494635799,
+      "loss": 1.0096,
       "step": 3840
     },
     {
       "epoch": 0.7246376811594203,
-      "grad_norm": 9.592608451843262,
-      "learning_rate": 2.565217391304348e-05,
-      "loss": 1.6145,
+      "grad_norm": 2.3253014087677,
+      "learning_rate": 0.00042753623188405797,
+      "loss": 0.8287,
       "step": 3850
     },
     {
       "epoch": 0.7265198569546396,
-      "grad_norm": 11.879024505615234,
-      "learning_rate": 2.5640880858272165e-05,
-      "loss": 1.6065,
+      "grad_norm": 2.6479291915893555,
+      "learning_rate": 0.00042734801430453605,
+      "loss": 0.9506,
       "step": 3860
     },
     {
       "epoch": 0.7284020327498588,
-      "grad_norm": 15.967653274536133,
-      "learning_rate": 2.5629587803500848e-05,
-      "loss": 1.716,
+      "grad_norm": 4.7788214683532715,
+      "learning_rate": 0.0004271597967250141,
+      "loss": 0.8718,
       "step": 3870
     },
     {
       "epoch": 0.7302842085450781,
-      "grad_norm": 20.394914627075195,
-      "learning_rate": 2.561829474872953e-05,
-      "loss": 1.4524,
+      "grad_norm": 1.1644917726516724,
+      "learning_rate": 0.0004269715791454922,
+      "loss": 0.774,
       "step": 3880
     },
     {
       "epoch": 0.7321663843402973,
-      "grad_norm": 8.076727867126465,
-      "learning_rate": 2.5607001693958214e-05,
-      "loss": 1.6307,
+      "grad_norm": 5.440783977508545,
+      "learning_rate": 0.0004267833615659703,
+      "loss": 1.0616,
       "step": 3890
     },
     {
       "epoch": 0.7340485601355167,
-      "grad_norm": 15.703822135925293,
-      "learning_rate": 2.55957086391869e-05,
-      "loss": 1.2349,
+      "grad_norm": 2.430136203765869,
+      "learning_rate": 0.00042659514398644835,
+      "loss": 0.543,
       "step": 3900
     },
     {
       "epoch": 0.7359307359307359,
-      "grad_norm": 13.136933326721191,
-      "learning_rate": 2.5584415584415586e-05,
-      "loss": 1.6267,
+      "grad_norm": 3.5641326904296875,
+      "learning_rate": 0.00042640692640692643,
+      "loss": 0.9902,
       "step": 3910
     },
     {
       "epoch": 0.7378129117259552,
-      "grad_norm": 10.711684226989746,
-      "learning_rate": 2.557312252964427e-05,
-      "loss": 1.5033,
+      "grad_norm": 2.292509078979492,
+      "learning_rate": 0.0004262187088274045,
+      "loss": 1.04,
       "step": 3920
     },
     {
       "epoch": 0.7396950875211745,
-      "grad_norm": 12.624408721923828,
-      "learning_rate": 2.5561829474872954e-05,
-      "loss": 1.516,
+      "grad_norm": 2.2531626224517822,
+      "learning_rate": 0.0004260304912478826,
+      "loss": 0.7291,
       "step": 3930
     },
     {
       "epoch": 0.7415772633163937,
-      "grad_norm": 5.2435479164123535,
-      "learning_rate": 2.555053642010164e-05,
-      "loss": 1.3178,
+      "grad_norm": 2.3294925689697266,
+      "learning_rate": 0.00042584227366836066,
+      "loss": 0.8943,
       "step": 3940
     },
     {
       "epoch": 0.743459439111613,
-      "grad_norm": 12.167356491088867,
-      "learning_rate": 2.5539243365330323e-05,
-      "loss": 1.7724,
+      "grad_norm": 2.969782590866089,
+      "learning_rate": 0.00042565405608883873,
+      "loss": 0.8468,
       "step": 3950
     },
     {
       "epoch": 0.7453416149068323,
-      "grad_norm": 11.038766860961914,
-      "learning_rate": 2.552795031055901e-05,
-      "loss": 1.397,
+      "grad_norm": 2.1566853523254395,
+      "learning_rate": 0.0004254658385093168,
+      "loss": 0.6394,
       "step": 3960
     },
     {
       "epoch": 0.7472237907020516,
-      "grad_norm": 11.564054489135742,
-      "learning_rate": 2.551665725578769e-05,
-      "loss": 1.5357,
+      "grad_norm": 5.316396713256836,
+      "learning_rate": 0.00042527762092979483,
+      "loss": 1.0201,
       "step": 3970
     },
     {
       "epoch": 0.7491059664972708,
-      "grad_norm": 19.608352661132812,
-      "learning_rate": 2.5505364201016375e-05,
-      "loss": 1.5665,
+      "grad_norm": 4.417022705078125,
+      "learning_rate": 0.0004250894033502729,
+      "loss": 0.8763,
       "step": 3980
     },
     {
       "epoch": 0.7509881422924901,
-      "grad_norm": 13.263577461242676,
-      "learning_rate": 2.549407114624506e-05,
-      "loss": 1.4692,
+      "grad_norm": 3.9584805965423584,
+      "learning_rate": 0.000424901185770751,
+      "loss": 0.8368,
       "step": 3990
     },
     {
       "epoch": 0.7528703180877094,
-      "grad_norm": 21.618656158447266,
-      "learning_rate": 2.5482778091473743e-05,
-      "loss": 1.5511,
+      "grad_norm": 4.402260780334473,
+      "learning_rate": 0.00042471296819122906,
+      "loss": 0.9425,
       "step": 4000
     },
     {
       "epoch": 0.7547524938829286,
-      "grad_norm": 4.379132270812988,
-      "learning_rate": 2.547148503670243e-05,
-      "loss": 1.6006,
+      "grad_norm": 1.8441309928894043,
+      "learning_rate": 0.00042452475061170714,
+      "loss": 0.9747,
       "step": 4010
     },
     {
       "epoch": 0.756634669678148,
-      "grad_norm": 5.841297626495361,
-      "learning_rate": 2.5460191981931112e-05,
-      "loss": 1.3704,
+      "grad_norm": 2.9513907432556152,
+      "learning_rate": 0.0004243365330321852,
+      "loss": 0.6434,
       "step": 4020
     },
     {
       "epoch": 0.7585168454733672,
-      "grad_norm": 9.923529624938965,
-      "learning_rate": 2.5448898927159798e-05,
-      "loss": 1.6851,
+      "grad_norm": 3.533721923828125,
+      "learning_rate": 0.0004241483154526633,
+      "loss": 1.06,
       "step": 4030
     },
     {
       "epoch": 0.7603990212685865,
-      "grad_norm": 6.657973766326904,
-      "learning_rate": 2.5437605872388484e-05,
-      "loss": 1.3996,
+      "grad_norm": 3.201486110687256,
+      "learning_rate": 0.00042396009787314137,
+      "loss": 0.7152,
       "step": 4040
     },
     {
       "epoch": 0.7622811970638057,
-      "grad_norm": 7.652090072631836,
-      "learning_rate": 2.5426312817617167e-05,
-      "loss": 1.3747,
+      "grad_norm": 3.1424193382263184,
+      "learning_rate": 0.00042377188029361944,
+      "loss": 0.8416,
       "step": 4050
     },
     {
       "epoch": 0.764163372859025,
-      "grad_norm": 6.833724498748779,
-      "learning_rate": 2.541501976284585e-05,
-      "loss": 1.3425,
+      "grad_norm": 3.6960856914520264,
+      "learning_rate": 0.00042358366271409747,
+      "loss": 0.7052,
       "step": 4060
     },
     {
       "epoch": 0.7660455486542443,
-      "grad_norm": 3.7029807567596436,
-      "learning_rate": 2.5403726708074532e-05,
-      "loss": 1.5997,
+      "grad_norm": 1.366542935371399,
+      "learning_rate": 0.00042339544513457554,
+      "loss": 1.1092,
       "step": 4070
     },
     {
       "epoch": 0.7679277244494636,
-      "grad_norm": 9.09876537322998,
-      "learning_rate": 2.539243365330322e-05,
-      "loss": 1.3796,
+      "grad_norm": 4.202461242675781,
+      "learning_rate": 0.0004232072275550536,
+      "loss": 0.735,
       "step": 4080
     },
     {
       "epoch": 0.7698099002446829,
-      "grad_norm": 6.666367053985596,
-      "learning_rate": 2.5381140598531905e-05,
-      "loss": 1.2452,
+      "grad_norm": 0.5352344512939453,
+      "learning_rate": 0.0004230190099755317,
+      "loss": 0.6184,
       "step": 4090
     },
     {
       "epoch": 0.7716920760399021,
-      "grad_norm": 53.3235969543457,
-      "learning_rate": 2.5369847543760587e-05,
-      "loss": 1.4592,
+      "grad_norm": 3.9562442302703857,
+      "learning_rate": 0.0004228307923960098,
+      "loss": 0.8374,
       "step": 4100
     },
     {
       "epoch": 0.7735742518351214,
-      "grad_norm": 14.514080047607422,
-      "learning_rate": 2.5358554488989273e-05,
-      "loss": 1.4565,
+      "grad_norm": 4.5890936851501465,
+      "learning_rate": 0.0004226425748164879,
+      "loss": 0.7613,
       "step": 4110
     },
     {
       "epoch": 0.7754564276303407,
-      "grad_norm": 4.159520626068115,
-      "learning_rate": 2.5347261434217956e-05,
-      "loss": 1.2396,
+      "grad_norm": 2.076005458831787,
+      "learning_rate": 0.000422454357236966,
+      "loss": 0.697,
       "step": 4120
     },
     {
       "epoch": 0.7773386034255599,
-      "grad_norm": 6.2468438148498535,
-      "learning_rate": 2.5335968379446642e-05,
-      "loss": 1.2646,
+      "grad_norm": 2.3517327308654785,
+      "learning_rate": 0.00042226613965744406,
+      "loss": 0.8244,
       "step": 4130
     },
     {
       "epoch": 0.7792207792207793,
-      "grad_norm": 14.043899536132812,
-      "learning_rate": 2.5324675324675325e-05,
-      "loss": 1.5409,
+      "grad_norm": 2.801920175552368,
+      "learning_rate": 0.0004220779220779221,
+      "loss": 0.9129,
       "step": 4140
     },
     {
       "epoch": 0.7811029550159985,
-      "grad_norm": 10.741174697875977,
-      "learning_rate": 2.5313382269904007e-05,
-      "loss": 1.2822,
+      "grad_norm": 3.0341615676879883,
+      "learning_rate": 0.00042188970449840015,
+      "loss": 0.6649,
       "step": 4150
     },
     {
       "epoch": 0.7829851308112178,
-      "grad_norm": 5.443540573120117,
-      "learning_rate": 2.5302089215132694e-05,
-      "loss": 1.561,
+      "grad_norm": 1.9421992301940918,
+      "learning_rate": 0.00042170148691887823,
+      "loss": 1.1939,
       "step": 4160
     },
     {
       "epoch": 0.784867306606437,
-      "grad_norm": 4.099931716918945,
-      "learning_rate": 2.529079616036138e-05,
-      "loss": 1.6496,
+      "grad_norm": 2.504599094390869,
+      "learning_rate": 0.0004215132693393563,
+      "loss": 1.0883,
       "step": 4170
     },
     {
       "epoch": 0.7867494824016563,
-      "grad_norm": 4.217696666717529,
-      "learning_rate": 2.5279503105590062e-05,
-      "loss": 1.3227,
+      "grad_norm": 2.8098371028900146,
+      "learning_rate": 0.0004213250517598344,
+      "loss": 0.7397,
       "step": 4180
     },
     {
       "epoch": 0.7886316581968756,
-      "grad_norm": 18.865585327148438,
-      "learning_rate": 2.526821005081875e-05,
-      "loss": 1.5359,
+      "grad_norm": 3.403374433517456,
+      "learning_rate": 0.00042113683418031246,
+      "loss": 0.9343,
       "step": 4190
     },
     {
       "epoch": 0.7905138339920948,
-      "grad_norm": 7.36782693862915,
-      "learning_rate": 2.525691699604743e-05,
-      "loss": 1.3648,
+      "grad_norm": 3.9628524780273438,
+      "learning_rate": 0.00042094861660079054,
+      "loss": 0.9282,
       "step": 4200
     },
     {
       "epoch": 0.7923960097873142,
-      "grad_norm": 11.05923843383789,
-      "learning_rate": 2.5245623941276117e-05,
-      "loss": 1.556,
+      "grad_norm": 4.7820820808410645,
+      "learning_rate": 0.0004207603990212686,
+      "loss": 1.0349,
       "step": 4210
     },
     {
       "epoch": 0.7942781855825334,
-      "grad_norm": 9.203524589538574,
-      "learning_rate": 2.5234330886504803e-05,
-      "loss": 1.4646,
+      "grad_norm": 3.7244391441345215,
+      "learning_rate": 0.0004205721814417467,
+      "loss": 0.9997,
       "step": 4220
     },
     {
       "epoch": 0.7961603613777527,
-      "grad_norm": 9.965381622314453,
-      "learning_rate": 2.5223037831733483e-05,
-      "loss": 1.4233,
+      "grad_norm": 3.3195786476135254,
+      "learning_rate": 0.0004203839638622247,
+      "loss": 0.8765,
       "step": 4230
     },
     {
       "epoch": 0.7980425371729719,
-      "grad_norm": 17.369232177734375,
-      "learning_rate": 2.521174477696217e-05,
-      "loss": 1.4067,
+      "grad_norm": 2.580580234527588,
+      "learning_rate": 0.0004201957462827028,
+      "loss": 0.797,
       "step": 4240
     },
     {
       "epoch": 0.7999247129681912,
-      "grad_norm": 4.989372253417969,
-      "learning_rate": 2.520045172219085e-05,
-      "loss": 1.773,
+      "grad_norm": 2.195201873779297,
+      "learning_rate": 0.00042000752870318086,
+      "loss": 1.167,
       "step": 4250
     },
     {
       "epoch": 0.8018068887634106,
-      "grad_norm": 10.048711776733398,
-      "learning_rate": 2.5189158667419537e-05,
-      "loss": 1.8644,
+      "grad_norm": 2.2976858615875244,
+      "learning_rate": 0.00041981931112365894,
+      "loss": 1.1836,
       "step": 4260
     },
     {
       "epoch": 0.8036890645586298,
-      "grad_norm": 4.219175338745117,
-      "learning_rate": 2.5177865612648223e-05,
-      "loss": 1.133,
+      "grad_norm": 2.622638702392578,
+      "learning_rate": 0.000419631093544137,
+      "loss": 0.7872,
       "step": 4270
     },
     {
       "epoch": 0.8055712403538491,
-      "grad_norm": 20.187997817993164,
-      "learning_rate": 2.5166572557876906e-05,
-      "loss": 1.5066,
+      "grad_norm": 3.801210403442383,
+      "learning_rate": 0.0004194428759646151,
+      "loss": 0.8956,
       "step": 4280
     },
     {
       "epoch": 0.8074534161490683,
-      "grad_norm": 7.595444202423096,
-      "learning_rate": 2.5155279503105592e-05,
-      "loss": 1.6849,
+      "grad_norm": 2.6912474632263184,
+      "learning_rate": 0.00041925465838509317,
+      "loss": 1.2465,
       "step": 4290
     },
     {
       "epoch": 0.8093355919442876,
-      "grad_norm": 14.513413429260254,
-      "learning_rate": 2.5143986448334275e-05,
-      "loss": 1.3276,
+      "grad_norm": 2.3820366859436035,
+      "learning_rate": 0.00041906644080557125,
+      "loss": 0.7562,
       "step": 4300
     },
     {
       "epoch": 0.8112177677395068,
-      "grad_norm": 15.310640335083008,
-      "learning_rate": 2.513269339356296e-05,
-      "loss": 1.3818,
+      "grad_norm": 3.9497461318969727,
+      "learning_rate": 0.0004188782232260493,
+      "loss": 0.8316,
       "step": 4310
     },
     {
       "epoch": 0.8130999435347261,
-      "grad_norm": 10.264412879943848,
-      "learning_rate": 2.5121400338791644e-05,
-      "loss": 1.2301,
+      "grad_norm": 1.9162503480911255,
+      "learning_rate": 0.0004186900056465274,
+      "loss": 0.5897,
       "step": 4320
     },
     {
       "epoch": 0.8149821193299455,
-      "grad_norm": 6.609447002410889,
-      "learning_rate": 2.5110107284020326e-05,
-      "loss": 1.646,
+      "grad_norm": 3.3370280265808105,
+      "learning_rate": 0.0004185017880670055,
+      "loss": 1.5245,
       "step": 4330
     },
     {
       "epoch": 0.8168642951251647,
-      "grad_norm": 5.502641677856445,
-      "learning_rate": 2.5098814229249012e-05,
-      "loss": 1.2485,
+      "grad_norm": 2.484536647796631,
+      "learning_rate": 0.00041831357048748355,
+      "loss": 0.7318,
       "step": 4340
     },
     {
       "epoch": 0.818746470920384,
-      "grad_norm": 10.554330825805664,
-      "learning_rate": 2.5087521174477695e-05,
-      "loss": 1.3115,
+      "grad_norm": 3.806140184402466,
+      "learning_rate": 0.00041812535290796163,
+      "loss": 0.7705,
       "step": 4350
     },
     {
       "epoch": 0.8206286467156032,
-      "grad_norm": 4.31608247756958,
-      "learning_rate": 2.507622811970638e-05,
-      "loss": 1.1539,
+      "grad_norm": 1.3592685461044312,
+      "learning_rate": 0.0004179371353284397,
+      "loss": 0.5629,
       "step": 4360
     },
     {
       "epoch": 0.8225108225108225,
-      "grad_norm": 8.371481895446777,
-      "learning_rate": 2.5064935064935067e-05,
-      "loss": 1.4917,
+      "grad_norm": 4.968286991119385,
+      "learning_rate": 0.0004177489177489178,
+      "loss": 1.035,
       "step": 4370
     },
     {
       "epoch": 0.8243929983060417,
-      "grad_norm": 19.33193016052246,
-      "learning_rate": 2.505364201016375e-05,
-      "loss": 1.2167,
+      "grad_norm": 2.162809133529663,
+      "learning_rate": 0.00041756070016939586,
+      "loss": 0.6883,
       "step": 4380
     },
     {
       "epoch": 0.8262751741012611,
-      "grad_norm": 14.059998512268066,
-      "learning_rate": 2.5042348955392436e-05,
-      "loss": 1.4155,
+      "grad_norm": 3.7793686389923096,
+      "learning_rate": 0.00041737248258987393,
+      "loss": 0.8916,
       "step": 4390
     },
     {
       "epoch": 0.8281573498964804,
-      "grad_norm": 19.17221450805664,
-      "learning_rate": 2.503105590062112e-05,
-      "loss": 1.6316,
+      "grad_norm": 5.029797077178955,
+      "learning_rate": 0.00041718426501035196,
+      "loss": 1.0611,
       "step": 4400
     },
     {
       "epoch": 0.8300395256916996,
-      "grad_norm": 3.8574047088623047,
-      "learning_rate": 2.50197628458498e-05,
-      "loss": 1.3991,
+      "grad_norm": 1.8391931056976318,
+      "learning_rate": 0.00041699604743083003,
+      "loss": 1.018,
       "step": 4410
     },
     {
       "epoch": 0.8319217014869189,
-      "grad_norm": 11.87222671508789,
-      "learning_rate": 2.5008469791078487e-05,
-      "loss": 1.5969,
+      "grad_norm": 2.113842725753784,
+      "learning_rate": 0.0004168078298513081,
+      "loss": 0.9259,
       "step": 4420
     },
     {
       "epoch": 0.8338038772821381,
-      "grad_norm": 5.910922050476074,
-      "learning_rate": 2.499717673630717e-05,
-      "loss": 1.4359,
+      "grad_norm": 1.8083542585372925,
+      "learning_rate": 0.0004166196122717862,
+      "loss": 0.8516,
       "step": 4430
     },
     {
       "epoch": 0.8356860530773574,
-      "grad_norm": 13.17109203338623,
-      "learning_rate": 2.4985883681535856e-05,
-      "loss": 1.3405,
+      "grad_norm": 4.179044723510742,
+      "learning_rate": 0.00041643139469226426,
+      "loss": 0.9547,
       "step": 4440
     },
     {
       "epoch": 0.8375682288725766,
-      "grad_norm": 10.746623039245605,
-      "learning_rate": 2.4974590626764542e-05,
-      "loss": 1.4437,
+      "grad_norm": 4.096711158752441,
+      "learning_rate": 0.00041624317711274234,
+      "loss": 1.0307,
       "step": 4450
     },
     {
       "epoch": 0.839450404667796,
-      "grad_norm": 18.121070861816406,
-      "learning_rate": 2.4963297571993225e-05,
-      "loss": 1.4979,
+      "grad_norm": 2.8632447719573975,
+      "learning_rate": 0.0004160549595332204,
+      "loss": 1.0436,
       "step": 4460
     },
     {
       "epoch": 0.8413325804630153,
-      "grad_norm": 8.530396461486816,
-      "learning_rate": 2.495200451722191e-05,
-      "loss": 1.7246,
+      "grad_norm": 3.738539695739746,
+      "learning_rate": 0.0004158667419536985,
+      "loss": 1.0449,
       "step": 4470
     },
     {
       "epoch": 0.8432147562582345,
-      "grad_norm": 16.504993438720703,
-      "learning_rate": 2.4940711462450594e-05,
-      "loss": 1.4092,
+      "grad_norm": 4.9758148193359375,
+      "learning_rate": 0.00041567852437417657,
+      "loss": 0.7716,
       "step": 4480
     },
     {
       "epoch": 0.8450969320534538,
-      "grad_norm": 7.536314010620117,
-      "learning_rate": 2.4929418407679276e-05,
-      "loss": 0.9995,
+      "grad_norm": 1.671683669090271,
+      "learning_rate": 0.0004154903067946546,
+      "loss": 0.5224,
       "step": 4490
     },
     {
       "epoch": 0.846979107848673,
-      "grad_norm": 5.592616081237793,
-      "learning_rate": 2.4918125352907963e-05,
-      "loss": 1.3312,
+      "grad_norm": 1.9305776357650757,
+      "learning_rate": 0.00041530208921513267,
+      "loss": 0.8533,
       "step": 4500
     },
     {
       "epoch": 0.8488612836438924,
-      "grad_norm": 18.722604751586914,
-      "learning_rate": 2.4906832298136645e-05,
-      "loss": 1.4819,
+      "grad_norm": 4.232862949371338,
+      "learning_rate": 0.00041511387163561074,
+      "loss": 0.9005,
       "step": 4510
     },
     {
       "epoch": 0.8507434594391116,
-      "grad_norm": 16.095701217651367,
-      "learning_rate": 2.489553924336533e-05,
-      "loss": 1.4257,
+      "grad_norm": 2.9629571437835693,
+      "learning_rate": 0.0004149256540560888,
+      "loss": 0.8665,
       "step": 4520
     },
     {
       "epoch": 0.8526256352343309,
-      "grad_norm": 11.100549697875977,
-      "learning_rate": 2.4884246188594014e-05,
-      "loss": 1.4673,
+      "grad_norm": 3.31140398979187,
+      "learning_rate": 0.0004147374364765669,
+      "loss": 0.9225,
       "step": 4530
     },
     {
       "epoch": 0.8545078110295502,
-      "grad_norm": 22.106538772583008,
-      "learning_rate": 2.48729531338227e-05,
-      "loss": 1.3014,
+      "grad_norm": 4.2920823097229,
+      "learning_rate": 0.000414549218897045,
+      "loss": 0.5833,
       "step": 4540
     },
     {
       "epoch": 0.8563899868247694,
-      "grad_norm": 9.377236366271973,
-      "learning_rate": 2.4861660079051386e-05,
-      "loss": 1.5026,
+      "grad_norm": 3.0021259784698486,
+      "learning_rate": 0.0004143610013175231,
+      "loss": 1.0226,
       "step": 4550
     },
     {
       "epoch": 0.8582721626199887,
-      "grad_norm": 7.03391695022583,
-      "learning_rate": 2.485036702428007e-05,
-      "loss": 1.3629,
+      "grad_norm": 2.458348035812378,
+      "learning_rate": 0.0004141727837380012,
+      "loss": 0.8703,
       "step": 4560
     },
     {
       "epoch": 0.8601543384152079,
-      "grad_norm": 19.828781127929688,
-      "learning_rate": 2.4839073969508755e-05,
-      "loss": 1.6024,
+      "grad_norm": 4.885277271270752,
+      "learning_rate": 0.00041398456615847926,
+      "loss": 1.175,
       "step": 4570
     },
     {
       "epoch": 0.8620365142104273,
-      "grad_norm": 14.324966430664062,
-      "learning_rate": 2.4827780914737434e-05,
-      "loss": 1.3317,
+      "grad_norm": 5.482606410980225,
+      "learning_rate": 0.0004137963485789573,
+      "loss": 0.8186,
       "step": 4580
     },
     {
       "epoch": 0.8639186900056465,
-      "grad_norm": 17.969161987304688,
-      "learning_rate": 2.481648785996612e-05,
-      "loss": 1.5452,
+      "grad_norm": 4.906975746154785,
+      "learning_rate": 0.00041360813099943535,
+      "loss": 0.9061,
       "step": 4590
     },
     {
       "epoch": 0.8658008658008658,
-      "grad_norm": 28.252727508544922,
-      "learning_rate": 2.4805194805194806e-05,
-      "loss": 1.3952,
+      "grad_norm": 3.1498374938964844,
+      "learning_rate": 0.00041341991341991343,
+      "loss": 0.9264,
       "step": 4600
     },
     {
       "epoch": 0.8676830415960851,
-      "grad_norm": 6.551124095916748,
-      "learning_rate": 2.479390175042349e-05,
-      "loss": 1.4017,
+      "grad_norm": 3.5111498832702637,
+      "learning_rate": 0.0004132316958403915,
+      "loss": 1.0421,
       "step": 4610
     },
     {
       "epoch": 0.8695652173913043,
-      "grad_norm": 9.968870162963867,
-      "learning_rate": 2.4782608695652175e-05,
-      "loss": 1.4794,
+      "grad_norm": 2.239441156387329,
+      "learning_rate": 0.0004130434782608696,
+      "loss": 0.922,
       "step": 4620
     },
     {
       "epoch": 0.8714473931865236,
-      "grad_norm": 10.416319847106934,
-      "learning_rate": 2.4771315640880858e-05,
-      "loss": 1.4633,
+      "grad_norm": 2.3396077156066895,
+      "learning_rate": 0.00041285526068134766,
+      "loss": 0.9517,
       "step": 4630
     },
     {
       "epoch": 0.8733295689817429,
-      "grad_norm": 12.675484657287598,
-      "learning_rate": 2.4760022586109544e-05,
-      "loss": 1.1787,
+      "grad_norm": 3.4688830375671387,
+      "learning_rate": 0.00041266704310182574,
+      "loss": 0.6071,
       "step": 4640
     },
     {
       "epoch": 0.8752117447769622,
-      "grad_norm": 4.599917411804199,
-      "learning_rate": 2.474872953133823e-05,
-      "loss": 1.4834,
+      "grad_norm": 3.2922751903533936,
+      "learning_rate": 0.0004124788255223038,
+      "loss": 0.892,
       "step": 4650
     },
     {
       "epoch": 0.8770939205721815,
-      "grad_norm": 23.191003799438477,
-      "learning_rate": 2.473743647656691e-05,
-      "loss": 1.2879,
+      "grad_norm": 2.0869195461273193,
+      "learning_rate": 0.00041229060794278183,
+      "loss": 0.8705,
       "step": 4660
     },
     {
       "epoch": 0.8789760963674007,
-      "grad_norm": 10.024175643920898,
-      "learning_rate": 2.4726143421795595e-05,
-      "loss": 1.1128,
+      "grad_norm": 0.24055927991867065,
+      "learning_rate": 0.0004121023903632599,
+      "loss": 0.722,
       "step": 4670
     },
     {
       "epoch": 0.88085827216262,
-      "grad_norm": 3.1574692726135254,
-      "learning_rate": 2.471485036702428e-05,
-      "loss": 1.2905,
+      "grad_norm": 1.8887193202972412,
+      "learning_rate": 0.000411914172783738,
+      "loss": 0.9261,
       "step": 4680
     },
     {
       "epoch": 0.8827404479578392,
-      "grad_norm": 5.446253776550293,
-      "learning_rate": 2.4703557312252964e-05,
-      "loss": 1.5343,
+      "grad_norm": 9.118612289428711,
+      "learning_rate": 0.00041172595520421606,
+      "loss": 1.1263,
       "step": 4690
     },
     {
       "epoch": 0.8846226237530586,
-      "grad_norm": 13.522638320922852,
-      "learning_rate": 2.469226425748165e-05,
-      "loss": 1.3689,
+      "grad_norm": 4.854977607727051,
+      "learning_rate": 0.00041153773762469414,
+      "loss": 0.7742,
       "step": 4700
     },
     {
       "epoch": 0.8865047995482778,
-      "grad_norm": 20.1780948638916,
-      "learning_rate": 2.4680971202710333e-05,
-      "loss": 1.4354,
+      "grad_norm": 2.078373908996582,
+      "learning_rate": 0.0004113495200451722,
+      "loss": 0.8702,
       "step": 4710
     },
     {
       "epoch": 0.8883869753434971,
-      "grad_norm": 10.403576850891113,
-      "learning_rate": 2.466967814793902e-05,
-      "loss": 1.5017,
+      "grad_norm": 3.6736721992492676,
+      "learning_rate": 0.0004111613024656503,
+      "loss": 0.9657,
       "step": 4720
     },
     {
       "epoch": 0.8902691511387164,
-      "grad_norm": 9.197463989257812,
-      "learning_rate": 2.4658385093167705e-05,
-      "loss": 1.2154,
+      "grad_norm": 3.352625846862793,
+      "learning_rate": 0.00041097308488612837,
+      "loss": 0.6156,
       "step": 4730
     },
     {
       "epoch": 0.8921513269339356,
-      "grad_norm": 11.027040481567383,
-      "learning_rate": 2.4647092038396388e-05,
-      "loss": 1.1206,
+      "grad_norm": 5.416534900665283,
+      "learning_rate": 0.00041078486730660645,
+      "loss": 0.8606,
       "step": 4740
     },
     {
       "epoch": 0.8940335027291549,
-      "grad_norm": 10.964394569396973,
-      "learning_rate": 2.463579898362507e-05,
-      "loss": 1.3562,
+      "grad_norm": 1.8365694284439087,
+      "learning_rate": 0.00041059664972708447,
+      "loss": 0.9732,
       "step": 4750
     },
     {
       "epoch": 0.8959156785243741,
-      "grad_norm": 8.044187545776367,
-      "learning_rate": 2.4624505928853753e-05,
-      "loss": 1.1975,
+      "grad_norm": 1.2886053323745728,
+      "learning_rate": 0.00041040843214756254,
+      "loss": 0.5367,
       "step": 4760
     },
     {
       "epoch": 0.8977978543195935,
-      "grad_norm": 12.479483604431152,
-      "learning_rate": 2.461321287408244e-05,
-      "loss": 1.3726,
+      "grad_norm": 1.6298965215682983,
+      "learning_rate": 0.0004102202145680407,
+      "loss": 0.8823,
       "step": 4770
     },
     {
       "epoch": 0.8996800301148127,
-      "grad_norm": 10.56410026550293,
-      "learning_rate": 2.4601919819311125e-05,
-      "loss": 1.3986,
+      "grad_norm": 3.4384331703186035,
+      "learning_rate": 0.00041003199698851875,
+      "loss": 0.9704,
       "step": 4780
     },
     {
       "epoch": 0.901562205910032,
-      "grad_norm": 13.470977783203125,
-      "learning_rate": 2.4590626764539808e-05,
-      "loss": 1.2694,
+      "grad_norm": 2.444072961807251,
+      "learning_rate": 0.00040984377940899683,
+      "loss": 0.7552,
       "step": 4790
     },
     {
       "epoch": 0.9034443817052513,
-      "grad_norm": 4.0369744300842285,
-      "learning_rate": 2.4579333709768494e-05,
-      "loss": 1.0665,
+      "grad_norm": 1.4015096426010132,
+      "learning_rate": 0.0004096555618294749,
+      "loss": 0.8217,
       "step": 4800
     },
     {
       "epoch": 0.9053265575004705,
-      "grad_norm": 15.835465431213379,
-      "learning_rate": 2.4568040654997177e-05,
-      "loss": 1.2974,
+      "grad_norm": 3.5089902877807617,
+      "learning_rate": 0.000409467344249953,
+      "loss": 0.8258,
       "step": 4810
     },
     {
       "epoch": 0.9072087332956899,
-      "grad_norm": 13.87853717803955,
-      "learning_rate": 2.4556747600225863e-05,
-      "loss": 1.3459,
+      "grad_norm": 3.8666586875915527,
+      "learning_rate": 0.00040927912667043106,
+      "loss": 0.8321,
       "step": 4820
     },
     {
       "epoch": 0.9090909090909091,
-      "grad_norm": 27.265249252319336,
-      "learning_rate": 2.454545454545455e-05,
-      "loss": 1.3109,
+      "grad_norm": 2.164626359939575,
+      "learning_rate": 0.00040909090909090913,
+      "loss": 0.8906,
       "step": 4830
     },
     {
       "epoch": 0.9109730848861284,
-      "grad_norm": 5.671602725982666,
-      "learning_rate": 2.4534161490683228e-05,
-      "loss": 1.6534,
+      "grad_norm": 1.8831568956375122,
+      "learning_rate": 0.00040890269151138716,
+      "loss": 1.1285,
       "step": 4840
     },
     {
       "epoch": 0.9128552606813476,
-      "grad_norm": 4.075274467468262,
-      "learning_rate": 2.4522868435911914e-05,
-      "loss": 1.3106,
+      "grad_norm": 2.536550998687744,
+      "learning_rate": 0.00040871447393186523,
+      "loss": 0.8967,
       "step": 4850
     },
     {
       "epoch": 0.9147374364765669,
-      "grad_norm": 17.051898956298828,
-      "learning_rate": 2.4511575381140597e-05,
-      "loss": 1.2559,
+      "grad_norm": 2.850433826446533,
+      "learning_rate": 0.0004085262563523433,
+      "loss": 0.7869,
       "step": 4860
     },
     {
       "epoch": 0.9166196122717862,
-      "grad_norm": 14.164190292358398,
-      "learning_rate": 2.4500282326369283e-05,
-      "loss": 1.4232,
+      "grad_norm": 3.7627031803131104,
+      "learning_rate": 0.0004083380387728214,
+      "loss": 0.8574,
       "step": 4870
     },
     {
       "epoch": 0.9185017880670054,
-      "grad_norm": 4.818867206573486,
-      "learning_rate": 2.448898927159797e-05,
-      "loss": 1.1879,
+      "grad_norm": 3.227433919906616,
+      "learning_rate": 0.00040814982119329946,
+      "loss": 0.6237,
       "step": 4880
     },
     {
       "epoch": 0.9203839638622248,
-      "grad_norm": 6.399837493896484,
-      "learning_rate": 2.4477696216826652e-05,
-      "loss": 1.3137,
+      "grad_norm": 2.9040145874023438,
+      "learning_rate": 0.00040796160361377754,
+      "loss": 0.8478,
       "step": 4890
     },
     {
       "epoch": 0.922266139657444,
-      "grad_norm": 26.00821304321289,
-      "learning_rate": 2.4466403162055338e-05,
-      "loss": 1.4008,
+      "grad_norm": 4.651259899139404,
+      "learning_rate": 0.0004077733860342556,
+      "loss": 0.9602,
       "step": 4900
     },
     {
       "epoch": 0.9241483154526633,
-      "grad_norm": 19.372074127197266,
-      "learning_rate": 2.445511010728402e-05,
-      "loss": 1.4288,
+      "grad_norm": 2.856947183609009,
+      "learning_rate": 0.0004075851684547337,
+      "loss": 1.1861,
       "step": 4910
     },
     {
       "epoch": 0.9260304912478825,
-      "grad_norm": 21.262939453125,
-      "learning_rate": 2.4443817052512707e-05,
-      "loss": 1.3958,
+      "grad_norm": 3.3068671226501465,
+      "learning_rate": 0.00040739695087521177,
+      "loss": 0.9059,
       "step": 4920
     },
     {
       "epoch": 0.9279126670431018,
-      "grad_norm": 8.744524955749512,
-      "learning_rate": 2.443252399774139e-05,
-      "loss": 1.2075,
+      "grad_norm": 2.0299556255340576,
+      "learning_rate": 0.0004072087332956898,
+      "loss": 0.9159,
       "step": 4930
     },
     {
       "epoch": 0.9297948428383211,
-      "grad_norm": 19.05074691772461,
-      "learning_rate": 2.4421230942970072e-05,
-      "loss": 1.4352,
+      "grad_norm": 4.704328536987305,
+      "learning_rate": 0.00040702051571616787,
+      "loss": 1.2919,
       "step": 4940
     },
     {
       "epoch": 0.9316770186335404,
-      "grad_norm": 4.271505355834961,
-      "learning_rate": 2.4409937888198758e-05,
-      "loss": 0.8662,
+      "grad_norm": 4.627618312835693,
+      "learning_rate": 0.00040683229813664594,
+      "loss": 0.5245,
       "step": 4950
     },
     {
       "epoch": 0.9335591944287597,
-      "grad_norm": 11.079838752746582,
-      "learning_rate": 2.4398644833427444e-05,
-      "loss": 0.9989,
+      "grad_norm": 4.851251125335693,
+      "learning_rate": 0.000406644080557124,
+      "loss": 0.6766,
       "step": 4960
     },
     {
       "epoch": 0.9354413702239789,
-      "grad_norm": 12.747160911560059,
-      "learning_rate": 2.4387351778656127e-05,
-      "loss": 1.3546,
+      "grad_norm": 3.3305420875549316,
+      "learning_rate": 0.0004064558629776021,
+      "loss": 0.966,
       "step": 4970
     },
     {
       "epoch": 0.9373235460191982,
-      "grad_norm": 9.06821060180664,
-      "learning_rate": 2.4376058723884813e-05,
-      "loss": 1.237,
+      "grad_norm": 3.7064177989959717,
+      "learning_rate": 0.0004062676453980802,
+      "loss": 0.8435,
       "step": 4980
     },
     {
       "epoch": 0.9392057218144174,
-      "grad_norm": 18.579593658447266,
-      "learning_rate": 2.4364765669113496e-05,
-      "loss": 1.0582,
+      "grad_norm": 3.640709161758423,
+      "learning_rate": 0.0004060794278185583,
+      "loss": 0.6872,
       "step": 4990
     },
     {
       "epoch": 0.9410878976096367,
-      "grad_norm": 25.581363677978516,
-      "learning_rate": 2.435347261434218e-05,
-      "loss": 1.1142,
+      "grad_norm": 4.062694549560547,
+      "learning_rate": 0.0004058912102390364,
+      "loss": 0.8231,
       "step": 5000
     },
     {
       "epoch": 0.9429700734048561,
-      "grad_norm": 3.7508864402770996,
-      "learning_rate": 2.4342179559570864e-05,
-      "loss": 1.1593,
+      "grad_norm": 5.525752544403076,
+      "learning_rate": 0.0004057029926595144,
+      "loss": 0.7747,
       "step": 5010
     },
     {
       "epoch": 0.9448522492000753,
-      "grad_norm": 12.199710845947266,
-      "learning_rate": 2.4330886504799547e-05,
-      "loss": 1.2923,
+      "grad_norm": 3.2349648475646973,
+      "learning_rate": 0.0004055147750799925,
+      "loss": 0.9788,
       "step": 5020
     },
     {
       "epoch": 0.9467344249952946,
-      "grad_norm": 15.573914527893066,
-      "learning_rate": 2.4319593450028233e-05,
-      "loss": 1.2529,
+      "grad_norm": 3.297736406326294,
+      "learning_rate": 0.00040532655750047055,
+      "loss": 0.9376,
       "step": 5030
     },
     {
       "epoch": 0.9486166007905138,
-      "grad_norm": 10.434758186340332,
-      "learning_rate": 2.4308300395256916e-05,
-      "loss": 1.2285,
+      "grad_norm": 4.1556549072265625,
+      "learning_rate": 0.00040513833992094863,
+      "loss": 0.8717,
       "step": 5040
     },
     {
       "epoch": 0.9504987765857331,
-      "grad_norm": 14.420498847961426,
-      "learning_rate": 2.4297007340485602e-05,
-      "loss": 1.3006,
+      "grad_norm": 4.240426063537598,
+      "learning_rate": 0.0004049501223414267,
+      "loss": 0.8023,
       "step": 5050
     },
     {
       "epoch": 0.9523809523809523,
-      "grad_norm": 32.99720764160156,
-      "learning_rate": 2.4285714285714288e-05,
-      "loss": 1.4461,
+      "grad_norm": 3.5286829471588135,
+      "learning_rate": 0.0004047619047619048,
+      "loss": 0.8408,
       "step": 5060
     },
     {
       "epoch": 0.9542631281761716,
-      "grad_norm": 4.45308780670166,
-      "learning_rate": 2.427442123094297e-05,
-      "loss": 1.1251,
+      "grad_norm": 3.1347432136535645,
+      "learning_rate": 0.00040457368718238286,
+      "loss": 0.6977,
       "step": 5070
     },
     {
       "epoch": 0.956145303971391,
-      "grad_norm": 23.680532455444336,
-      "learning_rate": 2.4263128176171657e-05,
-      "loss": 1.0521,
+      "grad_norm": 2.5033462047576904,
+      "learning_rate": 0.00040438546960286094,
+      "loss": 0.8087,
       "step": 5080
     },
     {
       "epoch": 0.9580274797666102,
-      "grad_norm": 7.266300201416016,
-      "learning_rate": 2.425183512140034e-05,
-      "loss": 1.2313,
+      "grad_norm": 1.7074840068817139,
+      "learning_rate": 0.000404197252023339,
+      "loss": 0.8927,
       "step": 5090
     },
     {
       "epoch": 0.9599096555618295,
-      "grad_norm": 9.216297149658203,
-      "learning_rate": 2.4240542066629022e-05,
-      "loss": 1.1339,
+      "grad_norm": 5.167369365692139,
+      "learning_rate": 0.00040400903444381703,
+      "loss": 0.9689,
       "step": 5100
     },
     {
       "epoch": 0.9617918313570487,
-      "grad_norm": 11.177430152893066,
-      "learning_rate": 2.4229249011857708e-05,
-      "loss": 1.1926,
+      "grad_norm": 2.3252077102661133,
+      "learning_rate": 0.0004038208168642951,
+      "loss": 0.8471,
       "step": 5110
     },
     {
       "epoch": 0.963674007152268,
-      "grad_norm": 21.9661808013916,
-      "learning_rate": 2.421795595708639e-05,
-      "loss": 1.4161,
+      "grad_norm": 3.6321616172790527,
+      "learning_rate": 0.0004036325992847732,
+      "loss": 0.9404,
       "step": 5120
     },
     {
       "epoch": 0.9655561829474872,
-      "grad_norm": 12.636842727661133,
-      "learning_rate": 2.4206662902315077e-05,
-      "loss": 1.1809,
+      "grad_norm": 3.3031723499298096,
+      "learning_rate": 0.00040344438170525126,
+      "loss": 0.7534,
       "step": 5130
     },
     {
       "epoch": 0.9674383587427066,
-      "grad_norm": 12.683283805847168,
-      "learning_rate": 2.419536984754376e-05,
-      "loss": 1.2115,
+      "grad_norm": 3.7766780853271484,
+      "learning_rate": 0.00040325616412572934,
+      "loss": 0.7902,
       "step": 5140
     },
     {
       "epoch": 0.9693205345379259,
-      "grad_norm": 14.036438941955566,
-      "learning_rate": 2.4184076792772446e-05,
-      "loss": 1.4084,
+      "grad_norm": 4.320493221282959,
+      "learning_rate": 0.0004030679465462074,
+      "loss": 0.8989,
       "step": 5150
     },
     {
       "epoch": 0.9712027103331451,
-      "grad_norm": 8.703644752502441,
-      "learning_rate": 2.4172783738001132e-05,
-      "loss": 1.3985,
+      "grad_norm": 3.0502378940582275,
+      "learning_rate": 0.0004028797289666855,
+      "loss": 1.1857,
       "step": 5160
     },
     {
       "epoch": 0.9730848861283644,
-      "grad_norm": 15.209477424621582,
-      "learning_rate": 2.4161490683229814e-05,
-      "loss": 1.3594,
+      "grad_norm": 6.050513744354248,
+      "learning_rate": 0.00040269151138716357,
+      "loss": 1.2846,
       "step": 5170
     },
     {
       "epoch": 0.9749670619235836,
-      "grad_norm": 8.397993087768555,
-      "learning_rate": 2.41501976284585e-05,
-      "loss": 1.2813,
+      "grad_norm": 2.5459749698638916,
+      "learning_rate": 0.00040250329380764165,
+      "loss": 0.7691,
       "step": 5180
     },
     {
       "epoch": 0.9768492377188029,
-      "grad_norm": 11.615766525268555,
-      "learning_rate": 2.4138904573687183e-05,
-      "loss": 1.1447,
+      "grad_norm": 3.656423568725586,
+      "learning_rate": 0.00040231507622811967,
+      "loss": 0.7498,
       "step": 5190
     },
     {
       "epoch": 0.9787314135140223,
-      "grad_norm": 3.9449048042297363,
-      "learning_rate": 2.4127611518915866e-05,
-      "loss": 1.0322,
+      "grad_norm": 3.0081403255462646,
+      "learning_rate": 0.00040212685864859774,
+      "loss": 0.7818,
       "step": 5200
     },
     {
       "epoch": 0.9806135893092415,
-      "grad_norm": 23.256526947021484,
-      "learning_rate": 2.4116318464144552e-05,
-      "loss": 1.0706,
+      "grad_norm": 6.50275182723999,
+      "learning_rate": 0.0004019386410690759,
+      "loss": 0.8074,
       "step": 5210
     },
     {
       "epoch": 0.9824957651044608,
-      "grad_norm": 6.652702808380127,
-      "learning_rate": 2.4105025409373235e-05,
-      "loss": 1.2482,
+      "grad_norm": 8.647149085998535,
+      "learning_rate": 0.00040175042348955395,
+      "loss": 0.9535,
       "step": 5220
     },
     {
       "epoch": 0.98437794089968,
-      "grad_norm": 27.787336349487305,
-      "learning_rate": 2.409373235460192e-05,
-      "loss": 1.2317,
+      "grad_norm": 1.342649221420288,
+      "learning_rate": 0.00040156220591003203,
+      "loss": 0.7369,
       "step": 5230
     },
     {
       "epoch": 0.9862601166948993,
-      "grad_norm": 4.631372451782227,
-      "learning_rate": 2.4082439299830607e-05,
-      "loss": 0.9322,
+      "grad_norm": 3.2202084064483643,
+      "learning_rate": 0.0004013739883305101,
+      "loss": 0.5109,
       "step": 5240
     },
     {
       "epoch": 0.9881422924901185,
-      "grad_norm": 7.675413131713867,
-      "learning_rate": 2.407114624505929e-05,
-      "loss": 1.1535,
+      "grad_norm": 3.334763765335083,
+      "learning_rate": 0.0004011857707509882,
+      "loss": 0.9238,
       "step": 5250
     },
     {
       "epoch": 0.9900244682853379,
-      "grad_norm": 11.056670188903809,
-      "learning_rate": 2.4059853190287976e-05,
-      "loss": 1.066,
+      "grad_norm": 1.7026392221450806,
+      "learning_rate": 0.00040099755317146626,
+      "loss": 0.841,
       "step": 5260
     },
     {
       "epoch": 0.9919066440805572,
-      "grad_norm": 2.178229808807373,
-      "learning_rate": 2.4048560135516655e-05,
-      "loss": 1.0266,
+      "grad_norm": 1.0753185749053955,
+      "learning_rate": 0.0004008093355919443,
+      "loss": 0.7236,
       "step": 5270
     },
     {
       "epoch": 0.9937888198757764,
-      "grad_norm": 5.711111068725586,
-      "learning_rate": 2.403726708074534e-05,
-      "loss": 1.2493,
+      "grad_norm": 2.5947108268737793,
+      "learning_rate": 0.00040062111801242236,
+      "loss": 1.0401,
       "step": 5280
     },
     {
       "epoch": 0.9956709956709957,
-      "grad_norm": 27.90021514892578,
-      "learning_rate": 2.4025974025974027e-05,
-      "loss": 1.1559,
+      "grad_norm": 2.397989273071289,
+      "learning_rate": 0.00040043290043290043,
+      "loss": 0.5921,
       "step": 5290
     },
     {
       "epoch": 0.9975531714662149,
-      "grad_norm": 8.907947540283203,
-      "learning_rate": 2.401468097120271e-05,
-      "loss": 1.0816,
+      "grad_norm": 2.8067519664764404,
+      "learning_rate": 0.0004002446828533785,
+      "loss": 0.9171,
       "step": 5300
     },
     {
       "epoch": 0.9994353472614342,
-      "grad_norm": 21.335691452026367,
-      "learning_rate": 2.4003387916431396e-05,
-      "loss": 1.0634,
+      "grad_norm": 3.1910457611083984,
+      "learning_rate": 0.0004000564652738566,
+      "loss": 0.5206,
       "step": 5310
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.8777333333333334,
-      "eval_loss": 0.7235307097434998,
-      "eval_runtime": 244.378,
-      "eval_samples_per_second": 30.69,
-      "eval_steps_per_second": 3.838,
+      "eval_accuracy": 0.9050666666666667,
+      "eval_loss": 0.33233147859573364,
+      "eval_runtime": 307.7593,
+      "eval_samples_per_second": 24.37,
+      "eval_steps_per_second": 3.048,
       "step": 5313
     },
     {
       "epoch": 1.0013175230566536,
-      "grad_norm": 5.803168773651123,
-      "learning_rate": 2.399209486166008e-05,
-      "loss": 1.4338,
+      "grad_norm": 2.1240122318267822,
+      "learning_rate": 0.00039986824769433466,
+      "loss": 1.0903,
       "step": 5320
     },
     {
       "epoch": 1.0031996988518728,
-      "grad_norm": 16.894943237304688,
-      "learning_rate": 2.3980801806888765e-05,
-      "loss": 1.3967,
+      "grad_norm": 2.710055351257324,
+      "learning_rate": 0.00039968003011481274,
+      "loss": 0.9562,
       "step": 5330
     },
     {
       "epoch": 1.005081874647092,
-      "grad_norm": 9.325886726379395,
-      "learning_rate": 2.396950875211745e-05,
-      "loss": 1.1418,
+      "grad_norm": 3.0997753143310547,
+      "learning_rate": 0.0003994918125352908,
+      "loss": 0.7154,
       "step": 5340
     },
     {
       "epoch": 1.0069640504423114,
-      "grad_norm": 15.385432243347168,
-      "learning_rate": 2.3958215697346133e-05,
-      "loss": 1.1674,
+      "grad_norm": 2.385953903198242,
+      "learning_rate": 0.0003993035949557689,
+      "loss": 0.933,
       "step": 5350
     },
     {
       "epoch": 1.0088462262375306,
-      "grad_norm": 16.328964233398438,
-      "learning_rate": 2.3946922642574816e-05,
-      "loss": 1.2397,
+      "grad_norm": 2.483708620071411,
+      "learning_rate": 0.0003991153773762469,
+      "loss": 0.7209,
       "step": 5360
     },
     {
       "epoch": 1.0107284020327498,
-      "grad_norm": 16.835899353027344,
-      "learning_rate": 2.39356295878035e-05,
-      "loss": 1.1855,
+      "grad_norm": 2.988840341567993,
+      "learning_rate": 0.000398927159796725,
+      "loss": 0.9888,
       "step": 5370
     },
     {
       "epoch": 1.012610577827969,
-      "grad_norm": 1.2168173789978027,
-      "learning_rate": 2.3924336533032185e-05,
-      "loss": 1.185,
+      "grad_norm": 0.3290826380252838,
+      "learning_rate": 0.00039873894221720307,
+      "loss": 0.9194,
       "step": 5380
     },
     {
       "epoch": 1.0144927536231885,
-      "grad_norm": 17.501239776611328,
-      "learning_rate": 2.391304347826087e-05,
-      "loss": 1.0554,
+      "grad_norm": 3.4336321353912354,
+      "learning_rate": 0.00039855072463768114,
+      "loss": 0.6798,
       "step": 5390
     },
     {
       "epoch": 1.0163749294184077,
-      "grad_norm": 7.978825569152832,
-      "learning_rate": 2.3901750423489554e-05,
-      "loss": 1.0179,
+      "grad_norm": 2.0959396362304688,
+      "learning_rate": 0.0003983625070581592,
+      "loss": 0.7914,
       "step": 5400
     },
     {
       "epoch": 1.018257105213627,
-      "grad_norm": 21.958223342895508,
-      "learning_rate": 2.389045736871824e-05,
-      "loss": 1.2771,
+      "grad_norm": 3.9184632301330566,
+      "learning_rate": 0.0003981742894786373,
+      "loss": 0.798,
       "step": 5410
     },
     {
       "epoch": 1.0201392810088463,
-      "grad_norm": 10.229572296142578,
-      "learning_rate": 2.3879164313946926e-05,
-      "loss": 0.9884,
+      "grad_norm": 5.386630535125732,
+      "learning_rate": 0.00039798607189911537,
+      "loss": 0.7589,
       "step": 5420
     },
     {
       "epoch": 1.0220214568040655,
-      "grad_norm": 10.10629653930664,
-      "learning_rate": 2.386787125917561e-05,
-      "loss": 1.2098,
+      "grad_norm": 3.6790056228637695,
+      "learning_rate": 0.0003977978543195935,
+      "loss": 1.0433,
       "step": 5430
     },
     {
       "epoch": 1.0239036325992847,
-      "grad_norm": 8.695528030395508,
-      "learning_rate": 2.3856578204404294e-05,
-      "loss": 1.0261,
+      "grad_norm": 3.881471872329712,
+      "learning_rate": 0.0003976096367400716,
+      "loss": 0.8501,
       "step": 5440
     },
     {
       "epoch": 1.025785808394504,
-      "grad_norm": 3.444990873336792,
-      "learning_rate": 2.3845285149632974e-05,
-      "loss": 1.0464,
+      "grad_norm": 2.0723538398742676,
+      "learning_rate": 0.0003974214191605496,
+      "loss": 0.8193,
       "step": 5450
     },
     {
       "epoch": 1.0276679841897234,
-      "grad_norm": 6.334211826324463,
-      "learning_rate": 2.383399209486166e-05,
-      "loss": 1.1021,
+      "grad_norm": 2.0003159046173096,
+      "learning_rate": 0.0003972332015810277,
+      "loss": 0.8174,
       "step": 5460
     },
     {
       "epoch": 1.0295501599849426,
-      "grad_norm": 7.3270978927612305,
-      "learning_rate": 2.3822699040090346e-05,
-      "loss": 1.0433,
+      "grad_norm": 1.533605933189392,
+      "learning_rate": 0.00039704498400150575,
+      "loss": 0.6906,
       "step": 5470
     },
     {
       "epoch": 1.0314323357801618,
-      "grad_norm": 19.58551788330078,
-      "learning_rate": 2.381140598531903e-05,
-      "loss": 0.9511,
+      "grad_norm": 3.0430614948272705,
+      "learning_rate": 0.00039685676642198383,
+      "loss": 0.6862,
       "step": 5480
     },
     {
       "epoch": 1.0333145115753812,
-      "grad_norm": 11.17794132232666,
-      "learning_rate": 2.3800112930547715e-05,
-      "loss": 1.3029,
+      "grad_norm": 4.184202671051025,
+      "learning_rate": 0.0003966685488424619,
+      "loss": 1.1837,
       "step": 5490
     },
     {
       "epoch": 1.0351966873706004,
-      "grad_norm": 20.934467315673828,
-      "learning_rate": 2.3788819875776397e-05,
-      "loss": 1.2481,
+      "grad_norm": 7.424351692199707,
+      "learning_rate": 0.00039648033126294,
+      "loss": 1.0442,
       "step": 5500
     },
     {
       "epoch": 1.0370788631658197,
-      "grad_norm": 5.4833083152771,
-      "learning_rate": 2.3777526821005083e-05,
-      "loss": 0.9677,
+      "grad_norm": 2.174436569213867,
+      "learning_rate": 0.00039629211368341806,
+      "loss": 0.7614,
       "step": 5510
     },
     {
       "epoch": 1.0389610389610389,
-      "grad_norm": 17.41068458557129,
-      "learning_rate": 2.376623376623377e-05,
-      "loss": 1.0862,
+      "grad_norm": 2.585970401763916,
+      "learning_rate": 0.00039610389610389614,
+      "loss": 0.7771,
       "step": 5520
     },
     {
       "epoch": 1.0408432147562583,
-      "grad_norm": 10.463713645935059,
-      "learning_rate": 2.3754940711462452e-05,
-      "loss": 1.1165,
+      "grad_norm": 2.9651038646698,
+      "learning_rate": 0.0003959156785243742,
+      "loss": 0.6379,
       "step": 5530
     },
     {
       "epoch": 1.0427253905514775,
-      "grad_norm": 4.4030232429504395,
-      "learning_rate": 2.3743647656691135e-05,
-      "loss": 0.8207,
+      "grad_norm": 2.588423013687134,
+      "learning_rate": 0.00039572746094485223,
+      "loss": 0.5153,
       "step": 5540
     },
     {
       "epoch": 1.0446075663466967,
-      "grad_norm": 8.17664909362793,
-      "learning_rate": 2.3732354601919818e-05,
-      "loss": 0.8548,
+      "grad_norm": 2.387333393096924,
+      "learning_rate": 0.0003955392433653303,
+      "loss": 0.5602,
       "step": 5550
     },
     {
       "epoch": 1.0464897421419161,
-      "grad_norm": 18.136306762695312,
-      "learning_rate": 2.3721061547148504e-05,
-      "loss": 1.0368,
+      "grad_norm": 2.9230353832244873,
+      "learning_rate": 0.0003953510257858084,
+      "loss": 0.7409,
       "step": 5560
     },
     {
       "epoch": 1.0483719179371354,
-      "grad_norm": 8.143134117126465,
-      "learning_rate": 2.370976849237719e-05,
-      "loss": 1.1101,
+      "grad_norm": 1.8143259286880493,
+      "learning_rate": 0.00039516280820628646,
+      "loss": 0.8831,
       "step": 5570
     },
     {
       "epoch": 1.0502540937323546,
-      "grad_norm": 17.662364959716797,
-      "learning_rate": 2.3698475437605872e-05,
-      "loss": 1.3467,
+      "grad_norm": 5.663400650024414,
+      "learning_rate": 0.00039497459062676454,
+      "loss": 1.0302,
       "step": 5580
     },
     {
       "epoch": 1.0521362695275738,
-      "grad_norm": 18.27084732055664,
-      "learning_rate": 2.368718238283456e-05,
-      "loss": 0.7784,
+      "grad_norm": 2.289546489715576,
+      "learning_rate": 0.0003947863730472426,
+      "loss": 0.4768,
       "step": 5590
     },
     {
       "epoch": 1.0540184453227932,
-      "grad_norm": 20.256084442138672,
-      "learning_rate": 2.367588932806324e-05,
-      "loss": 0.9596,
+      "grad_norm": 3.498897075653076,
+      "learning_rate": 0.0003945981554677207,
+      "loss": 0.8049,
       "step": 5600
     },
     {
       "epoch": 1.0559006211180124,
-      "grad_norm": 5.476850509643555,
-      "learning_rate": 2.3664596273291927e-05,
-      "loss": 1.2,
+      "grad_norm": 1.5211207866668701,
+      "learning_rate": 0.00039440993788819877,
+      "loss": 0.8888,
       "step": 5610
     },
     {
       "epoch": 1.0577827969132316,
-      "grad_norm": 13.74804973602295,
-      "learning_rate": 2.365330321852061e-05,
-      "loss": 1.3438,
+      "grad_norm": 2.7362804412841797,
+      "learning_rate": 0.0003942217203086768,
+      "loss": 1.1457,
       "step": 5620
     },
     {
       "epoch": 1.059664972708451,
-      "grad_norm": 7.789999008178711,
-      "learning_rate": 2.3642010163749293e-05,
-      "loss": 0.8268,
+      "grad_norm": 3.4366722106933594,
+      "learning_rate": 0.00039403350272915487,
+      "loss": 0.6405,
       "step": 5630
     },
     {
       "epoch": 1.0615471485036703,
-      "grad_norm": 10.119312286376953,
-      "learning_rate": 2.363071710897798e-05,
-      "loss": 1.03,
+      "grad_norm": 2.035243034362793,
+      "learning_rate": 0.00039384528514963294,
+      "loss": 0.724,
       "step": 5640
     },
     {
       "epoch": 1.0634293242988895,
-      "grad_norm": 17.798921585083008,
-      "learning_rate": 2.361942405420666e-05,
-      "loss": 0.9418,
+      "grad_norm": 3.6405818462371826,
+      "learning_rate": 0.0003936570675701111,
+      "loss": 0.7233,
       "step": 5650
     },
     {
       "epoch": 1.0653115000941087,
-      "grad_norm": 16.53011703491211,
-      "learning_rate": 2.3608130999435347e-05,
-      "loss": 1.0546,
+      "grad_norm": 1.665967583656311,
+      "learning_rate": 0.00039346884999058915,
+      "loss": 0.7456,
       "step": 5660
     },
     {
       "epoch": 1.0671936758893281,
-      "grad_norm": 8.095433235168457,
-      "learning_rate": 2.3596837944664034e-05,
-      "loss": 0.9267,
+      "grad_norm": 2.409528970718384,
+      "learning_rate": 0.00039328063241106723,
+      "loss": 0.6585,
       "step": 5670
     },
     {
       "epoch": 1.0690758516845473,
-      "grad_norm": 29.615936279296875,
-      "learning_rate": 2.3585544889892716e-05,
-      "loss": 1.2167,
+      "grad_norm": 4.774532794952393,
+      "learning_rate": 0.0003930924148315453,
+      "loss": 1.0393,
       "step": 5680
     },
     {
       "epoch": 1.0709580274797665,
-      "grad_norm": 2.900878667831421,
-      "learning_rate": 2.3574251835121402e-05,
-      "loss": 0.9583,
+      "grad_norm": 6.189159393310547,
+      "learning_rate": 0.0003929041972520234,
+      "loss": 0.7568,
       "step": 5690
     },
     {
       "epoch": 1.072840203274986,
-      "grad_norm": 2.4044950008392334,
-      "learning_rate": 2.356295878035009e-05,
-      "loss": 0.8682,
+      "grad_norm": 0.9074216485023499,
+      "learning_rate": 0.00039271597967250146,
+      "loss": 0.7797,
       "step": 5700
     },
     {
       "epoch": 1.0747223790702052,
-      "grad_norm": 3.0829226970672607,
-      "learning_rate": 2.3551665725578768e-05,
-      "loss": 0.9567,
+      "grad_norm": 1.9530513286590576,
+      "learning_rate": 0.0003925277620929795,
+      "loss": 0.7868,
       "step": 5710
     },
     {
       "epoch": 1.0766045548654244,
-      "grad_norm": 8.497003555297852,
-      "learning_rate": 2.3540372670807454e-05,
-      "loss": 0.889,
+      "grad_norm": 4.772610187530518,
+      "learning_rate": 0.00039233954451345756,
+      "loss": 0.5664,
       "step": 5720
     },
     {
       "epoch": 1.0784867306606438,
-      "grad_norm": 5.439596652984619,
-      "learning_rate": 2.3529079616036136e-05,
-      "loss": 1.1877,
+      "grad_norm": 1.7776116132736206,
+      "learning_rate": 0.00039215132693393563,
+      "loss": 1.1096,
       "step": 5730
     },
     {
       "epoch": 1.080368906455863,
-      "grad_norm": 13.089673042297363,
-      "learning_rate": 2.3517786561264823e-05,
-      "loss": 0.9833,
+      "grad_norm": 4.608347415924072,
+      "learning_rate": 0.0003919631093544137,
+      "loss": 0.779,
       "step": 5740
     },
     {
       "epoch": 1.0822510822510822,
-      "grad_norm": 8.794281959533691,
-      "learning_rate": 2.350649350649351e-05,
-      "loss": 1.4184,
+      "grad_norm": 3.2807509899139404,
+      "learning_rate": 0.0003917748917748918,
+      "loss": 1.134,
       "step": 5750
     },
     {
       "epoch": 1.0841332580463015,
-      "grad_norm": 11.838709831237793,
-      "learning_rate": 2.349520045172219e-05,
-      "loss": 1.2191,
+      "grad_norm": 2.146608591079712,
+      "learning_rate": 0.00039158667419536986,
+      "loss": 0.9571,
       "step": 5760
     },
     {
       "epoch": 1.0860154338415209,
-      "grad_norm": 12.768487930297852,
-      "learning_rate": 2.3483907396950877e-05,
-      "loss": 1.0469,
+      "grad_norm": 4.23948335647583,
+      "learning_rate": 0.00039139845661584794,
+      "loss": 0.8992,
       "step": 5770
     },
     {
       "epoch": 1.08789760963674,
-      "grad_norm": 16.489809036254883,
-      "learning_rate": 2.347261434217956e-05,
-      "loss": 1.2401,
+      "grad_norm": 4.417486667633057,
+      "learning_rate": 0.000391210239036326,
+      "loss": 0.9175,
       "step": 5780
     },
     {
       "epoch": 1.0897797854319593,
-      "grad_norm": 22.284759521484375,
-      "learning_rate": 2.3461321287408246e-05,
-      "loss": 1.1759,
+      "grad_norm": 3.7243125438690186,
+      "learning_rate": 0.0003910220214568041,
+      "loss": 0.8366,
       "step": 5790
     },
     {
       "epoch": 1.0916619612271785,
-      "grad_norm": 22.32616424560547,
-      "learning_rate": 2.345002823263693e-05,
-      "loss": 1.0558,
+      "grad_norm": 1.57295823097229,
+      "learning_rate": 0.0003908338038772821,
+      "loss": 0.8425,
       "step": 5800
     },
     {
       "epoch": 1.093544137022398,
-      "grad_norm": 6.583852767944336,
-      "learning_rate": 2.343873517786561e-05,
-      "loss": 1.1139,
+      "grad_norm": 2.6780879497528076,
+      "learning_rate": 0.0003906455862977602,
+      "loss": 0.8754,
       "step": 5810
     },
     {
       "epoch": 1.0954263128176172,
-      "grad_norm": 10.06447982788086,
-      "learning_rate": 2.3427442123094298e-05,
-      "loss": 1.0589,
+      "grad_norm": 0.9460309147834778,
+      "learning_rate": 0.00039045736871823827,
+      "loss": 0.5829,
       "step": 5820
     },
     {
       "epoch": 1.0973084886128364,
-      "grad_norm": 9.685362815856934,
-      "learning_rate": 2.341614906832298e-05,
-      "loss": 1.0719,
+      "grad_norm": 3.544323205947876,
+      "learning_rate": 0.00039026915113871634,
+      "loss": 0.7961,
       "step": 5830
     },
     {
       "epoch": 1.0991906644080558,
-      "grad_norm": 10.641950607299805,
-      "learning_rate": 2.3404856013551666e-05,
-      "loss": 0.8788,
+      "grad_norm": 4.158580303192139,
+      "learning_rate": 0.0003900809335591944,
+      "loss": 0.718,
       "step": 5840
     },
     {
       "epoch": 1.101072840203275,
-      "grad_norm": 9.749815940856934,
-      "learning_rate": 2.3393562958780352e-05,
-      "loss": 0.9172,
+      "grad_norm": 2.3165791034698486,
+      "learning_rate": 0.0003898927159796725,
+      "loss": 0.5707,
       "step": 5850
     },
     {
       "epoch": 1.1029550159984942,
-      "grad_norm": 15.738201141357422,
-      "learning_rate": 2.3382269904009035e-05,
-      "loss": 1.1344,
+      "grad_norm": 2.589895248413086,
+      "learning_rate": 0.00038970449840015057,
+      "loss": 0.9552,
       "step": 5860
     },
     {
       "epoch": 1.1048371917937136,
-      "grad_norm": 5.296823024749756,
-      "learning_rate": 2.337097684923772e-05,
-      "loss": 0.9969,
+      "grad_norm": 2.370169162750244,
+      "learning_rate": 0.0003895162808206287,
+      "loss": 0.7267,
       "step": 5870
     },
     {
       "epoch": 1.1067193675889329,
-      "grad_norm": 9.7347993850708,
-      "learning_rate": 2.33596837944664e-05,
-      "loss": 1.2592,
+      "grad_norm": 0.8681706786155701,
+      "learning_rate": 0.0003893280632411067,
+      "loss": 1.0151,
       "step": 5880
     },
     {
       "epoch": 1.108601543384152,
-      "grad_norm": 15.322550773620605,
-      "learning_rate": 2.3348390739695087e-05,
-      "loss": 0.8988,
+      "grad_norm": 0.5660446286201477,
+      "learning_rate": 0.0003891398456615848,
+      "loss": 0.6168,
       "step": 5890
     },
     {
       "epoch": 1.1104837191793713,
-      "grad_norm": 8.814277648925781,
-      "learning_rate": 2.3337097684923773e-05,
-      "loss": 0.9639,
+      "grad_norm": 2.0442614555358887,
+      "learning_rate": 0.0003889516280820629,
+      "loss": 0.603,
       "step": 5900
     },
     {
       "epoch": 1.1123658949745907,
-      "grad_norm": 9.318452835083008,
-      "learning_rate": 2.3325804630152455e-05,
-      "loss": 0.9157,
+      "grad_norm": 2.1244430541992188,
+      "learning_rate": 0.00038876341050254095,
+      "loss": 0.5466,
       "step": 5910
     },
     {
       "epoch": 1.11424807076981,
-      "grad_norm": 9.616998672485352,
-      "learning_rate": 2.331451157538114e-05,
-      "loss": 0.9965,
+      "grad_norm": 4.894069671630859,
+      "learning_rate": 0.00038857519292301903,
+      "loss": 0.7559,
       "step": 5920
     },
     {
       "epoch": 1.1161302465650291,
-      "grad_norm": 20.399810791015625,
-      "learning_rate": 2.3303218520609827e-05,
-      "loss": 0.9574,
+      "grad_norm": 4.432802200317383,
+      "learning_rate": 0.0003883869753434971,
+      "loss": 0.7947,
       "step": 5930
     },
     {
       "epoch": 1.1180124223602483,
-      "grad_norm": 25.457443237304688,
-      "learning_rate": 2.329192546583851e-05,
-      "loss": 0.948,
+      "grad_norm": 3.1599254608154297,
+      "learning_rate": 0.0003881987577639752,
+      "loss": 0.6203,
       "step": 5940
     },
     {
       "epoch": 1.1198945981554678,
-      "grad_norm": 10.880556106567383,
-      "learning_rate": 2.3280632411067196e-05,
-      "loss": 1.1425,
+      "grad_norm": 1.5855923891067505,
+      "learning_rate": 0.00038801054018445326,
+      "loss": 0.9732,
       "step": 5950
     },
     {
       "epoch": 1.121776773950687,
-      "grad_norm": 26.9040584564209,
-      "learning_rate": 2.326933935629588e-05,
-      "loss": 1.2454,
+      "grad_norm": 4.721362590789795,
+      "learning_rate": 0.00038782232260493134,
+      "loss": 1.12,
       "step": 5960
     },
     {
       "epoch": 1.1236589497459062,
-      "grad_norm": 12.784603118896484,
-      "learning_rate": 2.325804630152456e-05,
-      "loss": 1.0512,
+      "grad_norm": 2.280820846557617,
+      "learning_rate": 0.00038763410502540936,
+      "loss": 0.7846,
       "step": 5970
     },
     {
       "epoch": 1.1255411255411256,
-      "grad_norm": 6.434301853179932,
-      "learning_rate": 2.3246753246753248e-05,
-      "loss": 0.8505,
+      "grad_norm": 1.847952961921692,
+      "learning_rate": 0.00038744588744588743,
+      "loss": 0.6701,
       "step": 5980
     },
     {
       "epoch": 1.1274233013363448,
-      "grad_norm": 12.109904289245605,
-      "learning_rate": 2.323546019198193e-05,
-      "loss": 1.0467,
+      "grad_norm": 6.386806964874268,
+      "learning_rate": 0.0003872576698663655,
+      "loss": 0.8316,
       "step": 5990
     },
     {
       "epoch": 1.129305477131564,
-      "grad_norm": 11.11652660369873,
-      "learning_rate": 2.3224167137210616e-05,
-      "loss": 0.7246,
+      "grad_norm": 3.615067481994629,
+      "learning_rate": 0.0003870694522868436,
+      "loss": 0.6289,
       "step": 6000
     },
     {
       "epoch": 1.1311876529267835,
-      "grad_norm": 15.59671401977539,
-      "learning_rate": 2.32128740824393e-05,
-      "loss": 0.9292,
+      "grad_norm": 2.8482844829559326,
+      "learning_rate": 0.00038688123470732166,
+      "loss": 0.7003,
       "step": 6010
     },
     {
       "epoch": 1.1330698287220027,
-      "grad_norm": 6.132266521453857,
-      "learning_rate": 2.3201581027667985e-05,
-      "loss": 0.721,
+      "grad_norm": 1.1809179782867432,
+      "learning_rate": 0.00038669301712779974,
+      "loss": 0.5173,
       "step": 6020
     },
     {
       "epoch": 1.134952004517222,
-      "grad_norm": 3.5677640438079834,
-      "learning_rate": 2.319028797289667e-05,
-      "loss": 0.9246,
+      "grad_norm": 2.176791191101074,
+      "learning_rate": 0.0003865047995482778,
+      "loss": 0.7291,
       "step": 6030
     },
     {
       "epoch": 1.136834180312441,
-      "grad_norm": 15.319580078125,
-      "learning_rate": 2.3178994918125354e-05,
-      "loss": 0.9668,
+      "grad_norm": 2.8241522312164307,
+      "learning_rate": 0.0003863165819687559,
+      "loss": 0.6527,
       "step": 6040
     },
     {
       "epoch": 1.1387163561076605,
-      "grad_norm": 15.318434715270996,
-      "learning_rate": 2.316770186335404e-05,
-      "loss": 1.0329,
+      "grad_norm": 3.1613717079162598,
+      "learning_rate": 0.00038612836438923397,
+      "loss": 0.639,
       "step": 6050
     },
     {
       "epoch": 1.1405985319028797,
-      "grad_norm": 8.09421157836914,
-      "learning_rate": 2.315640880858272e-05,
-      "loss": 1.0839,
+      "grad_norm": 2.8154726028442383,
+      "learning_rate": 0.000385940146809712,
+      "loss": 0.7908,
       "step": 6060
     },
     {
       "epoch": 1.142480707698099,
-      "grad_norm": 33.64767837524414,
-      "learning_rate": 2.3145115753811405e-05,
-      "loss": 1.1178,
+      "grad_norm": 3.6952784061431885,
+      "learning_rate": 0.00038575192923019007,
+      "loss": 0.826,
       "step": 6070
     },
     {
       "epoch": 1.1443628834933182,
-      "grad_norm": 23.52187728881836,
-      "learning_rate": 2.313382269904009e-05,
-      "loss": 1.1034,
+      "grad_norm": 3.8638036251068115,
+      "learning_rate": 0.00038556371165066814,
+      "loss": 1.0331,
       "step": 6080
     },
     {
       "epoch": 1.1462450592885376,
-      "grad_norm": 13.548325538635254,
-      "learning_rate": 2.3122529644268774e-05,
-      "loss": 0.801,
+      "grad_norm": 3.39660906791687,
+      "learning_rate": 0.0003853754940711463,
+      "loss": 0.4696,
       "step": 6090
     },
     {
       "epoch": 1.1481272350837568,
-      "grad_norm": 19.462448120117188,
-      "learning_rate": 2.311123658949746e-05,
-      "loss": 0.8789,
+      "grad_norm": 3.148796319961548,
+      "learning_rate": 0.00038518727649162435,
+      "loss": 0.5888,
       "step": 6100
     },
     {
       "epoch": 1.150009410878976,
-      "grad_norm": 32.65730285644531,
-      "learning_rate": 2.3099943534726143e-05,
-      "loss": 1.2276,
+      "grad_norm": 1.6677438020706177,
+      "learning_rate": 0.00038499905891210243,
+      "loss": 1.0059,
       "step": 6110
     },
     {
       "epoch": 1.1518915866741954,
-      "grad_norm": 1.7889130115509033,
-      "learning_rate": 2.308865047995483e-05,
-      "loss": 0.9995,
+      "grad_norm": 0.408934623003006,
+      "learning_rate": 0.0003848108413325805,
+      "loss": 0.7653,
       "step": 6120
     },
     {
       "epoch": 1.1537737624694147,
-      "grad_norm": 2.802178382873535,
-      "learning_rate": 2.3077357425183515e-05,
-      "loss": 0.9639,
+      "grad_norm": 0.9431570172309875,
+      "learning_rate": 0.0003846226237530586,
+      "loss": 0.8325,
       "step": 6130
     },
     {
       "epoch": 1.1556559382646339,
-      "grad_norm": 30.473918914794922,
-      "learning_rate": 2.3066064370412198e-05,
-      "loss": 0.9189,
+      "grad_norm": 3.6121902465820312,
+      "learning_rate": 0.00038443440617353666,
+      "loss": 0.6998,
       "step": 6140
     },
     {
       "epoch": 1.1575381140598533,
-      "grad_norm": 12.530424118041992,
-      "learning_rate": 2.305477131564088e-05,
-      "loss": 1.1765,
+      "grad_norm": 4.1678948402404785,
+      "learning_rate": 0.0003842461885940147,
+      "loss": 0.9246,
       "step": 6150
     },
     {
       "epoch": 1.1594202898550725,
-      "grad_norm": 20.285593032836914,
-      "learning_rate": 2.3043478260869563e-05,
-      "loss": 1.0729,
+      "grad_norm": 2.139291763305664,
+      "learning_rate": 0.00038405797101449276,
+      "loss": 0.7167,
       "step": 6160
     },
     {
       "epoch": 1.1613024656502917,
-      "grad_norm": 17.06644058227539,
-      "learning_rate": 2.303218520609825e-05,
-      "loss": 1.1532,
+      "grad_norm": 3.6605629920959473,
+      "learning_rate": 0.00038386975343497083,
+      "loss": 0.9195,
       "step": 6170
     },
     {
       "epoch": 1.163184641445511,
-      "grad_norm": 13.454843521118164,
-      "learning_rate": 2.3020892151326935e-05,
-      "loss": 1.0656,
+      "grad_norm": 0.30113252997398376,
+      "learning_rate": 0.0003836815358554489,
+      "loss": 0.7755,
       "step": 6180
     },
     {
       "epoch": 1.1650668172407304,
-      "grad_norm": 14.98215103149414,
-      "learning_rate": 2.3009599096555618e-05,
-      "loss": 0.7957,
+      "grad_norm": 3.287532091140747,
+      "learning_rate": 0.000383493318275927,
+      "loss": 0.6074,
       "step": 6190
     },
     {
       "epoch": 1.1669489930359496,
-      "grad_norm": 12.102943420410156,
-      "learning_rate": 2.2998306041784304e-05,
-      "loss": 1.3556,
+      "grad_norm": 5.199028491973877,
+      "learning_rate": 0.00038330510069640506,
+      "loss": 1.1058,
       "step": 6200
     },
     {
       "epoch": 1.1688311688311688,
-      "grad_norm": 10.94884204864502,
-      "learning_rate": 2.298701298701299e-05,
-      "loss": 1.0466,
+      "grad_norm": 3.132143259048462,
+      "learning_rate": 0.00038311688311688314,
+      "loss": 0.8156,
       "step": 6210
     },
     {
       "epoch": 1.170713344626388,
-      "grad_norm": 5.2002177238464355,
-      "learning_rate": 2.2975719932241673e-05,
-      "loss": 1.1543,
+      "grad_norm": 4.496564865112305,
+      "learning_rate": 0.0003829286655373612,
+      "loss": 0.78,
       "step": 6220
     },
     {
       "epoch": 1.1725955204216074,
-      "grad_norm": 8.19458293914795,
-      "learning_rate": 2.2964426877470356e-05,
-      "loss": 0.7794,
+      "grad_norm": 2.972040891647339,
+      "learning_rate": 0.00038274044795783924,
+      "loss": 0.6054,
       "step": 6230
     },
     {
       "epoch": 1.1744776962168266,
-      "grad_norm": 7.3538970947265625,
-      "learning_rate": 2.2953133822699038e-05,
-      "loss": 0.7533,
+      "grad_norm": 3.018740177154541,
+      "learning_rate": 0.0003825522303783173,
+      "loss": 0.596,
       "step": 6240
     },
     {
       "epoch": 1.1763598720120458,
-      "grad_norm": 13.170063018798828,
-      "learning_rate": 2.2941840767927724e-05,
-      "loss": 1.1663,
+      "grad_norm": 4.458706378936768,
+      "learning_rate": 0.0003823640127987954,
+      "loss": 1.0045,
       "step": 6250
     },
     {
       "epoch": 1.1782420478072653,
-      "grad_norm": 22.807252883911133,
-      "learning_rate": 2.293054771315641e-05,
-      "loss": 0.9954,
+      "grad_norm": 2.131452798843384,
+      "learning_rate": 0.00038217579521927347,
+      "loss": 0.7716,
       "step": 6260
     },
     {
       "epoch": 1.1801242236024845,
-      "grad_norm": 4.545765399932861,
-      "learning_rate": 2.2919254658385093e-05,
-      "loss": 0.9198,
+      "grad_norm": 1.9051343202590942,
+      "learning_rate": 0.00038198757763975154,
+      "loss": 0.6796,
       "step": 6270
     },
     {
       "epoch": 1.1820063993977037,
-      "grad_norm": 8.624770164489746,
-      "learning_rate": 2.290796160361378e-05,
-      "loss": 0.9429,
+      "grad_norm": 2.997279644012451,
+      "learning_rate": 0.0003817993600602296,
+      "loss": 0.6617,
       "step": 6280
     },
     {
       "epoch": 1.1838885751929231,
-      "grad_norm": 12.10987377166748,
-      "learning_rate": 2.2896668548842462e-05,
-      "loss": 1.2487,
+      "grad_norm": 2.8390822410583496,
+      "learning_rate": 0.0003816111424807077,
+      "loss": 1.022,
       "step": 6290
     },
     {
       "epoch": 1.1857707509881423,
-      "grad_norm": 1.273734211921692,
-      "learning_rate": 2.2885375494071148e-05,
-      "loss": 0.9025,
+      "grad_norm": 0.6079533696174622,
+      "learning_rate": 0.00038142292490118577,
+      "loss": 0.6297,
       "step": 6300
     },
     {
       "epoch": 1.1876529267833615,
-      "grad_norm": 6.115864276885986,
-      "learning_rate": 2.2874082439299834e-05,
-      "loss": 0.9436,
+      "grad_norm": 2.57958984375,
+      "learning_rate": 0.0003812347073216639,
+      "loss": 0.701,
       "step": 6310
     },
     {
       "epoch": 1.1895351025785807,
-      "grad_norm": 6.128392696380615,
-      "learning_rate": 2.2862789384528513e-05,
-      "loss": 0.9464,
+      "grad_norm": 3.771247625350952,
+      "learning_rate": 0.0003810464897421419,
+      "loss": 0.7474,
       "step": 6320
     },
     {
       "epoch": 1.1914172783738002,
-      "grad_norm": 13.86414623260498,
-      "learning_rate": 2.28514963297572e-05,
-      "loss": 0.8451,
+      "grad_norm": 2.2047035694122314,
+      "learning_rate": 0.00038085827216262,
+      "loss": 0.5503,
       "step": 6330
     },
     {
       "epoch": 1.1932994541690194,
-      "grad_norm": 19.781795501708984,
-      "learning_rate": 2.2840203274985882e-05,
-      "loss": 0.7738,
+      "grad_norm": 4.219132900238037,
+      "learning_rate": 0.0003806700545830981,
+      "loss": 0.6213,
       "step": 6340
     },
     {
       "epoch": 1.1951816299642386,
-      "grad_norm": 30.369861602783203,
-      "learning_rate": 2.2828910220214568e-05,
-      "loss": 1.2183,
+      "grad_norm": 3.2710931301116943,
+      "learning_rate": 0.00038048183700357615,
+      "loss": 0.6869,
       "step": 6350
     },
     {
       "epoch": 1.1970638057594578,
-      "grad_norm": 12.43057918548584,
-      "learning_rate": 2.2817617165443254e-05,
-      "loss": 1.1899,
+      "grad_norm": 2.1299989223480225,
+      "learning_rate": 0.00038029361942405423,
+      "loss": 1.0877,
       "step": 6360
     },
     {
       "epoch": 1.1989459815546772,
-      "grad_norm": 9.697637557983398,
-      "learning_rate": 2.2806324110671937e-05,
-      "loss": 1.2109,
+      "grad_norm": 2.7965474128723145,
+      "learning_rate": 0.0003801054018445323,
+      "loss": 0.9136,
       "step": 6370
     },
     {
       "epoch": 1.2008281573498965,
-      "grad_norm": 33.04664993286133,
-      "learning_rate": 2.2795031055900623e-05,
-      "loss": 1.2716,
+      "grad_norm": 4.0959696769714355,
+      "learning_rate": 0.0003799171842650104,
+      "loss": 1.0894,
       "step": 6380
     },
     {
       "epoch": 1.2027103331451157,
-      "grad_norm": 16.485483169555664,
-      "learning_rate": 2.2783738001129306e-05,
-      "loss": 1.1587,
+      "grad_norm": 3.3608808517456055,
+      "learning_rate": 0.00037972896668548846,
+      "loss": 0.7197,
       "step": 6390
     },
     {
       "epoch": 1.204592508940335,
-      "grad_norm": 8.360411643981934,
-      "learning_rate": 2.2772444946357992e-05,
-      "loss": 1.0419,
+      "grad_norm": 2.8144304752349854,
+      "learning_rate": 0.00037954074910596654,
+      "loss": 0.7231,
       "step": 6400
     },
     {
       "epoch": 1.2064746847355543,
-      "grad_norm": 3.943709373474121,
-      "learning_rate": 2.2761151891586674e-05,
-      "loss": 1.0027,
+      "grad_norm": 0.6063675880432129,
+      "learning_rate": 0.00037935253152644456,
+      "loss": 0.6725,
       "step": 6410
     },
     {
       "epoch": 1.2083568605307735,
-      "grad_norm": 17.139930725097656,
-      "learning_rate": 2.2749858836815357e-05,
-      "loss": 1.1939,
+      "grad_norm": 1.4623092412948608,
+      "learning_rate": 0.00037916431394692263,
+      "loss": 1.0756,
       "step": 6420
     },
     {
       "epoch": 1.210239036325993,
-      "grad_norm": 14.557121276855469,
-      "learning_rate": 2.2738565782044043e-05,
-      "loss": 0.8377,
+      "grad_norm": 1.7580760717391968,
+      "learning_rate": 0.0003789760963674007,
+      "loss": 0.6686,
       "step": 6430
     },
     {
       "epoch": 1.2121212121212122,
-      "grad_norm": 17.73250961303711,
-      "learning_rate": 2.272727272727273e-05,
-      "loss": 0.9496,
+      "grad_norm": 8.147522926330566,
+      "learning_rate": 0.0003787878787878788,
+      "loss": 0.7119,
       "step": 6440
     },
     {
       "epoch": 1.2140033879164314,
-      "grad_norm": 15.301102638244629,
-      "learning_rate": 2.2715979672501412e-05,
-      "loss": 1.0668,
+      "grad_norm": 3.453514337539673,
+      "learning_rate": 0.00037859966120835686,
+      "loss": 0.8779,
       "step": 6450
     },
     {
       "epoch": 1.2158855637116506,
-      "grad_norm": 15.799238204956055,
-      "learning_rate": 2.2704686617730098e-05,
-      "loss": 1.0151,
+      "grad_norm": 2.9244697093963623,
+      "learning_rate": 0.00037841144362883494,
+      "loss": 0.7888,
       "step": 6460
     },
     {
       "epoch": 1.21776773950687,
-      "grad_norm": 20.844924926757812,
-      "learning_rate": 2.269339356295878e-05,
-      "loss": 0.9301,
+      "grad_norm": 4.729470729827881,
+      "learning_rate": 0.000378223226049313,
+      "loss": 0.6608,
       "step": 6470
     },
     {
       "epoch": 1.2196499153020892,
-      "grad_norm": 7.686192512512207,
-      "learning_rate": 2.2682100508187467e-05,
-      "loss": 0.797,
+      "grad_norm": 2.5494601726531982,
+      "learning_rate": 0.0003780350084697911,
+      "loss": 0.6861,
       "step": 6480
     },
     {
       "epoch": 1.2215320910973084,
-      "grad_norm": 15.735981941223145,
-      "learning_rate": 2.267080745341615e-05,
-      "loss": 1.0817,
+      "grad_norm": 4.567807674407959,
+      "learning_rate": 0.0003778467908902691,
+      "loss": 0.8115,
       "step": 6490
     },
     {
       "epoch": 1.2234142668925279,
-      "grad_norm": 10.269287109375,
-      "learning_rate": 2.2659514398644832e-05,
-      "loss": 1.1608,
+      "grad_norm": 2.6255970001220703,
+      "learning_rate": 0.0003776585733107472,
+      "loss": 0.976,
       "step": 6500
     },
     {
       "epoch": 1.225296442687747,
-      "grad_norm": 28.582277297973633,
-      "learning_rate": 2.2648221343873518e-05,
-      "loss": 0.8283,
+      "grad_norm": 2.7389473915100098,
+      "learning_rate": 0.00037747035573122527,
+      "loss": 0.6704,
       "step": 6510
     },
     {
       "epoch": 1.2271786184829663,
-      "grad_norm": 8.514233589172363,
-      "learning_rate": 2.26369282891022e-05,
-      "loss": 1.2402,
+      "grad_norm": 2.68717360496521,
+      "learning_rate": 0.00037728213815170334,
+      "loss": 0.9755,
       "step": 6520
     },
     {
       "epoch": 1.2290607942781855,
-      "grad_norm": 9.531684875488281,
-      "learning_rate": 2.2625635234330887e-05,
-      "loss": 1.1151,
+      "grad_norm": 1.1812397241592407,
+      "learning_rate": 0.0003770939205721814,
+      "loss": 0.9663,
       "step": 6530
     },
     {
       "epoch": 1.230942970073405,
-      "grad_norm": 10.337912559509277,
-      "learning_rate": 2.2614342179559573e-05,
-      "loss": 0.8757,
+      "grad_norm": 2.752870798110962,
+      "learning_rate": 0.00037690570299265955,
+      "loss": 0.6739,
       "step": 6540
     },
     {
       "epoch": 1.2328251458686241,
-      "grad_norm": 8.955284118652344,
-      "learning_rate": 2.2603049124788256e-05,
-      "loss": 0.76,
+      "grad_norm": 3.182645797729492,
+      "learning_rate": 0.00037671748541313763,
+      "loss": 0.5563,
       "step": 6550
     },
     {
       "epoch": 1.2347073216638433,
-      "grad_norm": 14.72435474395752,
-      "learning_rate": 2.2591756070016942e-05,
-      "loss": 0.9185,
+      "grad_norm": 3.5560567378997803,
+      "learning_rate": 0.0003765292678336157,
+      "loss": 0.9168,
       "step": 6560
     },
     {
       "epoch": 1.2365894974590628,
-      "grad_norm": 16.09954071044922,
-      "learning_rate": 2.2580463015245625e-05,
-      "loss": 1.313,
+      "grad_norm": 3.1835429668426514,
+      "learning_rate": 0.0003763410502540938,
+      "loss": 1.2578,
       "step": 6570
     },
     {
       "epoch": 1.238471673254282,
-      "grad_norm": 8.912398338317871,
-      "learning_rate": 2.2569169960474307e-05,
-      "loss": 0.8678,
+      "grad_norm": 3.730724334716797,
+      "learning_rate": 0.0003761528326745718,
+      "loss": 0.5987,
       "step": 6580
     },
     {
       "epoch": 1.2403538490495012,
-      "grad_norm": 8.538116455078125,
-      "learning_rate": 2.2557876905702993e-05,
-      "loss": 1.0072,
+      "grad_norm": 1.5414713621139526,
+      "learning_rate": 0.0003759646150950499,
+      "loss": 0.7506,
       "step": 6590
     },
     {
       "epoch": 1.2422360248447206,
-      "grad_norm": 14.951505661010742,
-      "learning_rate": 2.2546583850931676e-05,
-      "loss": 0.6922,
+      "grad_norm": 1.6156970262527466,
+      "learning_rate": 0.00037577639751552796,
+      "loss": 0.4276,
       "step": 6600
     },
     {
       "epoch": 1.2441182006399398,
-      "grad_norm": 8.427967071533203,
-      "learning_rate": 2.2535290796160362e-05,
-      "loss": 0.6708,
+      "grad_norm": 1.9981142282485962,
+      "learning_rate": 0.00037558817993600603,
+      "loss": 0.6881,
       "step": 6610
     },
     {
       "epoch": 1.246000376435159,
-      "grad_norm": 9.497509002685547,
-      "learning_rate": 2.2523997741389045e-05,
-      "loss": 1.0899,
+      "grad_norm": 1.740738034248352,
+      "learning_rate": 0.0003753999623564841,
+      "loss": 1.0043,
       "step": 6620
     },
     {
       "epoch": 1.2478825522303783,
-      "grad_norm": 30.798381805419922,
-      "learning_rate": 2.251270468661773e-05,
-      "loss": 0.9245,
+      "grad_norm": 1.780943512916565,
+      "learning_rate": 0.0003752117447769622,
+      "loss": 0.5519,
       "step": 6630
     },
     {
       "epoch": 1.2497647280255977,
-      "grad_norm": 12.764777183532715,
-      "learning_rate": 2.2501411631846417e-05,
-      "loss": 0.8884,
+      "grad_norm": 5.119179725646973,
+      "learning_rate": 0.00037502352719744026,
+      "loss": 0.9018,
       "step": 6640
     },
     {
       "epoch": 1.251646903820817,
-      "grad_norm": 6.89249324798584,
-      "learning_rate": 2.24901185770751e-05,
-      "loss": 0.9125,
+      "grad_norm": 2.3510608673095703,
+      "learning_rate": 0.00037483530961791834,
+      "loss": 0.6956,
       "step": 6650
     },
     {
       "epoch": 1.253529079616036,
-      "grad_norm": 9.787991523742676,
-      "learning_rate": 2.2478825522303786e-05,
-      "loss": 0.8764,
+      "grad_norm": 4.710165500640869,
+      "learning_rate": 0.0003746470920383964,
+      "loss": 0.5775,
       "step": 6660
     },
     {
       "epoch": 1.2554112554112553,
-      "grad_norm": 20.8765811920166,
-      "learning_rate": 2.246753246753247e-05,
-      "loss": 1.1334,
+      "grad_norm": 4.0059051513671875,
+      "learning_rate": 0.00037445887445887444,
+      "loss": 0.7891,
       "step": 6670
     },
     {
       "epoch": 1.2572934312064747,
-      "grad_norm": 19.523963928222656,
-      "learning_rate": 2.245623941276115e-05,
-      "loss": 1.2794,
+      "grad_norm": 3.0449984073638916,
+      "learning_rate": 0.0003742706568793525,
+      "loss": 1.1969,
       "step": 6680
     },
     {
       "epoch": 1.259175607001694,
-      "grad_norm": 17.773313522338867,
-      "learning_rate": 2.2444946357989837e-05,
-      "loss": 0.8465,
+      "grad_norm": 2.8127171993255615,
+      "learning_rate": 0.0003740824392998306,
+      "loss": 0.6317,
       "step": 6690
     },
     {
       "epoch": 1.2610577827969132,
-      "grad_norm": 11.96570110321045,
-      "learning_rate": 2.243365330321852e-05,
-      "loss": 1.1552,
+      "grad_norm": 3.5369529724121094,
+      "learning_rate": 0.00037389422172030867,
+      "loss": 1.0789,
       "step": 6700
     },
     {
       "epoch": 1.2629399585921326,
-      "grad_norm": 8.867982864379883,
-      "learning_rate": 2.2422360248447206e-05,
-      "loss": 1.011,
+      "grad_norm": 0.9900074005126953,
+      "learning_rate": 0.00037370600414078674,
+      "loss": 0.7734,
       "step": 6710
     },
     {
       "epoch": 1.2648221343873518,
-      "grad_norm": 29.349536895751953,
-      "learning_rate": 2.2411067193675892e-05,
-      "loss": 0.8635,
+      "grad_norm": 5.281899452209473,
+      "learning_rate": 0.0003735177865612648,
+      "loss": 0.6816,
       "step": 6720
     },
     {
       "epoch": 1.266704310182571,
-      "grad_norm": 12.906034469604492,
-      "learning_rate": 2.2399774138904575e-05,
-      "loss": 0.8193,
+      "grad_norm": 3.5994303226470947,
+      "learning_rate": 0.0003733295689817429,
+      "loss": 0.7046,
       "step": 6730
     },
     {
       "epoch": 1.2685864859777904,
-      "grad_norm": 12.88680648803711,
-      "learning_rate": 2.238848108413326e-05,
-      "loss": 1.264,
+      "grad_norm": 3.552111864089966,
+      "learning_rate": 0.00037314135140222097,
+      "loss": 0.9432,
       "step": 6740
     },
     {
       "epoch": 1.2704686617730097,
-      "grad_norm": 5.127486228942871,
-      "learning_rate": 2.2377188029361943e-05,
-      "loss": 0.839,
+      "grad_norm": 1.8412734270095825,
+      "learning_rate": 0.0003729531338226991,
+      "loss": 0.7193,
       "step": 6750
     },
     {
       "epoch": 1.2723508375682289,
-      "grad_norm": 5.232704162597656,
-      "learning_rate": 2.2365894974590626e-05,
-      "loss": 0.9379,
+      "grad_norm": 2.4162864685058594,
+      "learning_rate": 0.0003727649162431771,
+      "loss": 0.7516,
       "step": 6760
     },
     {
       "epoch": 1.274233013363448,
-      "grad_norm": 38.552825927734375,
-      "learning_rate": 2.2354601919819312e-05,
-      "loss": 0.9256,
+      "grad_norm": 3.4946014881134033,
+      "learning_rate": 0.0003725766986636552,
+      "loss": 0.8546,
       "step": 6770
     },
     {
       "epoch": 1.2761151891586673,
-      "grad_norm": 27.74937629699707,
-      "learning_rate": 2.2343308865047995e-05,
-      "loss": 0.9923,
+      "grad_norm": 2.2517666816711426,
+      "learning_rate": 0.0003723884810841333,
+      "loss": 0.8215,
       "step": 6780
     },
     {
       "epoch": 1.2779973649538867,
-      "grad_norm": 26.892642974853516,
-      "learning_rate": 2.233201581027668e-05,
-      "loss": 1.0955,
+      "grad_norm": 4.775296688079834,
+      "learning_rate": 0.00037220026350461135,
+      "loss": 0.8108,
       "step": 6790
     },
     {
       "epoch": 1.279879540749106,
-      "grad_norm": 11.914067268371582,
-      "learning_rate": 2.2320722755505364e-05,
-      "loss": 0.9401,
+      "grad_norm": 2.924154043197632,
+      "learning_rate": 0.00037201204592508943,
+      "loss": 0.6881,
       "step": 6800
     },
     {
       "epoch": 1.2817617165443251,
-      "grad_norm": 16.04292869567871,
-      "learning_rate": 2.230942970073405e-05,
-      "loss": 0.9384,
+      "grad_norm": 2.4805564880371094,
+      "learning_rate": 0.0003718238283455675,
+      "loss": 0.7581,
       "step": 6810
     },
     {
       "epoch": 1.2836438923395446,
-      "grad_norm": 7.591270923614502,
-      "learning_rate": 2.2298136645962736e-05,
-      "loss": 0.8969,
+      "grad_norm": 1.8149421215057373,
+      "learning_rate": 0.0003716356107660456,
+      "loss": 0.7755,
       "step": 6820
     },
     {
       "epoch": 1.2855260681347638,
-      "grad_norm": 23.051708221435547,
-      "learning_rate": 2.228684359119142e-05,
-      "loss": 0.7347,
+      "grad_norm": 2.719374418258667,
+      "learning_rate": 0.00037144739318652366,
+      "loss": 0.5672,
       "step": 6830
     },
     {
       "epoch": 1.287408243929983,
-      "grad_norm": 19.278038024902344,
-      "learning_rate": 2.22755505364201e-05,
-      "loss": 0.873,
+      "grad_norm": 1.9230657815933228,
+      "learning_rate": 0.0003712591756070017,
+      "loss": 0.7672,
       "step": 6840
     },
     {
       "epoch": 1.2892904197252024,
-      "grad_norm": 6.090950012207031,
-      "learning_rate": 2.2264257481648784e-05,
-      "loss": 0.8029,
+      "grad_norm": 1.078181266784668,
+      "learning_rate": 0.00037107095802747976,
+      "loss": 0.4504,
       "step": 6850
     },
     {
       "epoch": 1.2911725955204216,
-      "grad_norm": 7.98006010055542,
-      "learning_rate": 2.225296442687747e-05,
-      "loss": 1.2243,
+      "grad_norm": 2.379368543624878,
+      "learning_rate": 0.00037088274044795783,
+      "loss": 0.9793,
       "step": 6860
     },
     {
       "epoch": 1.2930547713156408,
-      "grad_norm": 10.448835372924805,
-      "learning_rate": 2.2241671372106156e-05,
-      "loss": 1.0374,
+      "grad_norm": 3.081979990005493,
+      "learning_rate": 0.0003706945228684359,
+      "loss": 0.9406,
       "step": 6870
     },
     {
       "epoch": 1.2949369471108603,
-      "grad_norm": 8.826786041259766,
-      "learning_rate": 2.223037831733484e-05,
-      "loss": 0.9484,
+      "grad_norm": 1.935721516609192,
+      "learning_rate": 0.000370506305288914,
+      "loss": 0.6767,
       "step": 6880
     },
     {
       "epoch": 1.2968191229060795,
-      "grad_norm": 7.370842933654785,
-      "learning_rate": 2.2219085262563525e-05,
-      "loss": 1.0126,
+      "grad_norm": 4.712551116943359,
+      "learning_rate": 0.00037031808770939206,
+      "loss": 0.8502,
       "step": 6890
     },
     {
       "epoch": 1.2987012987012987,
-      "grad_norm": 18.180156707763672,
-      "learning_rate": 2.2207792207792207e-05,
-      "loss": 1.2361,
+      "grad_norm": 4.596050262451172,
+      "learning_rate": 0.00037012987012987014,
+      "loss": 0.9557,
       "step": 6900
     },
     {
       "epoch": 1.300583474496518,
-      "grad_norm": 11.186609268188477,
-      "learning_rate": 2.2196499153020894e-05,
-      "loss": 1.0604,
+      "grad_norm": 2.515564203262329,
+      "learning_rate": 0.0003699416525503482,
+      "loss": 0.8905,
       "step": 6910
     },
     {
       "epoch": 1.3024656502917373,
-      "grad_norm": 20.87862777709961,
-      "learning_rate": 2.218520609824958e-05,
-      "loss": 0.8857,
+      "grad_norm": 1.882664442062378,
+      "learning_rate": 0.0003697534349708263,
+      "loss": 0.5869,
       "step": 6920
     },
     {
       "epoch": 1.3043478260869565,
-      "grad_norm": 12.313833236694336,
-      "learning_rate": 2.217391304347826e-05,
-      "loss": 0.9329,
+      "grad_norm": 2.301548719406128,
+      "learning_rate": 0.0003695652173913043,
+      "loss": 0.8646,
       "step": 6930
     },
     {
       "epoch": 1.3062300018821758,
-      "grad_norm": 9.278630256652832,
-      "learning_rate": 2.2162619988706945e-05,
-      "loss": 0.9219,
+      "grad_norm": 1.3468194007873535,
+      "learning_rate": 0.0003693769998117824,
+      "loss": 0.8281,
       "step": 6940
     },
     {
       "epoch": 1.308112177677395,
-      "grad_norm": 14.707159996032715,
-      "learning_rate": 2.215132693393563e-05,
-      "loss": 1.227,
+      "grad_norm": 2.1500511169433594,
+      "learning_rate": 0.00036918878223226047,
+      "loss": 0.9431,
       "step": 6950
     },
     {
       "epoch": 1.3099943534726144,
-      "grad_norm": 31.15422248840332,
-      "learning_rate": 2.2140033879164314e-05,
-      "loss": 0.7853,
+      "grad_norm": 3.4444191455841064,
+      "learning_rate": 0.00036900056465273854,
+      "loss": 0.6617,
       "step": 6960
     },
     {
       "epoch": 1.3118765292678336,
-      "grad_norm": 8.061625480651855,
-      "learning_rate": 2.2128740824393e-05,
-      "loss": 0.9793,
+      "grad_norm": 2.3794643878936768,
+      "learning_rate": 0.0003688123470732166,
+      "loss": 0.7803,
       "step": 6970
     },
     {
       "epoch": 1.3137587050630528,
-      "grad_norm": 9.95386791229248,
-      "learning_rate": 2.2117447769621683e-05,
-      "loss": 1.2524,
+      "grad_norm": 2.2236216068267822,
+      "learning_rate": 0.00036862412949369475,
+      "loss": 1.1628,
       "step": 6980
     },
     {
       "epoch": 1.3156408808582722,
-      "grad_norm": 6.791532516479492,
-      "learning_rate": 2.210615471485037e-05,
-      "loss": 0.8785,
+      "grad_norm": 1.8513524532318115,
+      "learning_rate": 0.00036843591191417283,
+      "loss": 0.7694,
       "step": 6990
     },
     {
       "epoch": 1.3175230566534915,
-      "grad_norm": 7.024257659912109,
-      "learning_rate": 2.2094861660079055e-05,
-      "loss": 0.9782,
+      "grad_norm": 5.277342796325684,
+      "learning_rate": 0.0003682476943346509,
+      "loss": 0.7636,
       "step": 7000
     },
     {
       "epoch": 1.3194052324487107,
-      "grad_norm": 7.559739589691162,
-      "learning_rate": 2.2083568605307737e-05,
-      "loss": 0.9149,
+      "grad_norm": 2.040113925933838,
+      "learning_rate": 0.000368059476755129,
+      "loss": 0.6748,
       "step": 7010
     },
     {
       "epoch": 1.32128740824393,
-      "grad_norm": 8.2528076171875,
-      "learning_rate": 2.207227555053642e-05,
-      "loss": 0.7117,
+      "grad_norm": 4.988844871520996,
+      "learning_rate": 0.000367871259175607,
+      "loss": 0.5714,
       "step": 7020
     },
     {
       "epoch": 1.3231695840391493,
-      "grad_norm": 21.831268310546875,
-      "learning_rate": 2.2060982495765103e-05,
-      "loss": 0.9922,
+      "grad_norm": 4.413284778594971,
+      "learning_rate": 0.0003676830415960851,
+      "loss": 0.7948,
       "step": 7030
     },
     {
       "epoch": 1.3250517598343685,
-      "grad_norm": 16.495553970336914,
-      "learning_rate": 2.204968944099379e-05,
-      "loss": 1.0543,
+      "grad_norm": 3.844486713409424,
+      "learning_rate": 0.00036749482401656316,
+      "loss": 0.8911,
       "step": 7040
     },
     {
       "epoch": 1.3269339356295877,
-      "grad_norm": 10.989031791687012,
-      "learning_rate": 2.2038396386222475e-05,
-      "loss": 0.9902,
+      "grad_norm": 3.5554070472717285,
+      "learning_rate": 0.00036730660643704123,
+      "loss": 0.7859,
       "step": 7050
     },
     {
       "epoch": 1.3288161114248072,
-      "grad_norm": 10.473587036132812,
-      "learning_rate": 2.2027103331451158e-05,
-      "loss": 0.8303,
+      "grad_norm": 2.1481847763061523,
+      "learning_rate": 0.0003671183888575193,
+      "loss": 0.7252,
       "step": 7060
     },
     {
       "epoch": 1.3306982872200264,
-      "grad_norm": 11.674604415893555,
-      "learning_rate": 2.2015810276679844e-05,
-      "loss": 0.79,
+      "grad_norm": 2.8761439323425293,
+      "learning_rate": 0.0003669301712779974,
+      "loss": 0.6704,
       "step": 7070
     },
     {
       "epoch": 1.3325804630152456,
-      "grad_norm": 8.529352188110352,
-      "learning_rate": 2.2004517221908526e-05,
-      "loss": 1.1502,
+      "grad_norm": 3.5740177631378174,
+      "learning_rate": 0.00036674195369847546,
+      "loss": 0.9431,
       "step": 7080
     },
     {
       "epoch": 1.3344626388104648,
-      "grad_norm": 14.45956802368164,
-      "learning_rate": 2.1993224167137212e-05,
-      "loss": 1.0809,
+      "grad_norm": 6.081122875213623,
+      "learning_rate": 0.00036655373611895354,
+      "loss": 0.8061,
       "step": 7090
     },
     {
       "epoch": 1.3363448146056842,
-      "grad_norm": 7.735544681549072,
-      "learning_rate": 2.1981931112365895e-05,
-      "loss": 1.4952,
+      "grad_norm": 3.7727537155151367,
+      "learning_rate": 0.00036636551853943156,
+      "loss": 1.3814,
       "step": 7100
     },
     {
       "epoch": 1.3382269904009034,
-      "grad_norm": 4.9393134117126465,
-      "learning_rate": 2.1970638057594578e-05,
-      "loss": 0.8674,
+      "grad_norm": 4.154539585113525,
+      "learning_rate": 0.00036617730095990964,
+      "loss": 0.6843,
       "step": 7110
     },
     {
       "epoch": 1.3401091661961226,
-      "grad_norm": 15.033855438232422,
-      "learning_rate": 2.1959345002823264e-05,
-      "loss": 0.8618,
+      "grad_norm": 5.381721019744873,
+      "learning_rate": 0.0003659890833803877,
+      "loss": 0.7377,
       "step": 7120
     },
     {
       "epoch": 1.341991341991342,
-      "grad_norm": 6.540989398956299,
-      "learning_rate": 2.1948051948051947e-05,
-      "loss": 0.8127,
+      "grad_norm": 2.434192657470703,
+      "learning_rate": 0.0003658008658008658,
+      "loss": 0.6333,
       "step": 7130
     },
     {
       "epoch": 1.3438735177865613,
-      "grad_norm": 5.148128032684326,
-      "learning_rate": 2.1936758893280633e-05,
-      "loss": 0.8892,
+      "grad_norm": 1.2883131504058838,
+      "learning_rate": 0.00036561264822134387,
+      "loss": 0.8143,
       "step": 7140
     },
     {
       "epoch": 1.3457556935817805,
-      "grad_norm": 54.40917205810547,
-      "learning_rate": 2.192546583850932e-05,
-      "loss": 1.0773,
+      "grad_norm": 6.206827640533447,
+      "learning_rate": 0.00036542443064182194,
+      "loss": 0.7948,
       "step": 7150
     },
     {
       "epoch": 1.347637869377,
-      "grad_norm": 20.81314468383789,
-      "learning_rate": 2.1914172783738e-05,
-      "loss": 1.0652,
+      "grad_norm": 2.471853017807007,
+      "learning_rate": 0.0003652362130623,
+      "loss": 0.807,
       "step": 7160
     },
     {
       "epoch": 1.3495200451722191,
-      "grad_norm": 7.903687000274658,
-      "learning_rate": 2.1902879728966687e-05,
-      "loss": 0.8812,
+      "grad_norm": 2.1814422607421875,
+      "learning_rate": 0.0003650479954827781,
+      "loss": 0.778,
       "step": 7170
     },
     {
       "epoch": 1.3514022209674383,
-      "grad_norm": 13.962177276611328,
-      "learning_rate": 2.189158667419537e-05,
-      "loss": 1.0603,
+      "grad_norm": 3.2641725540161133,
+      "learning_rate": 0.00036485977790325617,
+      "loss": 0.8233,
       "step": 7180
     },
     {
       "epoch": 1.3532843967626578,
-      "grad_norm": 4.5833892822265625,
-      "learning_rate": 2.1880293619424053e-05,
-      "loss": 1.2734,
+      "grad_norm": 0.6247275471687317,
+      "learning_rate": 0.0003646715603237342,
+      "loss": 1.2489,
       "step": 7190
     },
     {
       "epoch": 1.355166572557877,
-      "grad_norm": 13.545757293701172,
-      "learning_rate": 2.186900056465274e-05,
-      "loss": 0.9052,
+      "grad_norm": 2.727606773376465,
+      "learning_rate": 0.0003644833427442123,
+      "loss": 0.8297,
       "step": 7200
     },
     {
       "epoch": 1.3570487483530962,
-      "grad_norm": 31.69056510925293,
-      "learning_rate": 2.185770750988142e-05,
-      "loss": 0.8867,
+      "grad_norm": 3.449462413787842,
+      "learning_rate": 0.0003642951251646904,
+      "loss": 0.8319,
       "step": 7210
     },
     {
       "epoch": 1.3589309241483154,
-      "grad_norm": 13.554387092590332,
-      "learning_rate": 2.1846414455110108e-05,
-      "loss": 0.8744,
+      "grad_norm": 3.3109967708587646,
+      "learning_rate": 0.0003641069075851685,
+      "loss": 0.8439,
       "step": 7220
     },
     {
       "epoch": 1.3608130999435346,
-      "grad_norm": 4.856804370880127,
-      "learning_rate": 2.1835121400338794e-05,
-      "loss": 0.8372,
+      "grad_norm": 3.461745023727417,
+      "learning_rate": 0.00036391869000564655,
+      "loss": 0.807,
       "step": 7230
     },
     {
       "epoch": 1.362695275738754,
-      "grad_norm": 12.263862609863281,
-      "learning_rate": 2.1823828345567476e-05,
-      "loss": 1.0886,
+      "grad_norm": 3.968158721923828,
+      "learning_rate": 0.00036373047242612463,
+      "loss": 0.784,
       "step": 7240
     },
     {
       "epoch": 1.3645774515339733,
-      "grad_norm": 18.752655029296875,
-      "learning_rate": 2.1812535290796163e-05,
-      "loss": 0.7373,
+      "grad_norm": 2.976491928100586,
+      "learning_rate": 0.0003635422548466027,
+      "loss": 0.6565,
       "step": 7250
     },
     {
       "epoch": 1.3664596273291925,
-      "grad_norm": 22.475677490234375,
-      "learning_rate": 2.1801242236024845e-05,
-      "loss": 1.0194,
+      "grad_norm": 5.492811679840088,
+      "learning_rate": 0.0003633540372670808,
+      "loss": 0.889,
       "step": 7260
     },
     {
       "epoch": 1.368341803124412,
-      "grad_norm": 3.7865233421325684,
-      "learning_rate": 2.178994918125353e-05,
-      "loss": 0.9387,
+      "grad_norm": 2.031484365463257,
+      "learning_rate": 0.00036316581968755886,
+      "loss": 0.8284,
       "step": 7270
     },
     {
       "epoch": 1.370223978919631,
-      "grad_norm": 23.85765266418457,
-      "learning_rate": 2.1778656126482214e-05,
-      "loss": 1.1156,
+      "grad_norm": 2.7551069259643555,
+      "learning_rate": 0.0003629776021080369,
+      "loss": 0.9452,
       "step": 7280
     },
     {
       "epoch": 1.3721061547148503,
-      "grad_norm": 13.753911972045898,
-      "learning_rate": 2.1767363071710897e-05,
-      "loss": 0.89,
+      "grad_norm": 2.5312998294830322,
+      "learning_rate": 0.00036278938452851496,
+      "loss": 0.8268,
       "step": 7290
     },
     {
       "epoch": 1.3739883305100697,
-      "grad_norm": 13.338386535644531,
-      "learning_rate": 2.1756070016939583e-05,
-      "loss": 1.0544,
+      "grad_norm": 2.5570919513702393,
+      "learning_rate": 0.00036260116694899303,
+      "loss": 0.7484,
       "step": 7300
     },
     {
       "epoch": 1.375870506305289,
-      "grad_norm": 7.914046287536621,
-      "learning_rate": 2.1744776962168265e-05,
-      "loss": 0.9427,
+      "grad_norm": 2.838815450668335,
+      "learning_rate": 0.0003624129493694711,
+      "loss": 0.9233,
       "step": 7310
     },
     {
       "epoch": 1.3777526821005082,
-      "grad_norm": 33.587493896484375,
-      "learning_rate": 2.173348390739695e-05,
-      "loss": 1.066,
+      "grad_norm": 4.210334777832031,
+      "learning_rate": 0.0003622247317899492,
+      "loss": 0.9044,
       "step": 7320
     },
     {
       "epoch": 1.3796348578957276,
-      "grad_norm": 5.1027326583862305,
-      "learning_rate": 2.1722190852625638e-05,
-      "loss": 0.6101,
+      "grad_norm": 1.67719566822052,
+      "learning_rate": 0.00036203651421042726,
+      "loss": 0.4253,
       "step": 7330
     },
     {
       "epoch": 1.3815170336909468,
-      "grad_norm": 0.7987704277038574,
-      "learning_rate": 2.171089779785432e-05,
-      "loss": 1.1217,
+      "grad_norm": 0.12403179705142975,
+      "learning_rate": 0.00036184829663090534,
+      "loss": 0.6734,
       "step": 7340
     },
     {
       "epoch": 1.383399209486166,
-      "grad_norm": 9.396214485168457,
-      "learning_rate": 2.1699604743083006e-05,
-      "loss": 1.0526,
+      "grad_norm": 6.574319362640381,
+      "learning_rate": 0.0003616600790513834,
+      "loss": 0.8685,
       "step": 7350
     },
     {
       "epoch": 1.3852813852813852,
-      "grad_norm": 21.794408798217773,
-      "learning_rate": 2.1688311688311686e-05,
-      "loss": 0.8303,
+      "grad_norm": 1.8935904502868652,
+      "learning_rate": 0.00036147186147186144,
+      "loss": 0.7353,
       "step": 7360
     },
     {
       "epoch": 1.3871635610766044,
-      "grad_norm": 22.232450485229492,
-      "learning_rate": 2.1677018633540372e-05,
-      "loss": 1.1084,
+      "grad_norm": 5.131047248840332,
+      "learning_rate": 0.0003612836438923395,
+      "loss": 0.6971,
       "step": 7370
     },
     {
       "epoch": 1.3890457368718239,
-      "grad_norm": 7.189990520477295,
-      "learning_rate": 2.1665725578769058e-05,
-      "loss": 1.1183,
+      "grad_norm": 5.154910564422607,
+      "learning_rate": 0.0003610954263128176,
+      "loss": 0.9539,
       "step": 7380
     },
     {
       "epoch": 1.390927912667043,
-      "grad_norm": 4.046782970428467,
-      "learning_rate": 2.165443252399774e-05,
-      "loss": 0.6394,
+      "grad_norm": 1.3230202198028564,
+      "learning_rate": 0.00036090720873329567,
+      "loss": 0.514,
       "step": 7390
     },
     {
       "epoch": 1.3928100884622623,
-      "grad_norm": 6.194625377655029,
-      "learning_rate": 2.1643139469226427e-05,
-      "loss": 0.8828,
+      "grad_norm": 1.5253217220306396,
+      "learning_rate": 0.00036071899115377374,
+      "loss": 0.8314,
       "step": 7400
     },
     {
       "epoch": 1.3946922642574817,
-      "grad_norm": 15.320564270019531,
-      "learning_rate": 2.163184641445511e-05,
-      "loss": 0.8554,
+      "grad_norm": 3.205615520477295,
+      "learning_rate": 0.0003605307735742518,
+      "loss": 0.7711,
       "step": 7410
     },
     {
       "epoch": 1.396574440052701,
-      "grad_norm": 9.800172805786133,
-      "learning_rate": 2.1620553359683795e-05,
-      "loss": 0.9175,
+      "grad_norm": 2.362624168395996,
+      "learning_rate": 0.00036034255599472995,
+      "loss": 0.8926,
       "step": 7420
     },
     {
       "epoch": 1.3984566158479201,
-      "grad_norm": 4.258259296417236,
-      "learning_rate": 2.160926030491248e-05,
-      "loss": 0.872,
+      "grad_norm": 0.3860720992088318,
+      "learning_rate": 0.00036015433841520803,
+      "loss": 0.7266,
       "step": 7430
     },
     {
       "epoch": 1.4003387916431396,
-      "grad_norm": 16.88814353942871,
-      "learning_rate": 2.1597967250141164e-05,
-      "loss": 0.9596,
+      "grad_norm": 5.132907390594482,
+      "learning_rate": 0.0003599661208356861,
+      "loss": 0.8531,
       "step": 7440
     },
     {
       "epoch": 1.4022209674383588,
-      "grad_norm": 11.031537055969238,
-      "learning_rate": 2.1586674195369847e-05,
-      "loss": 0.8293,
+      "grad_norm": 2.3389828205108643,
+      "learning_rate": 0.0003597779032561641,
+      "loss": 0.5889,
       "step": 7450
     },
     {
       "epoch": 1.404103143233578,
-      "grad_norm": 20.199750900268555,
-      "learning_rate": 2.1575381140598533e-05,
-      "loss": 0.9311,
+      "grad_norm": 6.251373291015625,
+      "learning_rate": 0.0003595896856766422,
+      "loss": 0.8839,
       "step": 7460
     },
     {
       "epoch": 1.4059853190287974,
-      "grad_norm": 10.844975471496582,
-      "learning_rate": 2.1564088085827216e-05,
-      "loss": 0.8357,
+      "grad_norm": 3.035172939300537,
+      "learning_rate": 0.0003594014680971203,
+      "loss": 0.5464,
       "step": 7470
     },
     {
       "epoch": 1.4078674948240166,
-      "grad_norm": 26.47841453552246,
-      "learning_rate": 2.15527950310559e-05,
-      "loss": 0.9514,
+      "grad_norm": 5.207313060760498,
+      "learning_rate": 0.00035921325051759836,
+      "loss": 0.8323,
       "step": 7480
     },
     {
       "epoch": 1.4097496706192358,
-      "grad_norm": 27.28598976135254,
-      "learning_rate": 2.1541501976284584e-05,
-      "loss": 1.2359,
+      "grad_norm": 2.937911033630371,
+      "learning_rate": 0.00035902503293807643,
+      "loss": 1.0031,
       "step": 7490
     },
     {
       "epoch": 1.411631846414455,
-      "grad_norm": 22.267330169677734,
-      "learning_rate": 2.153020892151327e-05,
-      "loss": 0.7749,
+      "grad_norm": 3.0856587886810303,
+      "learning_rate": 0.0003588368153585545,
+      "loss": 0.6141,
       "step": 7500
     },
     {
       "epoch": 1.4135140222096743,
-      "grad_norm": 10.783427238464355,
-      "learning_rate": 2.1518915866741956e-05,
-      "loss": 0.7608,
+      "grad_norm": 3.810689926147461,
+      "learning_rate": 0.0003586485977790326,
+      "loss": 0.6492,
       "step": 7510
     },
     {
       "epoch": 1.4153961980048937,
-      "grad_norm": 12.085504531860352,
-      "learning_rate": 2.150762281197064e-05,
-      "loss": 0.8435,
+      "grad_norm": 3.4080021381378174,
+      "learning_rate": 0.00035846038019951066,
+      "loss": 0.4189,
       "step": 7520
     },
     {
       "epoch": 1.417278373800113,
-      "grad_norm": 4.772538185119629,
-      "learning_rate": 2.1496329757199325e-05,
-      "loss": 1.1515,
+      "grad_norm": 1.4870398044586182,
+      "learning_rate": 0.00035827216261998874,
+      "loss": 0.9096,
       "step": 7530
     },
     {
       "epoch": 1.419160549595332,
-      "grad_norm": 16.843612670898438,
-      "learning_rate": 2.1485036702428005e-05,
-      "loss": 0.6556,
+      "grad_norm": 4.46442174911499,
+      "learning_rate": 0.00035808394504046676,
+      "loss": 0.6143,
       "step": 7540
     },
     {
       "epoch": 1.4210427253905515,
-      "grad_norm": 16.49220085144043,
-      "learning_rate": 2.147374364765669e-05,
-      "loss": 0.9657,
+      "grad_norm": 4.368817329406738,
+      "learning_rate": 0.00035789572746094484,
+      "loss": 0.9284,
       "step": 7550
     },
     {
       "epoch": 1.4229249011857708,
-      "grad_norm": 9.660463333129883,
-      "learning_rate": 2.1462450592885377e-05,
-      "loss": 1.0375,
+      "grad_norm": 1.6569066047668457,
+      "learning_rate": 0.0003577075098814229,
+      "loss": 0.9065,
       "step": 7560
     },
     {
       "epoch": 1.42480707698099,
-      "grad_norm": 14.508869171142578,
-      "learning_rate": 2.145115753811406e-05,
-      "loss": 1.191,
+      "grad_norm": 7.855297565460205,
+      "learning_rate": 0.000357519292301901,
+      "loss": 1.0547,
       "step": 7570
     },
     {
       "epoch": 1.4266892527762094,
-      "grad_norm": 9.467730522155762,
-      "learning_rate": 2.1439864483342745e-05,
-      "loss": 0.911,
+      "grad_norm": 3.0609915256500244,
+      "learning_rate": 0.00035733107472237907,
+      "loss": 0.8127,
       "step": 7580
     },
     {
       "epoch": 1.4285714285714286,
-      "grad_norm": 9.366085052490234,
-      "learning_rate": 2.1428571428571428e-05,
-      "loss": 0.6016,
+      "grad_norm": 1.1458548307418823,
+      "learning_rate": 0.00035714285714285714,
+      "loss": 0.6241,
       "step": 7590
     },
     {
       "epoch": 1.4304536043666478,
-      "grad_norm": 6.722076892852783,
-      "learning_rate": 2.1417278373800114e-05,
-      "loss": 1.1092,
+      "grad_norm": 2.866499662399292,
+      "learning_rate": 0.0003569546395633352,
+      "loss": 0.7273,
       "step": 7600
     },
     {
       "epoch": 1.4323357801618672,
-      "grad_norm": 7.0276336669921875,
-      "learning_rate": 2.14059853190288e-05,
-      "loss": 0.8131,
+      "grad_norm": 4.291321754455566,
+      "learning_rate": 0.0003567664219838133,
+      "loss": 0.6635,
       "step": 7610
     },
     {
       "epoch": 1.4342179559570865,
-      "grad_norm": 9.224479675292969,
-      "learning_rate": 2.1394692264257483e-05,
-      "loss": 0.8081,
+      "grad_norm": 2.3771772384643555,
+      "learning_rate": 0.00035657820440429137,
+      "loss": 0.8284,
       "step": 7620
     },
     {
       "epoch": 1.4361001317523057,
-      "grad_norm": 18.966449737548828,
-      "learning_rate": 2.1383399209486166e-05,
-      "loss": 0.8604,
+      "grad_norm": 6.619555950164795,
+      "learning_rate": 0.0003563899868247694,
+      "loss": 0.7846,
       "step": 7630
     },
     {
       "epoch": 1.4379823075475249,
-      "grad_norm": 6.464701175689697,
-      "learning_rate": 2.137210615471485e-05,
-      "loss": 0.7825,
+      "grad_norm": 2.625793695449829,
+      "learning_rate": 0.0003562017692452475,
+      "loss": 0.6474,
       "step": 7640
     },
     {
       "epoch": 1.439864483342744,
-      "grad_norm": 31.871877670288086,
-      "learning_rate": 2.1360813099943534e-05,
-      "loss": 0.7266,
+      "grad_norm": 3.1928863525390625,
+      "learning_rate": 0.0003560135516657256,
+      "loss": 0.4706,
       "step": 7650
     },
     {
       "epoch": 1.4417466591379635,
-      "grad_norm": 0.4453635513782501,
-      "learning_rate": 2.134952004517222e-05,
-      "loss": 0.9575,
+      "grad_norm": 0.3685864210128784,
+      "learning_rate": 0.0003558253340862037,
+      "loss": 0.7511,
       "step": 7660
     },
     {
       "epoch": 1.4436288349331827,
-      "grad_norm": 20.737749099731445,
-      "learning_rate": 2.1338226990400903e-05,
-      "loss": 0.8743,
+      "grad_norm": 6.002312660217285,
+      "learning_rate": 0.00035563711650668175,
+      "loss": 0.652,
       "step": 7670
     },
     {
       "epoch": 1.445511010728402,
-      "grad_norm": 7.961155414581299,
-      "learning_rate": 2.132693393562959e-05,
-      "loss": 0.8414,
+      "grad_norm": 2.449984312057495,
+      "learning_rate": 0.00035544889892715983,
+      "loss": 0.4978,
       "step": 7680
     },
     {
       "epoch": 1.4473931865236214,
-      "grad_norm": 14.688976287841797,
-      "learning_rate": 2.1315640880858275e-05,
-      "loss": 0.8901,
+      "grad_norm": 5.680086612701416,
+      "learning_rate": 0.0003552606813476379,
+      "loss": 0.9043,
       "step": 7690
     },
     {
       "epoch": 1.4492753623188406,
-      "grad_norm": 1.112459421157837,
-      "learning_rate": 2.1304347826086958e-05,
-      "loss": 0.7949,
+      "grad_norm": 1.8493766784667969,
+      "learning_rate": 0.000355072463768116,
+      "loss": 0.7197,
       "step": 7700
     },
     {
       "epoch": 1.4511575381140598,
-      "grad_norm": 1.5727367401123047,
-      "learning_rate": 2.129305477131564e-05,
-      "loss": 0.834,
+      "grad_norm": 0.3454371392726898,
+      "learning_rate": 0.000354884246188594,
+      "loss": 0.6712,
       "step": 7710
     },
     {
       "epoch": 1.4530397139092792,
-      "grad_norm": 15.447263717651367,
-      "learning_rate": 2.1281761716544323e-05,
-      "loss": 0.8197,
+      "grad_norm": 3.6451921463012695,
+      "learning_rate": 0.0003546960286090721,
+      "loss": 0.8027,
       "step": 7720
     },
     {
       "epoch": 1.4549218897044984,
-      "grad_norm": 16.137331008911133,
-      "learning_rate": 2.127046866177301e-05,
-      "loss": 0.7958,
+      "grad_norm": 3.223358154296875,
+      "learning_rate": 0.00035450781102955016,
+      "loss": 0.7259,
       "step": 7730
     },
     {
       "epoch": 1.4568040654997176,
-      "grad_norm": 10.978745460510254,
-      "learning_rate": 2.1259175607001696e-05,
-      "loss": 1.0626,
+      "grad_norm": 3.4901061058044434,
+      "learning_rate": 0.00035431959345002823,
+      "loss": 0.8298,
       "step": 7740
     },
     {
       "epoch": 1.458686241294937,
-      "grad_norm": 13.138747215270996,
-      "learning_rate": 2.1247882552230378e-05,
-      "loss": 0.8053,
+      "grad_norm": 1.583277940750122,
+      "learning_rate": 0.0003541313758705063,
+      "loss": 0.6178,
       "step": 7750
     },
     {
       "epoch": 1.4605684170901563,
-      "grad_norm": 15.29321575164795,
-      "learning_rate": 2.1236589497459064e-05,
-      "loss": 1.0308,
+      "grad_norm": 1.1956707239151,
+      "learning_rate": 0.0003539431582909844,
+      "loss": 0.8466,
       "step": 7760
     },
     {
       "epoch": 1.4624505928853755,
-      "grad_norm": 7.9823527336120605,
-      "learning_rate": 2.1225296442687747e-05,
-      "loss": 1.0307,
+      "grad_norm": 2.058783769607544,
+      "learning_rate": 0.00035375494071146246,
+      "loss": 0.905,
       "step": 7770
     },
     {
       "epoch": 1.4643327686805947,
-      "grad_norm": 28.051477432250977,
-      "learning_rate": 2.1214003387916433e-05,
-      "loss": 0.9783,
+      "grad_norm": 8.526945114135742,
+      "learning_rate": 0.00035356672313194054,
+      "loss": 0.9288,
       "step": 7780
     },
     {
       "epoch": 1.466214944475814,
-      "grad_norm": 2.81699800491333,
-      "learning_rate": 2.120271033314512e-05,
-      "loss": 0.6847,
+      "grad_norm": 3.497265577316284,
+      "learning_rate": 0.0003533785055524186,
+      "loss": 0.6056,
       "step": 7790
     },
     {
       "epoch": 1.4680971202710333,
-      "grad_norm": 20.243371963500977,
-      "learning_rate": 2.11914172783738e-05,
-      "loss": 0.8305,
+      "grad_norm": 2.1482982635498047,
+      "learning_rate": 0.00035319028797289664,
+      "loss": 0.6909,
       "step": 7800
     },
     {
       "epoch": 1.4699792960662525,
-      "grad_norm": 5.7048115730285645,
-      "learning_rate": 2.1180124223602485e-05,
-      "loss": 0.6723,
+      "grad_norm": 2.2635204792022705,
+      "learning_rate": 0.0003530020703933747,
+      "loss": 0.6319,
       "step": 7810
     },
     {
       "epoch": 1.4718614718614718,
-      "grad_norm": 14.926733016967773,
-      "learning_rate": 2.1168831168831167e-05,
-      "loss": 0.9217,
+      "grad_norm": 2.105267286300659,
+      "learning_rate": 0.0003528138528138528,
+      "loss": 0.7792,
       "step": 7820
     },
     {
       "epoch": 1.4737436476566912,
-      "grad_norm": 22.269062042236328,
-      "learning_rate": 2.1157538114059853e-05,
-      "loss": 1.0226,
+      "grad_norm": 2.60502552986145,
+      "learning_rate": 0.00035262563523433087,
+      "loss": 0.8495,
       "step": 7830
     },
     {
       "epoch": 1.4756258234519104,
-      "grad_norm": 5.699794292449951,
-      "learning_rate": 2.114624505928854e-05,
-      "loss": 0.749,
+      "grad_norm": 1.7702150344848633,
+      "learning_rate": 0.00035243741765480894,
+      "loss": 0.6141,
       "step": 7840
     },
     {
       "epoch": 1.4775079992471296,
-      "grad_norm": 5.858451843261719,
-      "learning_rate": 2.1134952004517222e-05,
-      "loss": 0.7048,
+      "grad_norm": 4.5006184577941895,
+      "learning_rate": 0.000352249200075287,
+      "loss": 0.7274,
       "step": 7850
     },
     {
       "epoch": 1.479390175042349,
-      "grad_norm": 8.59745979309082,
-      "learning_rate": 2.1123658949745908e-05,
-      "loss": 0.9316,
+      "grad_norm": 3.0699360370635986,
+      "learning_rate": 0.00035206098249576515,
+      "loss": 0.8281,
       "step": 7860
     },
     {
       "epoch": 1.4812723508375683,
-      "grad_norm": 12.093297958374023,
-      "learning_rate": 2.111236589497459e-05,
-      "loss": 1.0178,
+      "grad_norm": 8.668073654174805,
+      "learning_rate": 0.00035187276491624323,
+      "loss": 0.8215,
       "step": 7870
     },
     {
       "epoch": 1.4831545266327875,
-      "grad_norm": 12.393393516540527,
-      "learning_rate": 2.1101072840203277e-05,
-      "loss": 0.7238,
+      "grad_norm": 7.539294242858887,
+      "learning_rate": 0.0003516845473367213,
+      "loss": 0.8282,
       "step": 7880
     },
     {
       "epoch": 1.485036702428007,
-      "grad_norm": 46.78508758544922,
-      "learning_rate": 2.108977978543196e-05,
-      "loss": 0.8131,
+      "grad_norm": 3.0696523189544678,
+      "learning_rate": 0.0003514963297571993,
+      "loss": 0.6651,
       "step": 7890
     },
     {
       "epoch": 1.486918878223226,
-      "grad_norm": 1.7145179510116577,
-      "learning_rate": 2.1078486730660642e-05,
-      "loss": 0.7799,
+      "grad_norm": 2.3803205490112305,
+      "learning_rate": 0.0003513081121776774,
+      "loss": 0.7852,
       "step": 7900
     },
     {
       "epoch": 1.4888010540184453,
-      "grad_norm": 24.29248809814453,
-      "learning_rate": 2.106719367588933e-05,
-      "loss": 1.083,
+      "grad_norm": 1.9712696075439453,
+      "learning_rate": 0.0003511198945981555,
+      "loss": 0.8236,
       "step": 7910
     },
     {
       "epoch": 1.4906832298136645,
-      "grad_norm": 10.044913291931152,
-      "learning_rate": 2.105590062111801e-05,
-      "loss": 0.7995,
+      "grad_norm": 3.2915306091308594,
+      "learning_rate": 0.00035093167701863356,
+      "loss": 0.7561,
       "step": 7920
     },
     {
       "epoch": 1.4925654056088837,
-      "grad_norm": 17.686159133911133,
-      "learning_rate": 2.1044607566346697e-05,
-      "loss": 1.23,
+      "grad_norm": 2.5508902072906494,
+      "learning_rate": 0.00035074345943911163,
+      "loss": 1.0034,
       "step": 7930
     },
     {
       "epoch": 1.4944475814041032,
-      "grad_norm": 12.747879981994629,
-      "learning_rate": 2.1033314511575383e-05,
-      "loss": 1.0773,
+      "grad_norm": 4.021931171417236,
+      "learning_rate": 0.0003505552418595897,
+      "loss": 0.9428,
       "step": 7940
     },
     {
       "epoch": 1.4963297571993224,
-      "grad_norm": 14.406421661376953,
-      "learning_rate": 2.1022021456804066e-05,
-      "loss": 0.8235,
+      "grad_norm": 3.7076730728149414,
+      "learning_rate": 0.0003503670242800678,
+      "loss": 0.7309,
       "step": 7950
     },
     {
       "epoch": 1.4982119329945416,
-      "grad_norm": 9.100472450256348,
-      "learning_rate": 2.1010728402032752e-05,
-      "loss": 0.5385,
+      "grad_norm": 2.309861421585083,
+      "learning_rate": 0.00035017880670054586,
+      "loss": 0.4758,
       "step": 7960
     },
     {
       "epoch": 1.500094108789761,
-      "grad_norm": 1.3265353441238403,
-      "learning_rate": 2.0999435347261435e-05,
-      "loss": 0.7218,
+      "grad_norm": 4.0389862060546875,
+      "learning_rate": 0.0003499905891210239,
+      "loss": 0.7024,
       "step": 7970
     },
     {
       "epoch": 1.5019762845849802,
-      "grad_norm": 22.250818252563477,
-      "learning_rate": 2.0988142292490117e-05,
-      "loss": 0.9737,
+      "grad_norm": 3.7609620094299316,
+      "learning_rate": 0.00034980237154150196,
+      "loss": 0.8486,
       "step": 7980
     },
     {
       "epoch": 1.5038584603801994,
-      "grad_norm": 6.612005710601807,
-      "learning_rate": 2.0976849237718803e-05,
-      "loss": 0.7305,
+      "grad_norm": 2.8926572799682617,
+      "learning_rate": 0.00034961415396198004,
+      "loss": 0.5895,
       "step": 7990
     },
     {
       "epoch": 1.5057406361754189,
-      "grad_norm": 26.889253616333008,
-      "learning_rate": 2.0965556182947486e-05,
-      "loss": 0.82,
+      "grad_norm": 4.276566982269287,
+      "learning_rate": 0.0003494259363824581,
+      "loss": 0.6477,
       "step": 8000
     },
     {
       "epoch": 1.507622811970638,
-      "grad_norm": 15.2909574508667,
-      "learning_rate": 2.0954263128176172e-05,
-      "loss": 0.8488,
+      "grad_norm": 3.2659835815429688,
+      "learning_rate": 0.0003492377188029362,
+      "loss": 0.7834,
       "step": 8010
     },
     {
       "epoch": 1.5095049877658573,
-      "grad_norm": 14.072726249694824,
-      "learning_rate": 2.0942970073404858e-05,
-      "loss": 0.7917,
+      "grad_norm": 2.6856377124786377,
+      "learning_rate": 0.00034904950122341427,
+      "loss": 0.6446,
       "step": 8020
     },
     {
       "epoch": 1.5113871635610767,
-      "grad_norm": 19.601593017578125,
-      "learning_rate": 2.093167701863354e-05,
-      "loss": 0.8319,
+      "grad_norm": 2.0098695755004883,
+      "learning_rate": 0.00034886128364389234,
+      "loss": 0.7457,
       "step": 8030
     },
     {
       "epoch": 1.513269339356296,
-      "grad_norm": 11.248756408691406,
-      "learning_rate": 2.0920383963862227e-05,
-      "loss": 0.8455,
+      "grad_norm": 2.1096489429473877,
+      "learning_rate": 0.0003486730660643704,
+      "loss": 0.7569,
       "step": 8040
     },
     {
       "epoch": 1.5151515151515151,
-      "grad_norm": 14.29757022857666,
-      "learning_rate": 2.090909090909091e-05,
-      "loss": 0.836,
+      "grad_norm": 2.1530377864837646,
+      "learning_rate": 0.0003484848484848485,
+      "loss": 0.6903,
       "step": 8050
     },
     {
       "epoch": 1.5170336909467346,
-      "grad_norm": 18.773252487182617,
-      "learning_rate": 2.0897797854319592e-05,
-      "loss": 0.7697,
+      "grad_norm": 3.4614245891571045,
+      "learning_rate": 0.0003482966309053265,
+      "loss": 0.7235,
       "step": 8060
     },
     {
       "epoch": 1.5189158667419536,
-      "grad_norm": 13.430651664733887,
-      "learning_rate": 2.088650479954828e-05,
-      "loss": 0.7545,
+      "grad_norm": 2.9231855869293213,
+      "learning_rate": 0.0003481084133258046,
+      "loss": 0.6586,
       "step": 8070
     },
     {
       "epoch": 1.520798042537173,
-      "grad_norm": 18.43965721130371,
-      "learning_rate": 2.087521174477696e-05,
-      "loss": 0.9061,
+      "grad_norm": 3.0614635944366455,
+      "learning_rate": 0.00034792019574628267,
+      "loss": 0.827,
       "step": 8080
     },
     {
       "epoch": 1.5226802183323922,
-      "grad_norm": 4.812375068664551,
-      "learning_rate": 2.0863918690005647e-05,
-      "loss": 0.7921,
+      "grad_norm": 1.5056312084197998,
+      "learning_rate": 0.0003477319781667608,
+      "loss": 0.7082,
       "step": 8090
     },
     {
       "epoch": 1.5245623941276114,
-      "grad_norm": 13.704950332641602,
-      "learning_rate": 2.085262563523433e-05,
-      "loss": 1.0409,
+      "grad_norm": 1.0233908891677856,
+      "learning_rate": 0.0003475437605872389,
+      "loss": 0.7483,
       "step": 8100
     },
     {
       "epoch": 1.5264445699228308,
-      "grad_norm": 7.7468414306640625,
-      "learning_rate": 2.0841332580463016e-05,
-      "loss": 0.8707,
+      "grad_norm": 2.48866868019104,
+      "learning_rate": 0.00034735554300771695,
+      "loss": 0.8398,
       "step": 8110
     },
     {
       "epoch": 1.52832674571805,
-      "grad_norm": 8.92183780670166,
-      "learning_rate": 2.0830039525691702e-05,
-      "loss": 1.0549,
+      "grad_norm": 2.297361373901367,
+      "learning_rate": 0.00034716732542819503,
+      "loss": 0.8928,
       "step": 8120
     },
     {
       "epoch": 1.5302089215132693,
-      "grad_norm": 12.238114356994629,
-      "learning_rate": 2.0818746470920385e-05,
-      "loss": 1.0467,
+      "grad_norm": 1.6928592920303345,
+      "learning_rate": 0.0003469791078486731,
+      "loss": 0.7641,
       "step": 8130
     },
     {
       "epoch": 1.5320910973084887,
-      "grad_norm": 15.757767677307129,
-      "learning_rate": 2.080745341614907e-05,
-      "loss": 0.8356,
+      "grad_norm": 3.405965805053711,
+      "learning_rate": 0.0003467908902691512,
+      "loss": 0.8176,
       "step": 8140
     },
     {
       "epoch": 1.533973273103708,
-      "grad_norm": 15.579939842224121,
-      "learning_rate": 2.079616036137775e-05,
-      "loss": 0.7802,
+      "grad_norm": 2.9637680053710938,
+      "learning_rate": 0.0003466026726896292,
+      "loss": 0.7238,
       "step": 8150
     },
     {
       "epoch": 1.5358554488989271,
-      "grad_norm": 7.932995796203613,
-      "learning_rate": 2.0784867306606436e-05,
-      "loss": 0.782,
+      "grad_norm": 1.8810489177703857,
+      "learning_rate": 0.0003464144551101073,
+      "loss": 0.788,
       "step": 8160
     },
     {
       "epoch": 1.5377376246941465,
-      "grad_norm": 34.61065673828125,
-      "learning_rate": 2.0773574251835122e-05,
-      "loss": 1.0706,
+      "grad_norm": 8.370185852050781,
+      "learning_rate": 0.00034622623753058536,
+      "loss": 1.0583,
       "step": 8170
     },
     {
       "epoch": 1.5396198004893658,
-      "grad_norm": 6.347330093383789,
-      "learning_rate": 2.0762281197063805e-05,
-      "loss": 0.862,
+      "grad_norm": 1.2462303638458252,
+      "learning_rate": 0.00034603801995106343,
+      "loss": 0.7773,
       "step": 8180
     },
     {
       "epoch": 1.541501976284585,
-      "grad_norm": 12.102984428405762,
-      "learning_rate": 2.075098814229249e-05,
-      "loss": 0.976,
+      "grad_norm": 2.5813164710998535,
+      "learning_rate": 0.0003458498023715415,
+      "loss": 0.7235,
       "step": 8190
     },
     {
       "epoch": 1.5433841520798044,
-      "grad_norm": 34.155494689941406,
-      "learning_rate": 2.0739695087521177e-05,
-      "loss": 0.9479,
+      "grad_norm": 3.9357566833496094,
+      "learning_rate": 0.0003456615847920196,
+      "loss": 0.7332,
       "step": 8200
     },
     {
       "epoch": 1.5452663278750234,
-      "grad_norm": 4.237630844116211,
-      "learning_rate": 2.072840203274986e-05,
-      "loss": 0.6754,
+      "grad_norm": 1.743631362915039,
+      "learning_rate": 0.00034547336721249766,
+      "loss": 0.6358,
       "step": 8210
     },
     {
       "epoch": 1.5471485036702428,
-      "grad_norm": 9.868550300598145,
-      "learning_rate": 2.0717108977978546e-05,
-      "loss": 0.6415,
+      "grad_norm": 1.980130672454834,
+      "learning_rate": 0.00034528514963297574,
+      "loss": 0.6059,
       "step": 8220
     },
     {
       "epoch": 1.549030679465462,
-      "grad_norm": 9.84210205078125,
-      "learning_rate": 2.070581592320723e-05,
-      "loss": 1.0436,
+      "grad_norm": 3.3148014545440674,
+      "learning_rate": 0.0003450969320534538,
+      "loss": 0.9266,
       "step": 8230
     },
     {
       "epoch": 1.5509128552606812,
-      "grad_norm": 6.024266242980957,
-      "learning_rate": 2.069452286843591e-05,
-      "loss": 0.749,
+      "grad_norm": 4.528003692626953,
+      "learning_rate": 0.00034490871447393184,
+      "loss": 0.7274,
       "step": 8240
     },
     {
       "epoch": 1.5527950310559007,
-      "grad_norm": 24.58380889892578,
-      "learning_rate": 2.0683229813664597e-05,
-      "loss": 1.102,
+      "grad_norm": 2.7898247241973877,
+      "learning_rate": 0.0003447204968944099,
+      "loss": 0.9216,
       "step": 8250
     },
     {
       "epoch": 1.5546772068511199,
-      "grad_norm": 15.549361228942871,
-      "learning_rate": 2.067193675889328e-05,
-      "loss": 0.619,
+      "grad_norm": 5.00869083404541,
+      "learning_rate": 0.000344532279314888,
+      "loss": 0.5532,
       "step": 8260
     },
     {
       "epoch": 1.556559382646339,
-      "grad_norm": 11.536001205444336,
-      "learning_rate": 2.0660643704121966e-05,
-      "loss": 0.7282,
+      "grad_norm": 1.2055312395095825,
+      "learning_rate": 0.00034434406173536607,
+      "loss": 0.5286,
       "step": 8270
     },
     {
       "epoch": 1.5584415584415585,
-      "grad_norm": 1.458553671836853,
-      "learning_rate": 2.064935064935065e-05,
-      "loss": 0.8282,
+      "grad_norm": 0.34826549887657166,
+      "learning_rate": 0.00034415584415584414,
+      "loss": 0.6315,
       "step": 8280
     },
     {
       "epoch": 1.5603237342367777,
-      "grad_norm": 22.097503662109375,
-      "learning_rate": 2.0638057594579335e-05,
-      "loss": 1.32,
+      "grad_norm": 1.7927368879318237,
+      "learning_rate": 0.0003439676265763222,
+      "loss": 1.0556,
       "step": 8290
     },
     {
       "epoch": 1.562205910031997,
-      "grad_norm": 1.0465900897979736,
-      "learning_rate": 2.062676453980802e-05,
-      "loss": 0.6825,
+      "grad_norm": 3.9631786346435547,
+      "learning_rate": 0.00034377940899680035,
+      "loss": 0.5423,
       "step": 8300
     },
     {
       "epoch": 1.5640880858272164,
-      "grad_norm": 4.5169782638549805,
-      "learning_rate": 2.0615471485036704e-05,
-      "loss": 1.0293,
+      "grad_norm": 1.3428804874420166,
+      "learning_rate": 0.00034359119141727843,
+      "loss": 0.9069,
       "step": 8310
     },
     {
       "epoch": 1.5659702616224356,
-      "grad_norm": 24.006298065185547,
-      "learning_rate": 2.0604178430265386e-05,
-      "loss": 0.4358,
+      "grad_norm": 2.6552329063415527,
+      "learning_rate": 0.00034340297383775645,
+      "loss": 0.3473,
       "step": 8320
     },
     {
       "epoch": 1.5678524374176548,
-      "grad_norm": 34.115413665771484,
-      "learning_rate": 2.059288537549407e-05,
-      "loss": 0.7429,
+      "grad_norm": 3.216970920562744,
+      "learning_rate": 0.0003432147562582345,
+      "loss": 0.6878,
       "step": 8330
     },
     {
       "epoch": 1.5697346132128742,
-      "grad_norm": 0.8864557147026062,
-      "learning_rate": 2.0581592320722755e-05,
-      "loss": 0.6736,
+      "grad_norm": 1.658892273902893,
+      "learning_rate": 0.0003430265386787126,
+      "loss": 0.6124,
       "step": 8340
     },
     {
       "epoch": 1.5716167890080932,
-      "grad_norm": 3.0202507972717285,
-      "learning_rate": 2.057029926595144e-05,
-      "loss": 0.6662,
+      "grad_norm": 1.2303948402404785,
+      "learning_rate": 0.0003428383210991907,
+      "loss": 0.456,
       "step": 8350
     },
     {
       "epoch": 1.5734989648033126,
-      "grad_norm": 16.685216903686523,
-      "learning_rate": 2.0559006211180124e-05,
-      "loss": 1.1072,
+      "grad_norm": 3.7826428413391113,
+      "learning_rate": 0.00034265010351966876,
+      "loss": 0.8232,
       "step": 8360
     },
     {
       "epoch": 1.5753811405985318,
-      "grad_norm": 14.748647689819336,
-      "learning_rate": 2.054771315640881e-05,
-      "loss": 0.6912,
+      "grad_norm": 2.520322799682617,
+      "learning_rate": 0.00034246188594014683,
+      "loss": 0.4863,
       "step": 8370
     },
     {
       "epoch": 1.577263316393751,
-      "grad_norm": 19.778024673461914,
-      "learning_rate": 2.0536420101637493e-05,
-      "loss": 1.0492,
+      "grad_norm": 3.2636899948120117,
+      "learning_rate": 0.0003422736683606249,
+      "loss": 0.9228,
       "step": 8380
     },
     {
       "epoch": 1.5791454921889705,
-      "grad_norm": 1.4614105224609375,
-      "learning_rate": 2.052512704686618e-05,
-      "loss": 1.0389,
+      "grad_norm": 0.5615966320037842,
+      "learning_rate": 0.000342085450781103,
+      "loss": 0.8344,
       "step": 8390
     },
     {
       "epoch": 1.5810276679841897,
-      "grad_norm": 2.2111265659332275,
-      "learning_rate": 2.0513833992094865e-05,
-      "loss": 0.7137,
+      "grad_norm": 1.5312647819519043,
+      "learning_rate": 0.00034189723320158106,
+      "loss": 0.5623,
       "step": 8400
     },
     {
       "epoch": 1.582909843779409,
-      "grad_norm": 25.663915634155273,
-      "learning_rate": 2.0502540937323544e-05,
-      "loss": 0.8827,
+      "grad_norm": 2.7488324642181396,
+      "learning_rate": 0.0003417090156220591,
+      "loss": 0.7472,
       "step": 8410
     },
     {
       "epoch": 1.5847920195746283,
-      "grad_norm": 32.34187316894531,
-      "learning_rate": 2.049124788255223e-05,
-      "loss": 1.359,
+      "grad_norm": 2.1798367500305176,
+      "learning_rate": 0.00034152079804253716,
+      "loss": 1.2064,
       "step": 8420
     },
     {
       "epoch": 1.5866741953698476,
-      "grad_norm": 9.49844741821289,
-      "learning_rate": 2.0479954827780913e-05,
-      "loss": 0.7594,
+      "grad_norm": 1.7178131341934204,
+      "learning_rate": 0.00034133258046301524,
+      "loss": 0.7014,
       "step": 8430
     },
     {
       "epoch": 1.5885563711650668,
-      "grad_norm": 8.139872550964355,
-      "learning_rate": 2.04686617730096e-05,
-      "loss": 0.9236,
+      "grad_norm": 1.6949135065078735,
+      "learning_rate": 0.0003411443628834933,
+      "loss": 0.7091,
       "step": 8440
     },
     {
       "epoch": 1.5904385469602862,
-      "grad_norm": 4.609650135040283,
-      "learning_rate": 2.0457368718238285e-05,
-      "loss": 1.0342,
+      "grad_norm": 2.0318641662597656,
+      "learning_rate": 0.0003409561453039714,
+      "loss": 0.8617,
       "step": 8450
     },
     {
       "epoch": 1.5923207227555054,
-      "grad_norm": 13.851377487182617,
-      "learning_rate": 2.0446075663466968e-05,
-      "loss": 0.985,
+      "grad_norm": 2.8767454624176025,
+      "learning_rate": 0.00034076792772444947,
+      "loss": 0.7656,
       "step": 8460
     },
     {
       "epoch": 1.5942028985507246,
-      "grad_norm": 17.376338958740234,
-      "learning_rate": 2.0434782608695654e-05,
-      "loss": 0.8613,
+      "grad_norm": 5.067991256713867,
+      "learning_rate": 0.00034057971014492754,
+      "loss": 0.7261,
       "step": 8470
     },
     {
       "epoch": 1.596085074345944,
-      "grad_norm": 16.559776306152344,
-      "learning_rate": 2.042348955392434e-05,
-      "loss": 0.6948,
+      "grad_norm": 2.4623968601226807,
+      "learning_rate": 0.0003403914925654056,
+      "loss": 0.6039,
       "step": 8480
     },
     {
       "epoch": 1.597967250141163,
-      "grad_norm": 28.88027000427246,
-      "learning_rate": 2.0412196499153023e-05,
-      "loss": 0.7705,
+      "grad_norm": 2.616107940673828,
+      "learning_rate": 0.0003402032749858837,
+      "loss": 0.5545,
       "step": 8490
     },
     {
       "epoch": 1.5998494259363825,
-      "grad_norm": 25.906251907348633,
-      "learning_rate": 2.0400903444381705e-05,
-      "loss": 0.8089,
+      "grad_norm": 2.3766703605651855,
+      "learning_rate": 0.0003400150574063617,
+      "loss": 0.569,
       "step": 8500
     },
     {
       "epoch": 1.601731601731602,
-      "grad_norm": 3.415189266204834,
-      "learning_rate": 2.0389610389610388e-05,
-      "loss": 0.4868,
+      "grad_norm": 1.4852138757705688,
+      "learning_rate": 0.0003398268398268398,
+      "loss": 0.4487,
       "step": 8510
     },
     {
       "epoch": 1.6036137775268209,
-      "grad_norm": 7.368528366088867,
-      "learning_rate": 2.0378317334839074e-05,
-      "loss": 0.6249,
+      "grad_norm": 10.471375465393066,
+      "learning_rate": 0.00033963862224731787,
+      "loss": 0.4925,
       "step": 8520
     },
     {
       "epoch": 1.6054959533220403,
-      "grad_norm": 12.94930362701416,
-      "learning_rate": 2.036702428006776e-05,
-      "loss": 0.8234,
+      "grad_norm": 3.325183629989624,
+      "learning_rate": 0.000339450404667796,
+      "loss": 0.5884,
       "step": 8530
     },
     {
       "epoch": 1.6073781291172595,
-      "grad_norm": 1.4030075073242188,
-      "learning_rate": 2.0355731225296443e-05,
-      "loss": 0.7259,
+      "grad_norm": 0.5877097249031067,
+      "learning_rate": 0.0003392621870882741,
+      "loss": 0.669,
       "step": 8540
     },
     {
       "epoch": 1.6092603049124787,
-      "grad_norm": 21.617116928100586,
-      "learning_rate": 2.034443817052513e-05,
-      "loss": 1.0274,
+      "grad_norm": 1.0777863264083862,
+      "learning_rate": 0.00033907396950875215,
+      "loss": 0.9738,
       "step": 8550
     },
     {
       "epoch": 1.6111424807076982,
-      "grad_norm": 14.721446990966797,
-      "learning_rate": 2.033314511575381e-05,
-      "loss": 0.7038,
+      "grad_norm": 4.0034685134887695,
+      "learning_rate": 0.00033888575192923023,
+      "loss": 0.5929,
       "step": 8560
     },
     {
       "epoch": 1.6130246565029174,
-      "grad_norm": 37.175865173339844,
-      "learning_rate": 2.0321852060982498e-05,
-      "loss": 0.8129,
+      "grad_norm": 0.5476213693618774,
+      "learning_rate": 0.0003386975343497083,
+      "loss": 0.7154,
       "step": 8570
     },
     {
       "epoch": 1.6149068322981366,
-      "grad_norm": 18.872203826904297,
-      "learning_rate": 2.031055900621118e-05,
-      "loss": 0.7054,
+      "grad_norm": 2.511484384536743,
+      "learning_rate": 0.00033850931677018633,
+      "loss": 0.5125,
       "step": 8580
     },
     {
       "epoch": 1.616789008093356,
-      "grad_norm": 18.754467010498047,
-      "learning_rate": 2.0299265951439863e-05,
-      "loss": 0.7435,
+      "grad_norm": 5.705085277557373,
+      "learning_rate": 0.0003383210991906644,
+      "loss": 0.6806,
       "step": 8590
     },
     {
       "epoch": 1.6186711838885752,
-      "grad_norm": 17.79302406311035,
-      "learning_rate": 2.028797289666855e-05,
-      "loss": 0.8313,
+      "grad_norm": 5.182740211486816,
+      "learning_rate": 0.0003381328816111425,
+      "loss": 0.7014,
       "step": 8600
     },
     {
       "epoch": 1.6205533596837944,
-      "grad_norm": 17.659814834594727,
-      "learning_rate": 2.0276679841897232e-05,
-      "loss": 0.7874,
+      "grad_norm": 5.4863080978393555,
+      "learning_rate": 0.00033794466403162056,
+      "loss": 0.664,
       "step": 8610
     },
     {
       "epoch": 1.6224355354790139,
-      "grad_norm": 14.71423625946045,
-      "learning_rate": 2.0265386787125918e-05,
-      "loss": 0.83,
+      "grad_norm": 2.908485174179077,
+      "learning_rate": 0.00033775644645209863,
+      "loss": 0.6387,
       "step": 8620
     },
     {
       "epoch": 1.6243177112742329,
-      "grad_norm": 17.527603149414062,
-      "learning_rate": 2.0254093732354604e-05,
-      "loss": 1.0773,
+      "grad_norm": 4.561892032623291,
+      "learning_rate": 0.0003375682288725767,
+      "loss": 0.9372,
       "step": 8630
     },
     {
       "epoch": 1.6261998870694523,
-      "grad_norm": 12.642634391784668,
-      "learning_rate": 2.0242800677583287e-05,
-      "loss": 0.9208,
+      "grad_norm": 1.4518635272979736,
+      "learning_rate": 0.0003373800112930548,
+      "loss": 0.9095,
       "step": 8640
     },
     {
       "epoch": 1.6280820628646717,
-      "grad_norm": 1.2495397329330444,
-      "learning_rate": 2.0231507622811973e-05,
-      "loss": 0.9029,
+      "grad_norm": 0.5780901908874512,
+      "learning_rate": 0.00033719179371353286,
+      "loss": 0.8878,
       "step": 8650
     },
     {
       "epoch": 1.6299642386598907,
-      "grad_norm": 0.4178260564804077,
-      "learning_rate": 2.0220214568040655e-05,
-      "loss": 0.6525,
+      "grad_norm": 0.13929180800914764,
+      "learning_rate": 0.00033700357613401094,
+      "loss": 0.6141,
       "step": 8660
     },
     {
       "epoch": 1.6318464144551101,
-      "grad_norm": 12.84122085571289,
-      "learning_rate": 2.0208921513269338e-05,
-      "loss": 0.6563,
+      "grad_norm": 0.9517858624458313,
+      "learning_rate": 0.00033681535855448896,
+      "loss": 0.4543,
       "step": 8670
     },
     {
       "epoch": 1.6337285902503293,
-      "grad_norm": 14.375418663024902,
-      "learning_rate": 2.0197628458498024e-05,
-      "loss": 0.9404,
+      "grad_norm": 7.420551300048828,
+      "learning_rate": 0.00033662714097496704,
+      "loss": 0.8612,
       "step": 8680
     },
     {
       "epoch": 1.6356107660455486,
-      "grad_norm": 5.309718132019043,
-      "learning_rate": 2.0186335403726707e-05,
-      "loss": 0.8255,
+      "grad_norm": 3.4701571464538574,
+      "learning_rate": 0.0003364389233954451,
+      "loss": 0.8368,
       "step": 8690
     },
     {
       "epoch": 1.637492941840768,
-      "grad_norm": 19.447336196899414,
-      "learning_rate": 2.0175042348955393e-05,
-      "loss": 0.9448,
+      "grad_norm": 4.836475372314453,
+      "learning_rate": 0.0003362507058159232,
+      "loss": 0.8814,
       "step": 8700
     },
     {
       "epoch": 1.6393751176359872,
-      "grad_norm": 5.150761127471924,
-      "learning_rate": 2.016374929418408e-05,
-      "loss": 0.7773,
+      "grad_norm": 2.2584071159362793,
+      "learning_rate": 0.00033606248823640127,
+      "loss": 0.6878,
       "step": 8710
     },
     {
       "epoch": 1.6412572934312064,
-      "grad_norm": 10.67483139038086,
-      "learning_rate": 2.015245623941276e-05,
-      "loss": 0.6024,
+      "grad_norm": 2.989353656768799,
+      "learning_rate": 0.00033587427065687934,
+      "loss": 0.5362,
       "step": 8720
     },
     {
       "epoch": 1.6431394692264258,
-      "grad_norm": 18.900775909423828,
-      "learning_rate": 2.0141163184641448e-05,
-      "loss": 0.6435,
+      "grad_norm": 4.017577171325684,
+      "learning_rate": 0.0003356860530773574,
+      "loss": 0.5628,
       "step": 8730
     },
     {
       "epoch": 1.645021645021645,
-      "grad_norm": 7.011523246765137,
-      "learning_rate": 2.012987012987013e-05,
-      "loss": 0.9581,
+      "grad_norm": 2.840949773788452,
+      "learning_rate": 0.0003354978354978355,
+      "loss": 0.6433,
       "step": 8740
     },
     {
       "epoch": 1.6469038208168643,
-      "grad_norm": 20.141672134399414,
-      "learning_rate": 2.0118577075098816e-05,
-      "loss": 0.8604,
+      "grad_norm": 2.3696165084838867,
+      "learning_rate": 0.00033530961791831363,
+      "loss": 0.8615,
       "step": 8750
     },
     {
       "epoch": 1.6487859966120837,
-      "grad_norm": 19.226642608642578,
-      "learning_rate": 2.01072840203275e-05,
-      "loss": 0.8389,
+      "grad_norm": 2.571803092956543,
+      "learning_rate": 0.00033512140033879165,
+      "loss": 0.7667,
       "step": 8760
     },
     {
       "epoch": 1.6506681724073027,
-      "grad_norm": 12.022441864013672,
-      "learning_rate": 2.0095990965556182e-05,
-      "loss": 0.8879,
+      "grad_norm": 4.093145847320557,
+      "learning_rate": 0.0003349331827592697,
+      "loss": 0.7379,
       "step": 8770
     },
     {
       "epoch": 1.6525503482025221,
-      "grad_norm": 16.575210571289062,
-      "learning_rate": 2.0084697910784868e-05,
-      "loss": 0.9649,
+      "grad_norm": 4.968115329742432,
+      "learning_rate": 0.0003347449651797478,
+      "loss": 0.9293,
       "step": 8780
     },
     {
       "epoch": 1.6544325239977415,
-      "grad_norm": 13.532410621643066,
-      "learning_rate": 2.007340485601355e-05,
-      "loss": 1.1161,
+      "grad_norm": 4.6447248458862305,
+      "learning_rate": 0.0003345567476002259,
+      "loss": 1.1339,
       "step": 8790
     },
     {
       "epoch": 1.6563146997929605,
-      "grad_norm": 13.928875923156738,
-      "learning_rate": 2.0062111801242237e-05,
-      "loss": 0.57,
+      "grad_norm": 2.701174736022949,
+      "learning_rate": 0.00033436853002070396,
+      "loss": 0.6028,
       "step": 8800
     },
     {
       "epoch": 1.65819687558818,
-      "grad_norm": 9.75232982635498,
-      "learning_rate": 2.0050818746470923e-05,
-      "loss": 0.5475,
+      "grad_norm": 10.108606338500977,
+      "learning_rate": 0.00033418031244118203,
+      "loss": 0.593,
       "step": 8810
     },
     {
       "epoch": 1.6600790513833992,
-      "grad_norm": 19.574495315551758,
-      "learning_rate": 2.0039525691699605e-05,
-      "loss": 0.8012,
+      "grad_norm": 2.6389102935791016,
+      "learning_rate": 0.0003339920948616601,
+      "loss": 0.7667,
       "step": 8820
     },
     {
       "epoch": 1.6619612271786184,
-      "grad_norm": 8.311776161193848,
-      "learning_rate": 2.002823263692829e-05,
-      "loss": 0.8989,
+      "grad_norm": 8.076936721801758,
+      "learning_rate": 0.0003338038772821382,
+      "loss": 0.7453,
       "step": 8830
     },
     {
       "epoch": 1.6638434029738378,
-      "grad_norm": 3.7794203758239746,
-      "learning_rate": 2.0016939582156974e-05,
-      "loss": 0.7828,
+      "grad_norm": 0.3804866075515747,
+      "learning_rate": 0.00033361565970261626,
+      "loss": 0.822,
       "step": 8840
     },
     {
       "epoch": 1.665725578769057,
-      "grad_norm": 22.881847381591797,
-      "learning_rate": 2.0005646527385657e-05,
-      "loss": 1.0597,
+      "grad_norm": 4.877434253692627,
+      "learning_rate": 0.0003334274421230943,
+      "loss": 0.8363,
       "step": 8850
     },
     {
       "epoch": 1.6676077545642762,
-      "grad_norm": 23.629314422607422,
-      "learning_rate": 1.9994353472614343e-05,
-      "loss": 0.905,
+      "grad_norm": 2.456123113632202,
+      "learning_rate": 0.00033323922454357236,
+      "loss": 0.796,
       "step": 8860
     },
     {
       "epoch": 1.6694899303594957,
-      "grad_norm": 19.81004524230957,
-      "learning_rate": 1.9983060417843026e-05,
-      "loss": 1.0952,
+      "grad_norm": 4.350971698760986,
+      "learning_rate": 0.00033305100696405044,
+      "loss": 0.998,
       "step": 8870
     },
     {
       "epoch": 1.6713721061547149,
-      "grad_norm": 8.309386253356934,
-      "learning_rate": 1.9971767363071712e-05,
-      "loss": 0.7427,
+      "grad_norm": 2.929431200027466,
+      "learning_rate": 0.0003328627893845285,
+      "loss": 0.5895,
       "step": 8880
     },
     {
       "epoch": 1.673254281949934,
-      "grad_norm": 6.19067907333374,
-      "learning_rate": 1.9960474308300394e-05,
-      "loss": 0.7889,
+      "grad_norm": 0.23275358974933624,
+      "learning_rate": 0.0003326745718050066,
+      "loss": 0.6795,
       "step": 8890
     },
     {
       "epoch": 1.6751364577451535,
-      "grad_norm": 2.2768592834472656,
-      "learning_rate": 1.994918125352908e-05,
-      "loss": 0.8985,
+      "grad_norm": 2.660062074661255,
+      "learning_rate": 0.00033248635422548467,
+      "loss": 0.7639,
       "step": 8900
     },
     {
       "epoch": 1.6770186335403725,
-      "grad_norm": 7.979760646820068,
-      "learning_rate": 1.9937888198757767e-05,
-      "loss": 0.6913,
+      "grad_norm": 3.0600240230560303,
+      "learning_rate": 0.00033229813664596274,
+      "loss": 0.7491,
       "step": 8910
     },
     {
       "epoch": 1.678900809335592,
-      "grad_norm": 10.671245574951172,
-      "learning_rate": 1.992659514398645e-05,
-      "loss": 0.995,
+      "grad_norm": 2.6997058391571045,
+      "learning_rate": 0.0003321099190664408,
+      "loss": 0.9388,
       "step": 8920
     },
     {
       "epoch": 1.6807829851308114,
-      "grad_norm": 8.021512031555176,
-      "learning_rate": 1.9915302089215132e-05,
-      "loss": 1.1053,
+      "grad_norm": 3.960489511489868,
+      "learning_rate": 0.00033192170148691884,
+      "loss": 0.8138,
       "step": 8930
     },
     {
       "epoch": 1.6826651609260304,
-      "grad_norm": 10.151590347290039,
-      "learning_rate": 1.9904009034443818e-05,
-      "loss": 0.7479,
+      "grad_norm": 1.3196567296981812,
+      "learning_rate": 0.0003317334839073969,
+      "loss": 0.6019,
       "step": 8940
     },
     {
       "epoch": 1.6845473367212498,
-      "grad_norm": 15.058424949645996,
-      "learning_rate": 1.98927159796725e-05,
-      "loss": 0.9105,
+      "grad_norm": 3.450125217437744,
+      "learning_rate": 0.000331545266327875,
+      "loss": 0.7857,
       "step": 8950
     },
     {
       "epoch": 1.686429512516469,
-      "grad_norm": 0.7414102554321289,
-      "learning_rate": 1.9881422924901187e-05,
-      "loss": 0.7906,
+      "grad_norm": 3.65901517868042,
+      "learning_rate": 0.00033135704874835307,
+      "loss": 0.6392,
       "step": 8960
     },
     {
       "epoch": 1.6883116883116882,
-      "grad_norm": 16.738800048828125,
-      "learning_rate": 1.987012987012987e-05,
-      "loss": 0.891,
+      "grad_norm": 2.7813315391540527,
+      "learning_rate": 0.0003311688311688312,
+      "loss": 0.842,
       "step": 8970
     },
     {
       "epoch": 1.6901938641069076,
-      "grad_norm": 11.300554275512695,
-      "learning_rate": 1.9858836815358556e-05,
-      "loss": 0.849,
+      "grad_norm": 1.5750610828399658,
+      "learning_rate": 0.0003309806135893093,
+      "loss": 0.7625,
       "step": 8980
     },
     {
       "epoch": 1.6920760399021268,
-      "grad_norm": 23.922866821289062,
-      "learning_rate": 1.984754376058724e-05,
-      "loss": 1.182,
+      "grad_norm": 5.054027080535889,
+      "learning_rate": 0.00033079239600978735,
+      "loss": 0.9992,
       "step": 8990
     },
     {
       "epoch": 1.693958215697346,
-      "grad_norm": 15.910576820373535,
-      "learning_rate": 1.9836250705815924e-05,
-      "loss": 0.692,
+      "grad_norm": 1.384577751159668,
+      "learning_rate": 0.00033060417843026543,
+      "loss": 0.5368,
       "step": 9000
     },
     {
       "epoch": 1.6958403914925655,
-      "grad_norm": 12.029934883117676,
-      "learning_rate": 1.982495765104461e-05,
-      "loss": 0.8861,
+      "grad_norm": 2.0076184272766113,
+      "learning_rate": 0.0003304159608507435,
+      "loss": 0.8174,
       "step": 9010
     },
     {
       "epoch": 1.6977225672877847,
-      "grad_norm": 11.433530807495117,
-      "learning_rate": 1.981366459627329e-05,
-      "loss": 0.8564,
+      "grad_norm": 2.57498836517334,
+      "learning_rate": 0.00033022774327122153,
+      "loss": 0.7774,
       "step": 9020
     },
     {
       "epoch": 1.699604743083004,
-      "grad_norm": 16.016300201416016,
-      "learning_rate": 1.9802371541501976e-05,
-      "loss": 0.7998,
+      "grad_norm": 4.454178333282471,
+      "learning_rate": 0.0003300395256916996,
+      "loss": 0.6064,
       "step": 9030
     },
     {
       "epoch": 1.7014869188782233,
-      "grad_norm": 13.163691520690918,
-      "learning_rate": 1.9791078486730662e-05,
-      "loss": 0.8364,
+      "grad_norm": 3.815958023071289,
+      "learning_rate": 0.0003298513081121777,
+      "loss": 0.7997,
       "step": 9040
     },
     {
       "epoch": 1.7033690946734426,
-      "grad_norm": 15.584578514099121,
-      "learning_rate": 1.9779785431959345e-05,
-      "loss": 0.8937,
+      "grad_norm": 3.1478142738342285,
+      "learning_rate": 0.00032966309053265576,
+      "loss": 0.7746,
       "step": 9050
     },
     {
       "epoch": 1.7052512704686618,
-      "grad_norm": 34.573814392089844,
-      "learning_rate": 1.976849237718803e-05,
-      "loss": 0.4642,
+      "grad_norm": 2.721916675567627,
+      "learning_rate": 0.00032947487295313383,
+      "loss": 0.4329,
       "step": 9060
     },
     {
       "epoch": 1.7071334462638812,
-      "grad_norm": 3.945563316345215,
-      "learning_rate": 1.9757199322416713e-05,
-      "loss": 0.9052,
+      "grad_norm": 1.7750840187072754,
+      "learning_rate": 0.0003292866553736119,
+      "loss": 0.8113,
       "step": 9070
     },
     {
       "epoch": 1.7090156220591002,
-      "grad_norm": 14.851612091064453,
-      "learning_rate": 1.97459062676454e-05,
-      "loss": 0.766,
+      "grad_norm": 4.334325313568115,
+      "learning_rate": 0.00032909843779409,
+      "loss": 0.6708,
       "step": 9080
     },
     {
       "epoch": 1.7108977978543196,
-      "grad_norm": 20.84731101989746,
-      "learning_rate": 1.9734613212874085e-05,
-      "loss": 0.7886,
+      "grad_norm": 4.547921657562256,
+      "learning_rate": 0.00032891022021456806,
+      "loss": 0.6698,
       "step": 9090
     },
     {
       "epoch": 1.7127799736495388,
-      "grad_norm": 14.792206764221191,
-      "learning_rate": 1.9723320158102768e-05,
-      "loss": 0.8115,
+      "grad_norm": 5.556723117828369,
+      "learning_rate": 0.00032872200263504614,
+      "loss": 0.6937,
       "step": 9100
     },
     {
       "epoch": 1.714662149444758,
-      "grad_norm": 8.901041030883789,
-      "learning_rate": 1.971202710333145e-05,
-      "loss": 0.9397,
+      "grad_norm": 3.2558112144470215,
+      "learning_rate": 0.00032853378505552416,
+      "loss": 0.7803,
       "step": 9110
     },
     {
       "epoch": 1.7165443252399775,
-      "grad_norm": 8.225708961486816,
-      "learning_rate": 1.9700734048560134e-05,
-      "loss": 1.1384,
+      "grad_norm": 3.6377789974212646,
+      "learning_rate": 0.00032834556747600224,
+      "loss": 0.8596,
       "step": 9120
     },
     {
       "epoch": 1.7184265010351967,
-      "grad_norm": 5.585742473602295,
-      "learning_rate": 1.968944099378882e-05,
-      "loss": 0.794,
+      "grad_norm": 2.1207480430603027,
+      "learning_rate": 0.0003281573498964803,
+      "loss": 0.6312,
       "step": 9130
     },
     {
       "epoch": 1.7203086768304159,
-      "grad_norm": 4.999485969543457,
-      "learning_rate": 1.9678147939017506e-05,
-      "loss": 0.6178,
+      "grad_norm": 2.050334930419922,
+      "learning_rate": 0.0003279691323169584,
+      "loss": 0.6679,
       "step": 9140
     },
     {
       "epoch": 1.7221908526256353,
-      "grad_norm": 15.160903930664062,
-      "learning_rate": 1.966685488424619e-05,
-      "loss": 1.1031,
+      "grad_norm": 2.546337604522705,
+      "learning_rate": 0.00032778091473743647,
+      "loss": 0.9513,
       "step": 9150
     },
     {
       "epoch": 1.7240730284208545,
-      "grad_norm": 27.700048446655273,
-      "learning_rate": 1.9655561829474874e-05,
-      "loss": 0.9744,
+      "grad_norm": 3.854846715927124,
+      "learning_rate": 0.00032759269715791454,
+      "loss": 0.7858,
       "step": 9160
     },
     {
       "epoch": 1.7259552042160737,
-      "grad_norm": 15.794424057006836,
-      "learning_rate": 1.9644268774703557e-05,
-      "loss": 0.9782,
+      "grad_norm": 5.680496692657471,
+      "learning_rate": 0.0003274044795783926,
+      "loss": 1.0537,
       "step": 9170
     },
     {
       "epoch": 1.7278373800112932,
-      "grad_norm": 13.252302169799805,
-      "learning_rate": 1.9632975719932243e-05,
-      "loss": 0.9227,
+      "grad_norm": 2.7812232971191406,
+      "learning_rate": 0.0003272162619988707,
+      "loss": 0.8572,
       "step": 9180
     },
     {
       "epoch": 1.7297195558065124,
-      "grad_norm": 19.900714874267578,
-      "learning_rate": 1.9621682665160926e-05,
-      "loss": 0.7203,
+      "grad_norm": 2.869856357574463,
+      "learning_rate": 0.0003270280444193488,
+      "loss": 0.7442,
       "step": 9190
     },
     {
       "epoch": 1.7316017316017316,
-      "grad_norm": 8.167475700378418,
-      "learning_rate": 1.961038961038961e-05,
-      "loss": 0.9278,
+      "grad_norm": 1.521945595741272,
+      "learning_rate": 0.00032683982683982685,
+      "loss": 0.8358,
       "step": 9200
     },
     {
       "epoch": 1.733483907396951,
-      "grad_norm": 18.376384735107422,
-      "learning_rate": 1.9599096555618295e-05,
-      "loss": 1.005,
+      "grad_norm": 2.468168258666992,
+      "learning_rate": 0.0003266516092603049,
+      "loss": 0.9282,
       "step": 9210
     },
     {
       "epoch": 1.73536608319217,
-      "grad_norm": 12.970340728759766,
-      "learning_rate": 1.958780350084698e-05,
-      "loss": 0.7519,
+      "grad_norm": 1.156663417816162,
+      "learning_rate": 0.000326463391680783,
+      "loss": 0.6668,
       "step": 9220
     },
     {
       "epoch": 1.7372482589873894,
-      "grad_norm": 10.988688468933105,
-      "learning_rate": 1.9576510446075663e-05,
-      "loss": 0.8544,
+      "grad_norm": 2.1691527366638184,
+      "learning_rate": 0.0003262751741012611,
+      "loss": 0.8898,
       "step": 9230
     },
     {
       "epoch": 1.7391304347826086,
-      "grad_norm": 7.907069206237793,
-      "learning_rate": 1.956521739130435e-05,
-      "loss": 1.2205,
+      "grad_norm": 2.910538911819458,
+      "learning_rate": 0.00032608695652173916,
+      "loss": 0.8819,
       "step": 9240
     },
     {
       "epoch": 1.7410126105778279,
-      "grad_norm": 24.321718215942383,
-      "learning_rate": 1.9553924336533032e-05,
-      "loss": 0.7836,
+      "grad_norm": 5.341407299041748,
+      "learning_rate": 0.00032589873894221723,
+      "loss": 0.7117,
       "step": 9250
     },
     {
       "epoch": 1.7428947863730473,
-      "grad_norm": 11.95626449584961,
-      "learning_rate": 1.9542631281761718e-05,
-      "loss": 0.7509,
+      "grad_norm": 2.8497936725616455,
+      "learning_rate": 0.0003257105213626953,
+      "loss": 0.5795,
       "step": 9260
     },
     {
       "epoch": 1.7447769621682665,
-      "grad_norm": 1.1062142848968506,
-      "learning_rate": 1.9531338226990404e-05,
-      "loss": 0.5899,
+      "grad_norm": 0.632068932056427,
+      "learning_rate": 0.0003255223037831734,
+      "loss": 0.5179,
       "step": 9270
     },
     {
       "epoch": 1.7466591379634857,
-      "grad_norm": 12.256535530090332,
-      "learning_rate": 1.9520045172219084e-05,
-      "loss": 0.7338,
+      "grad_norm": 8.0675630569458,
+      "learning_rate": 0.0003253340862036514,
+      "loss": 0.5875,
       "step": 9280
     },
     {
       "epoch": 1.7485413137587051,
-      "grad_norm": 28.20414924621582,
-      "learning_rate": 1.950875211744777e-05,
-      "loss": 0.848,
+      "grad_norm": 7.597952365875244,
+      "learning_rate": 0.0003251458686241295,
+      "loss": 0.7736,
       "step": 9290
     },
     {
       "epoch": 1.7504234895539243,
-      "grad_norm": 10.83683967590332,
-      "learning_rate": 1.9497459062676452e-05,
-      "loss": 1.0118,
+      "grad_norm": 4.242430210113525,
+      "learning_rate": 0.00032495765104460756,
+      "loss": 0.9301,
       "step": 9300
     },
     {
       "epoch": 1.7523056653491436,
-      "grad_norm": 12.281524658203125,
-      "learning_rate": 1.948616600790514e-05,
-      "loss": 0.7269,
+      "grad_norm": 0.9840834736824036,
+      "learning_rate": 0.00032476943346508564,
+      "loss": 0.5589,
       "step": 9310
     },
     {
       "epoch": 1.754187841144363,
-      "grad_norm": 3.447277545928955,
-      "learning_rate": 1.9474872953133825e-05,
-      "loss": 0.6222,
+      "grad_norm": 2.141883134841919,
+      "learning_rate": 0.0003245812158855637,
+      "loss": 0.7073,
       "step": 9320
     },
     {
       "epoch": 1.7560700169395822,
-      "grad_norm": 21.582672119140625,
-      "learning_rate": 1.9463579898362507e-05,
-      "loss": 0.9156,
+      "grad_norm": 3.3031342029571533,
+      "learning_rate": 0.0003243929983060418,
+      "loss": 0.7915,
       "step": 9330
     },
     {
       "epoch": 1.7579521927348014,
-      "grad_norm": 8.327972412109375,
-      "learning_rate": 1.9452286843591193e-05,
-      "loss": 0.7264,
+      "grad_norm": 3.9373908042907715,
+      "learning_rate": 0.00032420478072651987,
+      "loss": 0.5134,
       "step": 9340
     },
     {
       "epoch": 1.7598343685300208,
-      "grad_norm": 9.982987403869629,
-      "learning_rate": 1.9440993788819876e-05,
-      "loss": 0.5234,
+      "grad_norm": 3.2525720596313477,
+      "learning_rate": 0.00032401656314699794,
+      "loss": 0.4895,
       "step": 9350
     },
     {
       "epoch": 1.7617165443252398,
-      "grad_norm": 18.895675659179688,
-      "learning_rate": 1.9429700734048562e-05,
-      "loss": 0.6,
+      "grad_norm": 3.599588394165039,
+      "learning_rate": 0.000323828345567476,
+      "loss": 0.5655,
       "step": 9360
     },
     {
       "epoch": 1.7635987201204593,
-      "grad_norm": 3.999253749847412,
-      "learning_rate": 1.9418407679277245e-05,
-      "loss": 0.8148,
+      "grad_norm": 1.8969835042953491,
+      "learning_rate": 0.00032364012798795404,
+      "loss": 0.6763,
       "step": 9370
     },
     {
       "epoch": 1.7654808959156785,
-      "grad_norm": 7.097409725189209,
-      "learning_rate": 1.9407114624505927e-05,
-      "loss": 0.6838,
+      "grad_norm": 4.947686195373535,
+      "learning_rate": 0.0003234519104084321,
+      "loss": 0.6181,
       "step": 9380
     },
     {
       "epoch": 1.7673630717108977,
-      "grad_norm": 10.425472259521484,
-      "learning_rate": 1.9395821569734614e-05,
-      "loss": 0.7291,
+      "grad_norm": 1.9779622554779053,
+      "learning_rate": 0.0003232636928289102,
+      "loss": 0.7254,
       "step": 9390
     },
     {
       "epoch": 1.7692452475061171,
-      "grad_norm": 15.1722993850708,
-      "learning_rate": 1.9384528514963296e-05,
-      "loss": 0.7761,
+      "grad_norm": 2.9042108058929443,
+      "learning_rate": 0.00032307547524938827,
+      "loss": 0.6338,
       "step": 9400
     },
     {
       "epoch": 1.7711274233013363,
-      "grad_norm": 11.412665367126465,
-      "learning_rate": 1.9373235460191982e-05,
-      "loss": 0.904,
+      "grad_norm": 3.2115519046783447,
+      "learning_rate": 0.0003228872576698664,
+      "loss": 0.8777,
       "step": 9410
     },
     {
       "epoch": 1.7730095990965555,
-      "grad_norm": 19.17182159423828,
-      "learning_rate": 1.936194240542067e-05,
-      "loss": 0.9217,
+      "grad_norm": 3.146235466003418,
+      "learning_rate": 0.0003226990400903445,
+      "loss": 0.7832,
       "step": 9420
     },
     {
       "epoch": 1.774891774891775,
-      "grad_norm": 4.5345845222473145,
-      "learning_rate": 1.935064935064935e-05,
-      "loss": 0.6485,
+      "grad_norm": 2.5843141078948975,
+      "learning_rate": 0.00032251082251082255,
+      "loss": 0.4189,
       "step": 9430
     },
     {
       "epoch": 1.7767739506869942,
-      "grad_norm": 24.256860733032227,
-      "learning_rate": 1.9339356295878037e-05,
-      "loss": 1.0271,
+      "grad_norm": 5.2682271003723145,
+      "learning_rate": 0.00032232260493130063,
+      "loss": 0.9093,
       "step": 9440
     },
     {
       "epoch": 1.7786561264822134,
-      "grad_norm": 13.506421089172363,
-      "learning_rate": 1.932806324110672e-05,
-      "loss": 0.9675,
+      "grad_norm": 2.954876184463501,
+      "learning_rate": 0.0003221343873517787,
+      "loss": 0.7608,
       "step": 9450
     },
     {
       "epoch": 1.7805383022774328,
-      "grad_norm": 20.032943725585938,
-      "learning_rate": 1.9316770186335403e-05,
-      "loss": 0.8068,
+      "grad_norm": 1.4374929666519165,
+      "learning_rate": 0.00032194616977225673,
+      "loss": 0.629,
       "step": 9460
     },
     {
       "epoch": 1.782420478072652,
-      "grad_norm": 30.704687118530273,
-      "learning_rate": 1.930547713156409e-05,
-      "loss": 0.4575,
+      "grad_norm": 0.3903244137763977,
+      "learning_rate": 0.0003217579521927348,
+      "loss": 0.4012,
       "step": 9470
     },
     {
       "epoch": 1.7843026538678712,
-      "grad_norm": 29.728713989257812,
-      "learning_rate": 1.929418407679277e-05,
-      "loss": 0.9168,
+      "grad_norm": 2.7502241134643555,
+      "learning_rate": 0.0003215697346132129,
+      "loss": 0.7284,
       "step": 9480
     },
     {
       "epoch": 1.7861848296630907,
-      "grad_norm": 21.02536964416504,
-      "learning_rate": 1.9282891022021457e-05,
-      "loss": 1.0197,
+      "grad_norm": 4.083826065063477,
+      "learning_rate": 0.00032138151703369096,
+      "loss": 0.9896,
       "step": 9490
     },
     {
       "epoch": 1.7880670054583097,
-      "grad_norm": 0.558627724647522,
-      "learning_rate": 1.9271597967250143e-05,
-      "loss": 0.7217,
+      "grad_norm": 0.2609062194824219,
+      "learning_rate": 0.00032119329945416903,
+      "loss": 0.6251,
       "step": 9500
     },
     {
       "epoch": 1.789949181253529,
-      "grad_norm": 18.84478759765625,
-      "learning_rate": 1.9260304912478826e-05,
-      "loss": 0.7922,
+      "grad_norm": 3.0656192302703857,
+      "learning_rate": 0.0003210050818746471,
+      "loss": 0.5306,
       "step": 9510
     },
     {
       "epoch": 1.7918313570487485,
-      "grad_norm": 17.696800231933594,
-      "learning_rate": 1.9249011857707512e-05,
-      "loss": 1.0852,
+      "grad_norm": 5.871590614318848,
+      "learning_rate": 0.0003208168642951252,
+      "loss": 0.9829,
       "step": 9520
     },
     {
       "epoch": 1.7937135328439675,
-      "grad_norm": 20.650711059570312,
-      "learning_rate": 1.9237718802936195e-05,
-      "loss": 0.7487,
+      "grad_norm": 2.4569530487060547,
+      "learning_rate": 0.00032062864671560326,
+      "loss": 0.7821,
       "step": 9530
     },
     {
       "epoch": 1.795595708639187,
-      "grad_norm": 11.20978832244873,
-      "learning_rate": 1.9226425748164878e-05,
-      "loss": 1.1077,
+      "grad_norm": 2.2387983798980713,
+      "learning_rate": 0.0003204404291360813,
+      "loss": 0.8974,
       "step": 9540
     },
     {
       "epoch": 1.7974778844344061,
-      "grad_norm": 9.016942977905273,
-      "learning_rate": 1.9215132693393564e-05,
-      "loss": 0.7859,
+      "grad_norm": 1.3203648328781128,
+      "learning_rate": 0.00032025221155655936,
+      "loss": 0.7528,
       "step": 9550
     },
     {
       "epoch": 1.7993600602296254,
-      "grad_norm": 18.516727447509766,
-      "learning_rate": 1.9203839638622246e-05,
-      "loss": 0.6019,
+      "grad_norm": 3.0121521949768066,
+      "learning_rate": 0.00032006399397703744,
+      "loss": 0.6661,
       "step": 9560
     },
     {
       "epoch": 1.8012422360248448,
-      "grad_norm": 14.529791831970215,
-      "learning_rate": 1.9192546583850932e-05,
-      "loss": 0.8048,
+      "grad_norm": 3.89560866355896,
+      "learning_rate": 0.0003198757763975155,
+      "loss": 0.6831,
       "step": 9570
     },
     {
       "epoch": 1.803124411820064,
-      "grad_norm": 13.547689437866211,
-      "learning_rate": 1.9181253529079615e-05,
-      "loss": 1.0294,
+      "grad_norm": 2.6359190940856934,
+      "learning_rate": 0.0003196875588179936,
+      "loss": 0.9127,
       "step": 9580
     },
     {
       "epoch": 1.8050065876152832,
-      "grad_norm": 11.232539176940918,
-      "learning_rate": 1.91699604743083e-05,
-      "loss": 0.798,
+      "grad_norm": 3.913572072982788,
+      "learning_rate": 0.00031949934123847167,
+      "loss": 0.7298,
       "step": 9590
     },
     {
       "epoch": 1.8068887634105026,
-      "grad_norm": 13.629487037658691,
-      "learning_rate": 1.9158667419536987e-05,
-      "loss": 0.6611,
+      "grad_norm": 2.637514591217041,
+      "learning_rate": 0.00031931112365894974,
+      "loss": 0.6781,
       "step": 9600
     },
     {
       "epoch": 1.8087709392057219,
-      "grad_norm": 9.937800407409668,
-      "learning_rate": 1.914737436476567e-05,
-      "loss": 0.5109,
+      "grad_norm": 3.258054494857788,
+      "learning_rate": 0.0003191229060794278,
+      "loss": 0.4907,
       "step": 9610
     },
     {
       "epoch": 1.810653115000941,
-      "grad_norm": 14.390426635742188,
-      "learning_rate": 1.9136081309994356e-05,
-      "loss": 0.8687,
+      "grad_norm": 2.188828468322754,
+      "learning_rate": 0.0003189346884999059,
+      "loss": 0.9083,
       "step": 9620
     },
     {
       "epoch": 1.8125352907961605,
-      "grad_norm": 21.730375289916992,
-      "learning_rate": 1.9124788255223035e-05,
-      "loss": 1.0062,
+      "grad_norm": 6.632510185241699,
+      "learning_rate": 0.0003187464709203839,
+      "loss": 0.834,
       "step": 9630
     },
     {
       "epoch": 1.8144174665913795,
-      "grad_norm": 17.650148391723633,
-      "learning_rate": 1.911349520045172e-05,
-      "loss": 0.7727,
+      "grad_norm": 1.7166807651519775,
+      "learning_rate": 0.00031855825334086205,
+      "loss": 0.5749,
       "step": 9640
     },
     {
       "epoch": 1.816299642386599,
-      "grad_norm": 13.232230186462402,
-      "learning_rate": 1.9102202145680407e-05,
-      "loss": 0.987,
+      "grad_norm": 3.0382962226867676,
+      "learning_rate": 0.0003183700357613401,
+      "loss": 0.8518,
       "step": 9650
     },
     {
       "epoch": 1.8181818181818183,
-      "grad_norm": 21.150760650634766,
-      "learning_rate": 1.909090909090909e-05,
-      "loss": 1.2057,
+      "grad_norm": 2.7788681983947754,
+      "learning_rate": 0.0003181818181818182,
+      "loss": 0.9275,
       "step": 9660
     },
     {
       "epoch": 1.8200639939770373,
-      "grad_norm": 15.0765962600708,
-      "learning_rate": 1.9079616036137776e-05,
-      "loss": 0.9163,
+      "grad_norm": 2.9813215732574463,
+      "learning_rate": 0.0003179936006022963,
+      "loss": 0.9127,
       "step": 9670
     },
     {
       "epoch": 1.8219461697722568,
-      "grad_norm": 16.338342666625977,
-      "learning_rate": 1.906832298136646e-05,
-      "loss": 0.6045,
+      "grad_norm": 4.174839496612549,
+      "learning_rate": 0.00031780538302277436,
+      "loss": 0.5552,
       "step": 9680
     },
     {
       "epoch": 1.823828345567476,
-      "grad_norm": 28.87519073486328,
-      "learning_rate": 1.9057029926595145e-05,
-      "loss": 1.0076,
+      "grad_norm": 4.95925235748291,
+      "learning_rate": 0.00031761716544325243,
+      "loss": 0.8057,
       "step": 9690
     },
     {
       "epoch": 1.8257105213626952,
-      "grad_norm": 20.42050552368164,
-      "learning_rate": 1.904573687182383e-05,
-      "loss": 1.0798,
+      "grad_norm": 5.213214874267578,
+      "learning_rate": 0.0003174289478637305,
+      "loss": 0.8951,
       "step": 9700
     },
     {
       "epoch": 1.8275926971579146,
-      "grad_norm": 14.828946113586426,
-      "learning_rate": 1.9034443817052514e-05,
-      "loss": 1.1065,
+      "grad_norm": 4.759930610656738,
+      "learning_rate": 0.0003172407302842086,
+      "loss": 0.9093,
       "step": 9710
     },
     {
       "epoch": 1.8294748729531338,
-      "grad_norm": 17.2311954498291,
-      "learning_rate": 1.9023150762281196e-05,
-      "loss": 0.8212,
+      "grad_norm": 2.821655035018921,
+      "learning_rate": 0.0003170525127046866,
+      "loss": 0.8012,
       "step": 9720
     },
     {
       "epoch": 1.831357048748353,
-      "grad_norm": 17.343181610107422,
-      "learning_rate": 1.9011857707509883e-05,
-      "loss": 1.0861,
+      "grad_norm": 0.5296029448509216,
+      "learning_rate": 0.0003168642951251647,
+      "loss": 0.9112,
       "step": 9730
     },
     {
       "epoch": 1.8332392245435725,
-      "grad_norm": 15.251105308532715,
-      "learning_rate": 1.9000564652738565e-05,
-      "loss": 0.8787,
+      "grad_norm": 4.0418701171875,
+      "learning_rate": 0.00031667607754564276,
+      "loss": 1.006,
       "step": 9740
     },
     {
       "epoch": 1.8351214003387917,
-      "grad_norm": 7.921949863433838,
-      "learning_rate": 1.898927159796725e-05,
-      "loss": 1.5269,
+      "grad_norm": 2.679586410522461,
+      "learning_rate": 0.00031648785996612084,
+      "loss": 1.1059,
       "step": 9750
     },
     {
       "epoch": 1.8370035761340109,
-      "grad_norm": 11.668200492858887,
-      "learning_rate": 1.8977978543195934e-05,
-      "loss": 0.7354,
+      "grad_norm": 1.4031389951705933,
+      "learning_rate": 0.0003162996423865989,
+      "loss": 0.5594,
       "step": 9760
     },
     {
       "epoch": 1.8388857519292303,
-      "grad_norm": 27.10658073425293,
-      "learning_rate": 1.896668548842462e-05,
-      "loss": 0.8172,
+      "grad_norm": 2.9446029663085938,
+      "learning_rate": 0.000316111424807077,
+      "loss": 0.7634,
       "step": 9770
     },
     {
       "epoch": 1.8407679277244493,
-      "grad_norm": 16.760330200195312,
-      "learning_rate": 1.8955392433653306e-05,
-      "loss": 0.8198,
+      "grad_norm": 2.208606243133545,
+      "learning_rate": 0.00031592320722755507,
+      "loss": 0.7794,
       "step": 9780
     },
     {
       "epoch": 1.8426501035196687,
-      "grad_norm": 16.54085350036621,
-      "learning_rate": 1.894409937888199e-05,
-      "loss": 0.8366,
+      "grad_norm": 4.726104259490967,
+      "learning_rate": 0.00031573498964803314,
+      "loss": 0.7705,
       "step": 9790
     },
     {
       "epoch": 1.8445322793148882,
-      "grad_norm": 12.325793266296387,
-      "learning_rate": 1.893280632411067e-05,
-      "loss": 0.7522,
+      "grad_norm": 4.122304916381836,
+      "learning_rate": 0.00031554677206851116,
+      "loss": 0.7017,
       "step": 9800
     },
     {
       "epoch": 1.8464144551101072,
-      "grad_norm": 21.898651123046875,
-      "learning_rate": 1.8921513269339354e-05,
-      "loss": 0.7455,
+      "grad_norm": 5.453103542327881,
+      "learning_rate": 0.00031535855448898924,
+      "loss": 0.7014,
       "step": 9810
     },
     {
       "epoch": 1.8482966309053266,
-      "grad_norm": 8.566146850585938,
-      "learning_rate": 1.891022021456804e-05,
-      "loss": 0.7789,
+      "grad_norm": 2.251783609390259,
+      "learning_rate": 0.0003151703369094673,
+      "loss": 0.7531,
       "step": 9820
     },
     {
       "epoch": 1.8501788067005458,
-      "grad_norm": 13.785645484924316,
-      "learning_rate": 1.8898927159796726e-05,
-      "loss": 0.7233,
+      "grad_norm": 3.082606792449951,
+      "learning_rate": 0.0003149821193299454,
+      "loss": 0.6417,
       "step": 9830
     },
     {
       "epoch": 1.852060982495765,
-      "grad_norm": 15.743664741516113,
-      "learning_rate": 1.888763410502541e-05,
-      "loss": 0.9657,
+      "grad_norm": 3.380202293395996,
+      "learning_rate": 0.00031479390175042347,
+      "loss": 0.9505,
       "step": 9840
     },
     {
       "epoch": 1.8539431582909844,
-      "grad_norm": 32.00886535644531,
-      "learning_rate": 1.8876341050254095e-05,
-      "loss": 0.7379,
+      "grad_norm": 5.193691253662109,
+      "learning_rate": 0.00031460568417090155,
+      "loss": 0.763,
       "step": 9850
     },
     {
       "epoch": 1.8558253340862036,
-      "grad_norm": 7.174389839172363,
-      "learning_rate": 1.8865047995482778e-05,
-      "loss": 0.8011,
+      "grad_norm": 2.495882511138916,
+      "learning_rate": 0.0003144174665913797,
+      "loss": 0.7655,
       "step": 9860
     },
     {
       "epoch": 1.8577075098814229,
-      "grad_norm": 20.806581497192383,
-      "learning_rate": 1.8853754940711464e-05,
-      "loss": 0.8431,
+      "grad_norm": 2.836336374282837,
+      "learning_rate": 0.00031422924901185775,
+      "loss": 0.8337,
       "step": 9870
     },
     {
       "epoch": 1.8595896856766423,
-      "grad_norm": 21.621334075927734,
-      "learning_rate": 1.884246188594015e-05,
-      "loss": 0.9357,
+      "grad_norm": 1.8456288576126099,
+      "learning_rate": 0.00031404103143233583,
+      "loss": 0.8305,
       "step": 9880
     },
     {
       "epoch": 1.8614718614718615,
-      "grad_norm": 4.900033473968506,
-      "learning_rate": 1.883116883116883e-05,
-      "loss": 1.0361,
+      "grad_norm": 1.920587420463562,
+      "learning_rate": 0.00031385281385281385,
+      "loss": 0.8039,
       "step": 9890
     },
     {
       "epoch": 1.8633540372670807,
-      "grad_norm": 19.732091903686523,
-      "learning_rate": 1.8819875776397515e-05,
-      "loss": 0.6999,
+      "grad_norm": 5.574342727661133,
+      "learning_rate": 0.00031366459627329193,
+      "loss": 0.6394,
       "step": 9900
     },
     {
       "epoch": 1.8652362130623001,
-      "grad_norm": 41.449031829833984,
-      "learning_rate": 1.8808582721626198e-05,
-      "loss": 0.9272,
+      "grad_norm": 1.785643458366394,
+      "learning_rate": 0.00031347637869377,
+      "loss": 1.0156,
       "step": 9910
     },
     {
       "epoch": 1.8671183888575191,
-      "grad_norm": 21.400299072265625,
-      "learning_rate": 1.8797289666854884e-05,
-      "loss": 0.9431,
+      "grad_norm": 3.011502742767334,
+      "learning_rate": 0.0003132881611142481,
+      "loss": 0.9452,
       "step": 9920
     },
     {
       "epoch": 1.8690005646527386,
-      "grad_norm": 17.52521514892578,
-      "learning_rate": 1.878599661208357e-05,
-      "loss": 0.6384,
+      "grad_norm": 3.422855854034424,
+      "learning_rate": 0.00031309994353472616,
+      "loss": 0.4303,
       "step": 9930
     },
     {
       "epoch": 1.870882740447958,
-      "grad_norm": 12.966445922851562,
-      "learning_rate": 1.8774703557312253e-05,
-      "loss": 0.899,
+      "grad_norm": 4.045591354370117,
+      "learning_rate": 0.00031291172595520423,
+      "loss": 0.9059,
       "step": 9940
     },
     {
       "epoch": 1.872764916243177,
-      "grad_norm": 23.314476013183594,
-      "learning_rate": 1.876341050254094e-05,
-      "loss": 0.7705,
+      "grad_norm": 4.1779937744140625,
+      "learning_rate": 0.0003127235083756823,
+      "loss": 0.7193,
       "step": 9950
     },
     {
       "epoch": 1.8746470920383964,
-      "grad_norm": 14.737226486206055,
-      "learning_rate": 1.8752117447769625e-05,
-      "loss": 0.7806,
+      "grad_norm": 5.253387451171875,
+      "learning_rate": 0.0003125352907961604,
+      "loss": 0.7338,
       "step": 9960
     },
     {
       "epoch": 1.8765292678336156,
-      "grad_norm": 9.335084915161133,
-      "learning_rate": 1.8740824392998308e-05,
-      "loss": 0.8484,
+      "grad_norm": 1.7977385520935059,
+      "learning_rate": 0.00031234707321663846,
+      "loss": 0.7472,
       "step": 9970
     },
     {
       "epoch": 1.8784114436288348,
-      "grad_norm": 5.914954662322998,
-      "learning_rate": 1.872953133822699e-05,
-      "loss": 0.9675,
+      "grad_norm": 0.8867546319961548,
+      "learning_rate": 0.0003121588556371165,
+      "loss": 0.8371,
       "step": 9980
     },
     {
       "epoch": 1.8802936194240543,
-      "grad_norm": 13.312799453735352,
-      "learning_rate": 1.8718238283455673e-05,
-      "loss": 0.7423,
+      "grad_norm": 1.7153702974319458,
+      "learning_rate": 0.00031197063805759456,
+      "loss": 0.6443,
       "step": 9990
     },
     {
       "epoch": 1.8821757952192735,
-      "grad_norm": 10.06735610961914,
-      "learning_rate": 1.870694522868436e-05,
-      "loss": 1.0082,
+      "grad_norm": 0.5608319044113159,
+      "learning_rate": 0.00031178242047807264,
+      "loss": 0.7286,
       "step": 10000
     },
     {
       "epoch": 1.8840579710144927,
-      "grad_norm": 13.502887725830078,
-      "learning_rate": 1.8695652173913045e-05,
-      "loss": 0.7395,
+      "grad_norm": 1.7437026500701904,
+      "learning_rate": 0.0003115942028985507,
+      "loss": 0.6849,
       "step": 10010
     },
     {
       "epoch": 1.8859401468097121,
-      "grad_norm": 19.229324340820312,
-      "learning_rate": 1.8684359119141728e-05,
-      "loss": 1.0014,
+      "grad_norm": 3.591338872909546,
+      "learning_rate": 0.0003114059853190288,
+      "loss": 0.8433,
       "step": 10020
     },
     {
       "epoch": 1.8878223226049313,
-      "grad_norm": 15.201828002929688,
-      "learning_rate": 1.8673066064370414e-05,
-      "loss": 0.8707,
+      "grad_norm": 3.8551323413848877,
+      "learning_rate": 0.00031121776773950687,
+      "loss": 0.8481,
       "step": 10030
     },
     {
       "epoch": 1.8897044984001505,
-      "grad_norm": 28.766010284423828,
-      "learning_rate": 1.8661773009599097e-05,
-      "loss": 0.7835,
+      "grad_norm": 2.6424741744995117,
+      "learning_rate": 0.00031102955015998494,
+      "loss": 0.6314,
       "step": 10040
     },
     {
       "epoch": 1.89158667419537,
-      "grad_norm": 2.099621295928955,
-      "learning_rate": 1.8650479954827783e-05,
-      "loss": 0.8257,
+      "grad_norm": 2.6313867568969727,
+      "learning_rate": 0.000310841332580463,
+      "loss": 0.7825,
       "step": 10050
     },
     {
       "epoch": 1.8934688499905892,
-      "grad_norm": 11.689505577087402,
-      "learning_rate": 1.8639186900056465e-05,
-      "loss": 0.8311,
+      "grad_norm": 1.8866409063339233,
+      "learning_rate": 0.00031065311500094104,
+      "loss": 0.6601,
       "step": 10060
     },
     {
       "epoch": 1.8953510257858084,
-      "grad_norm": 1.5185855627059937,
-      "learning_rate": 1.8627893845285148e-05,
-      "loss": 0.7473,
+      "grad_norm": 0.6496089100837708,
+      "learning_rate": 0.0003104648974214191,
+      "loss": 0.7797,
       "step": 10070
     },
     {
       "epoch": 1.8972332015810278,
-      "grad_norm": 16.93794822692871,
-      "learning_rate": 1.8616600790513834e-05,
-      "loss": 0.8576,
+      "grad_norm": 3.582031726837158,
+      "learning_rate": 0.00031027667984189725,
+      "loss": 0.8081,
       "step": 10080
     },
     {
       "epoch": 1.8991153773762468,
-      "grad_norm": 12.918668746948242,
-      "learning_rate": 1.8605307735742517e-05,
-      "loss": 0.6406,
+      "grad_norm": 3.5374810695648193,
+      "learning_rate": 0.0003100884622623753,
+      "loss": 0.6228,
       "step": 10090
     },
     {
       "epoch": 1.9009975531714662,
-      "grad_norm": 38.772605895996094,
-      "learning_rate": 1.8594014680971203e-05,
-      "loss": 1.0629,
+      "grad_norm": 2.6551787853240967,
+      "learning_rate": 0.0003099002446828534,
+      "loss": 0.9884,
       "step": 10100
     },
     {
       "epoch": 1.9028797289666854,
-      "grad_norm": 8.743882179260254,
-      "learning_rate": 1.858272162619989e-05,
-      "loss": 0.8832,
+      "grad_norm": 2.751276969909668,
+      "learning_rate": 0.0003097120271033315,
+      "loss": 0.6363,
       "step": 10110
     },
     {
       "epoch": 1.9047619047619047,
-      "grad_norm": 18.133054733276367,
-      "learning_rate": 1.8571428571428572e-05,
-      "loss": 0.5021,
+      "grad_norm": 0.6503388285636902,
+      "learning_rate": 0.00030952380952380956,
+      "loss": 0.5765,
       "step": 10120
     },
     {
       "epoch": 1.906644080557124,
-      "grad_norm": 18.592235565185547,
-      "learning_rate": 1.8560135516657258e-05,
-      "loss": 0.6759,
+      "grad_norm": 1.9065133333206177,
+      "learning_rate": 0.00030933559194428763,
+      "loss": 0.6963,
       "step": 10130
     },
     {
       "epoch": 1.9085262563523433,
-      "grad_norm": 18.375886917114258,
-      "learning_rate": 1.854884246188594e-05,
-      "loss": 0.5934,
+      "grad_norm": 2.041531801223755,
+      "learning_rate": 0.0003091473743647657,
+      "loss": 0.4221,
       "step": 10140
     },
     {
       "epoch": 1.9104084321475625,
-      "grad_norm": 9.453605651855469,
-      "learning_rate": 1.8537549407114623e-05,
-      "loss": 0.8925,
+      "grad_norm": 1.8870247602462769,
+      "learning_rate": 0.00030895915678524373,
+      "loss": 0.7617,
       "step": 10150
     },
     {
       "epoch": 1.912290607942782,
-      "grad_norm": 20.740285873413086,
-      "learning_rate": 1.852625635234331e-05,
-      "loss": 0.931,
+      "grad_norm": 3.5730133056640625,
+      "learning_rate": 0.0003087709392057218,
+      "loss": 0.9216,
       "step": 10160
     },
     {
       "epoch": 1.9141727837380011,
-      "grad_norm": 21.86232566833496,
-      "learning_rate": 1.8514963297571992e-05,
-      "loss": 1.0436,
+      "grad_norm": 3.7722654342651367,
+      "learning_rate": 0.0003085827216261999,
+      "loss": 0.9224,
       "step": 10170
     },
     {
       "epoch": 1.9160549595332204,
-      "grad_norm": 31.541776657104492,
-      "learning_rate": 1.8503670242800678e-05,
-      "loss": 0.8187,
+      "grad_norm": 3.3979384899139404,
+      "learning_rate": 0.00030839450404667796,
+      "loss": 0.894,
       "step": 10180
     },
     {
       "epoch": 1.9179371353284398,
-      "grad_norm": 3.2206337451934814,
-      "learning_rate": 1.849237718802936e-05,
-      "loss": 0.6882,
+      "grad_norm": 1.5552890300750732,
+      "learning_rate": 0.00030820628646715604,
+      "loss": 0.6415,
       "step": 10190
     },
     {
       "epoch": 1.919819311123659,
-      "grad_norm": 1.3856621980667114,
-      "learning_rate": 1.8481084133258047e-05,
-      "loss": 0.6564,
+      "grad_norm": 0.8526018857955933,
+      "learning_rate": 0.0003080180688876341,
+      "loss": 0.5772,
       "step": 10200
     },
     {
       "epoch": 1.9217014869188782,
-      "grad_norm": 45.16568374633789,
-      "learning_rate": 1.8469791078486733e-05,
-      "loss": 0.7004,
+      "grad_norm": 2.919462203979492,
+      "learning_rate": 0.0003078298513081122,
+      "loss": 0.5405,
       "step": 10210
     },
     {
       "epoch": 1.9235836627140976,
-      "grad_norm": 12.040491104125977,
-      "learning_rate": 1.8458498023715416e-05,
-      "loss": 0.8935,
+      "grad_norm": 2.8013722896575928,
+      "learning_rate": 0.00030764163372859027,
+      "loss": 1.0859,
       "step": 10220
     },
     {
       "epoch": 1.9254658385093166,
-      "grad_norm": 1.0651041269302368,
-      "learning_rate": 1.84472049689441e-05,
-      "loss": 0.9316,
+      "grad_norm": 0.8399747610092163,
+      "learning_rate": 0.00030745341614906834,
+      "loss": 0.9671,
       "step": 10230
     },
     {
       "epoch": 1.927348014304536,
-      "grad_norm": 24.65813636779785,
-      "learning_rate": 1.8435911914172784e-05,
-      "loss": 1.1372,
+      "grad_norm": 3.0836198329925537,
+      "learning_rate": 0.00030726519856954636,
+      "loss": 1.083,
       "step": 10240
     },
     {
       "epoch": 1.9292301900997553,
-      "grad_norm": 11.479782104492188,
-      "learning_rate": 1.8424618859401467e-05,
-      "loss": 1.0353,
+      "grad_norm": 5.9964776039123535,
+      "learning_rate": 0.00030707698099002444,
+      "loss": 1.0115,
       "step": 10250
     },
     {
       "epoch": 1.9311123658949745,
-      "grad_norm": 12.16601848602295,
-      "learning_rate": 1.8413325804630153e-05,
-      "loss": 0.7897,
+      "grad_norm": 0.20760832726955414,
+      "learning_rate": 0.0003068887634105025,
+      "loss": 0.6393,
       "step": 10260
     },
     {
       "epoch": 1.932994541690194,
-      "grad_norm": 17.336091995239258,
-      "learning_rate": 1.8402032749858836e-05,
-      "loss": 0.8251,
+      "grad_norm": 4.026695251464844,
+      "learning_rate": 0.0003067005458309806,
+      "loss": 0.673,
       "step": 10270
     },
     {
       "epoch": 1.9348767174854131,
-      "grad_norm": 9.198282241821289,
-      "learning_rate": 1.8390739695087522e-05,
-      "loss": 0.7357,
+      "grad_norm": 2.453230142593384,
+      "learning_rate": 0.00030651232825145867,
+      "loss": 0.8183,
       "step": 10280
     },
     {
       "epoch": 1.9367588932806323,
-      "grad_norm": 12.248185157775879,
-      "learning_rate": 1.8379446640316208e-05,
-      "loss": 0.7766,
+      "grad_norm": 0.2803904116153717,
+      "learning_rate": 0.00030632411067193675,
+      "loss": 0.716,
       "step": 10290
     },
     {
       "epoch": 1.9386410690758518,
-      "grad_norm": 0.7270705103874207,
-      "learning_rate": 1.836815358554489e-05,
-      "loss": 0.5565,
+      "grad_norm": 0.20611296594142914,
+      "learning_rate": 0.0003061358930924149,
+      "loss": 0.5156,
       "step": 10300
     },
     {
       "epoch": 1.940523244871071,
-      "grad_norm": 18.369525909423828,
-      "learning_rate": 1.8356860530773577e-05,
-      "loss": 1.054,
+      "grad_norm": 7.233029365539551,
+      "learning_rate": 0.00030594767551289295,
+      "loss": 0.9044,
       "step": 10310
     },
     {
       "epoch": 1.9424054206662902,
-      "grad_norm": 3.5889933109283447,
-      "learning_rate": 1.834556747600226e-05,
-      "loss": 1.0666,
+      "grad_norm": 0.5617844462394714,
+      "learning_rate": 0.00030575945793337103,
+      "loss": 1.0455,
       "step": 10320
     },
     {
       "epoch": 1.9442875964615096,
-      "grad_norm": 16.404603958129883,
-      "learning_rate": 1.8334274421230942e-05,
-      "loss": 0.5469,
+      "grad_norm": 2.7063069343566895,
+      "learning_rate": 0.00030557124035384905,
+      "loss": 0.4773,
       "step": 10330
     },
     {
       "epoch": 1.9461697722567288,
-      "grad_norm": 5.033247470855713,
-      "learning_rate": 1.8322981366459628e-05,
-      "loss": 0.7387,
+      "grad_norm": 4.051327228546143,
+      "learning_rate": 0.00030538302277432713,
+      "loss": 0.665,
       "step": 10340
     },
     {
       "epoch": 1.948051948051948,
-      "grad_norm": 20.548690795898438,
-      "learning_rate": 1.831168831168831e-05,
-      "loss": 0.6346,
+      "grad_norm": 3.9234256744384766,
+      "learning_rate": 0.0003051948051948052,
+      "loss": 0.5562,
       "step": 10350
     },
     {
       "epoch": 1.9499341238471675,
-      "grad_norm": 19.923683166503906,
-      "learning_rate": 1.8300395256916997e-05,
-      "loss": 0.6418,
+      "grad_norm": 2.521334648132324,
+      "learning_rate": 0.0003050065876152833,
+      "loss": 0.6868,
       "step": 10360
     },
     {
       "epoch": 1.9518162996423865,
-      "grad_norm": 14.779675483703613,
-      "learning_rate": 1.828910220214568e-05,
-      "loss": 0.6338,
+      "grad_norm": 2.2171504497528076,
+      "learning_rate": 0.00030481837003576136,
+      "loss": 0.4326,
       "step": 10370
     },
     {
       "epoch": 1.9536984754376059,
-      "grad_norm": 24.59194564819336,
-      "learning_rate": 1.8277809147374366e-05,
-      "loss": 0.6815,
+      "grad_norm": 0.8069236278533936,
+      "learning_rate": 0.00030463015245623943,
+      "loss": 0.637,
       "step": 10380
     },
     {
       "epoch": 1.955580651232825,
-      "grad_norm": 6.548442840576172,
-      "learning_rate": 1.8266516092603052e-05,
-      "loss": 0.9073,
+      "grad_norm": 2.4707930088043213,
+      "learning_rate": 0.0003044419348767175,
+      "loss": 0.6729,
       "step": 10390
     },
     {
       "epoch": 1.9574628270280443,
-      "grad_norm": 25.888690948486328,
-      "learning_rate": 1.8255223037831734e-05,
-      "loss": 0.8705,
+      "grad_norm": 2.727604389190674,
+      "learning_rate": 0.0003042537172971956,
+      "loss": 0.7498,
       "step": 10400
     },
     {
       "epoch": 1.9593450028232637,
-      "grad_norm": 14.560491561889648,
-      "learning_rate": 1.8243929983060417e-05,
-      "loss": 1.0138,
+      "grad_norm": 3.5283005237579346,
+      "learning_rate": 0.0003040654997176736,
+      "loss": 0.9411,
       "step": 10410
     },
     {
       "epoch": 1.961227178618483,
-      "grad_norm": 18.18019676208496,
-      "learning_rate": 1.82326369282891e-05,
-      "loss": 0.7988,
+      "grad_norm": 3.2777674198150635,
+      "learning_rate": 0.0003038772821381517,
+      "loss": 0.6871,
       "step": 10420
     },
     {
       "epoch": 1.9631093544137022,
-      "grad_norm": 6.463456153869629,
-      "learning_rate": 1.8221343873517786e-05,
-      "loss": 0.6973,
+      "grad_norm": 2.1917569637298584,
+      "learning_rate": 0.00030368906455862976,
+      "loss": 0.603,
       "step": 10430
     },
     {
       "epoch": 1.9649915302089216,
-      "grad_norm": 6.2463226318359375,
-      "learning_rate": 1.8210050818746472e-05,
-      "loss": 0.6555,
+      "grad_norm": 0.9027181267738342,
+      "learning_rate": 0.00030350084697910784,
+      "loss": 0.4535,
       "step": 10440
     },
     {
       "epoch": 1.9668737060041408,
-      "grad_norm": 18.08149528503418,
-      "learning_rate": 1.8198757763975155e-05,
-      "loss": 0.6245,
+      "grad_norm": 3.275017261505127,
+      "learning_rate": 0.0003033126293995859,
+      "loss": 0.6109,
       "step": 10450
     },
     {
       "epoch": 1.96875588179936,
-      "grad_norm": 17.84767723083496,
-      "learning_rate": 1.818746470920384e-05,
-      "loss": 1.1495,
+      "grad_norm": 2.2948038578033447,
+      "learning_rate": 0.000303124411820064,
+      "loss": 0.9541,
       "step": 10460
     },
     {
       "epoch": 1.9706380575945794,
-      "grad_norm": 6.831356048583984,
-      "learning_rate": 1.8176171654432527e-05,
-      "loss": 0.8015,
+      "grad_norm": 2.1932127475738525,
+      "learning_rate": 0.00030293619424054207,
+      "loss": 0.7815,
       "step": 10470
     },
     {
       "epoch": 1.9725202333897986,
-      "grad_norm": 14.89473819732666,
-      "learning_rate": 1.816487859966121e-05,
-      "loss": 0.8459,
+      "grad_norm": 2.2697620391845703,
+      "learning_rate": 0.00030274797666102014,
+      "loss": 0.7469,
       "step": 10480
     },
     {
       "epoch": 1.9744024091850179,
-      "grad_norm": 18.0239315032959,
-      "learning_rate": 1.8153585544889896e-05,
-      "loss": 0.891,
+      "grad_norm": 3.1242027282714844,
+      "learning_rate": 0.0003025597590814982,
+      "loss": 0.8609,
       "step": 10490
     },
     {
       "epoch": 1.9762845849802373,
-      "grad_norm": 15.742325782775879,
-      "learning_rate": 1.8142292490118575e-05,
-      "loss": 1.2079,
+      "grad_norm": 2.455536365509033,
+      "learning_rate": 0.00030237154150197624,
+      "loss": 0.9871,
       "step": 10500
     },
     {
       "epoch": 1.9781667607754563,
-      "grad_norm": 6.855963230133057,
-      "learning_rate": 1.813099943534726e-05,
-      "loss": 0.7506,
+      "grad_norm": 1.7314091920852661,
+      "learning_rate": 0.0003021833239224543,
+      "loss": 0.7157,
       "step": 10510
     },
     {
       "epoch": 1.9800489365706757,
-      "grad_norm": 23.201900482177734,
-      "learning_rate": 1.8119706380575947e-05,
-      "loss": 0.854,
+      "grad_norm": 2.10880708694458,
+      "learning_rate": 0.00030199510634293245,
+      "loss": 0.7884,
       "step": 10520
     },
     {
       "epoch": 1.981931112365895,
-      "grad_norm": 11.044225692749023,
-      "learning_rate": 1.810841332580463e-05,
-      "loss": 0.9139,
+      "grad_norm": 3.019819974899292,
+      "learning_rate": 0.0003018068887634105,
+      "loss": 0.7561,
       "step": 10530
     },
     {
       "epoch": 1.9838132881611141,
-      "grad_norm": 10.52823257446289,
-      "learning_rate": 1.8097120271033316e-05,
-      "loss": 0.8894,
+      "grad_norm": 1.5337930917739868,
+      "learning_rate": 0.0003016186711838886,
+      "loss": 0.7473,
       "step": 10540
     },
     {
       "epoch": 1.9856954639563336,
-      "grad_norm": 11.858381271362305,
-      "learning_rate": 1.8085827216262e-05,
-      "loss": 0.5292,
+      "grad_norm": 4.329298973083496,
+      "learning_rate": 0.0003014304536043667,
+      "loss": 0.4503,
       "step": 10550
     },
     {
       "epoch": 1.9875776397515528,
-      "grad_norm": 20.706117630004883,
-      "learning_rate": 1.8074534161490685e-05,
-      "loss": 0.7082,
+      "grad_norm": 4.061004638671875,
+      "learning_rate": 0.00030124223602484476,
+      "loss": 0.5419,
       "step": 10560
     },
     {
       "epoch": 1.989459815546772,
-      "grad_norm": 9.068899154663086,
-      "learning_rate": 1.806324110671937e-05,
-      "loss": 0.9249,
+      "grad_norm": 2.1341538429260254,
+      "learning_rate": 0.00030105401844532283,
+      "loss": 0.873,
       "step": 10570
     },
     {
       "epoch": 1.9913419913419914,
-      "grad_norm": 12.875887870788574,
-      "learning_rate": 1.8051948051948053e-05,
-      "loss": 0.7443,
+      "grad_norm": 3.725860834121704,
+      "learning_rate": 0.0003008658008658009,
+      "loss": 0.7146,
       "step": 10580
     },
     {
       "epoch": 1.9932241671372106,
-      "grad_norm": 15.794068336486816,
-      "learning_rate": 1.8040654997176736e-05,
-      "loss": 0.7504,
+      "grad_norm": 4.153800010681152,
+      "learning_rate": 0.00030067758328627893,
+      "loss": 0.6242,
       "step": 10590
     },
     {
       "epoch": 1.9951063429324298,
-      "grad_norm": 9.319598197937012,
-      "learning_rate": 1.802936194240542e-05,
-      "loss": 0.7693,
+      "grad_norm": 2.7031161785125732,
+      "learning_rate": 0.000300489365706757,
+      "loss": 0.7066,
       "step": 10600
     },
     {
       "epoch": 1.9969885187276493,
-      "grad_norm": 7.584019660949707,
-      "learning_rate": 1.8018068887634105e-05,
-      "loss": 0.8374,
+      "grad_norm": 2.1315081119537354,
+      "learning_rate": 0.0003003011481272351,
+      "loss": 0.7769,
       "step": 10610
     },
     {
       "epoch": 1.9988706945228685,
-      "grad_norm": 7.562906742095947,
-      "learning_rate": 1.800677583286279e-05,
-      "loss": 0.7839,
+      "grad_norm": 2.260913610458374,
+      "learning_rate": 0.00030011293054771316,
+      "loss": 0.7608,
       "step": 10620
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.9056,
-      "eval_loss": 0.3731466829776764,
-      "eval_runtime": 275.5436,
-      "eval_samples_per_second": 27.219,
-      "eval_steps_per_second": 3.404,
+      "eval_accuracy": 0.9138666666666667,
+      "eval_loss": 0.2929131090641022,
+      "eval_runtime": 173.9091,
+      "eval_samples_per_second": 43.126,
+      "eval_steps_per_second": 5.394,
       "step": 10626
     },
     {
       "epoch": 2.0007528703180877,
-      "grad_norm": 26.831989288330078,
-      "learning_rate": 1.7995482778091474e-05,
-      "loss": 0.8433,
+      "grad_norm": 4.4684343338012695,
+      "learning_rate": 0.00029992471296819124,
+      "loss": 0.7618,
       "step": 10630
     },
     {
       "epoch": 2.002635046113307,
-      "grad_norm": 14.433839797973633,
-      "learning_rate": 1.798418972332016e-05,
-      "loss": 0.874,
+      "grad_norm": 2.1581475734710693,
+      "learning_rate": 0.0002997364953886693,
+      "loss": 0.7593,
       "step": 10640
     },
     {
       "epoch": 2.004517221908526,
-      "grad_norm": 6.329076290130615,
-      "learning_rate": 1.7972896668548842e-05,
-      "loss": 0.953,
+      "grad_norm": 3.0635669231414795,
+      "learning_rate": 0.0002995482778091474,
+      "loss": 1.0145,
       "step": 10650
     },
     {
       "epoch": 2.0063993977037455,
-      "grad_norm": 28.34707260131836,
-      "learning_rate": 1.796160361377753e-05,
-      "loss": 0.7231,
+      "grad_norm": 1.1936360597610474,
+      "learning_rate": 0.00029936006022962547,
+      "loss": 0.5021,
       "step": 10660
     },
     {
       "epoch": 2.008281573498965,
-      "grad_norm": 6.1005778312683105,
-      "learning_rate": 1.795031055900621e-05,
-      "loss": 1.0041,
+      "grad_norm": 2.743591547012329,
+      "learning_rate": 0.0002991718426501035,
+      "loss": 1.0748,
       "step": 10670
     },
     {
       "epoch": 2.010163749294184,
-      "grad_norm": 5.388144493103027,
-      "learning_rate": 1.7939017504234894e-05,
-      "loss": 0.71,
+      "grad_norm": 3.357374668121338,
+      "learning_rate": 0.00029898362507058156,
+      "loss": 0.7152,
       "step": 10680
     },
     {
       "epoch": 2.0120459250894034,
-      "grad_norm": 22.38304901123047,
-      "learning_rate": 1.792772444946358e-05,
-      "loss": 0.5202,
+      "grad_norm": 3.9789493083953857,
+      "learning_rate": 0.00029879540749105964,
+      "loss": 0.3214,
       "step": 10690
     },
     {
       "epoch": 2.013928100884623,
-      "grad_norm": 6.131063938140869,
-      "learning_rate": 1.7916431394692266e-05,
-      "loss": 0.8542,
+      "grad_norm": 9.24729061126709,
+      "learning_rate": 0.0002986071899115377,
+      "loss": 0.8628,
       "step": 10700
     },
     {
       "epoch": 2.015810276679842,
-      "grad_norm": 18.350811004638672,
-      "learning_rate": 1.790513833992095e-05,
-      "loss": 0.8689,
+      "grad_norm": 4.8567280769348145,
+      "learning_rate": 0.0002984189723320158,
+      "loss": 0.8648,
       "step": 10710
     },
     {
       "epoch": 2.0176924524750612,
-      "grad_norm": 1.005018949508667,
-      "learning_rate": 1.7893845285149635e-05,
-      "loss": 0.8021,
+      "grad_norm": 0.8375732898712158,
+      "learning_rate": 0.00029823075475249387,
+      "loss": 0.792,
       "step": 10720
     },
     {
       "epoch": 2.0195746282702802,
-      "grad_norm": 11.301158905029297,
-      "learning_rate": 1.7882552230378317e-05,
-      "loss": 0.7946,
+      "grad_norm": 2.4992613792419434,
+      "learning_rate": 0.00029804253717297195,
+      "loss": 0.7085,
       "step": 10730
     },
     {
       "epoch": 2.0214568040654997,
-      "grad_norm": 22.24584197998047,
-      "learning_rate": 1.7871259175607003e-05,
-      "loss": 0.8225,
+      "grad_norm": 4.913736820220947,
+      "learning_rate": 0.0002978543195934501,
+      "loss": 0.6829,
       "step": 10740
     },
     {
       "epoch": 2.023338979860719,
-      "grad_norm": 4.584529399871826,
-      "learning_rate": 1.785996612083569e-05,
-      "loss": 0.9226,
+      "grad_norm": 3.4823250770568848,
+      "learning_rate": 0.00029766610201392815,
+      "loss": 0.7979,
       "step": 10750
     },
     {
       "epoch": 2.025221155655938,
-      "grad_norm": 9.03012466430664,
-      "learning_rate": 1.784867306606437e-05,
-      "loss": 0.7807,
+      "grad_norm": 3.6224143505096436,
+      "learning_rate": 0.0002974778844344062,
+      "loss": 0.781,
       "step": 10760
     },
     {
       "epoch": 2.0271033314511575,
-      "grad_norm": 14.892461776733398,
-      "learning_rate": 1.7837380011293055e-05,
-      "loss": 0.5403,
+      "grad_norm": 2.89218807220459,
+      "learning_rate": 0.00029728966685488425,
+      "loss": 0.5409,
       "step": 10770
     },
     {
       "epoch": 2.028985507246377,
-      "grad_norm": 18.233808517456055,
-      "learning_rate": 1.7826086956521738e-05,
-      "loss": 0.7445,
+      "grad_norm": 2.972883462905884,
+      "learning_rate": 0.00029710144927536233,
+      "loss": 0.5425,
       "step": 10780
     },
     {
       "epoch": 2.030867683041596,
-      "grad_norm": 7.108969211578369,
-      "learning_rate": 1.7814793901750424e-05,
-      "loss": 0.7014,
+      "grad_norm": 2.568448781967163,
+      "learning_rate": 0.0002969132316958404,
+      "loss": 0.7254,
       "step": 10790
     },
     {
       "epoch": 2.0327498588368154,
-      "grad_norm": 3.6049118041992188,
-      "learning_rate": 1.780350084697911e-05,
-      "loss": 0.6114,
+      "grad_norm": 1.1434894800186157,
+      "learning_rate": 0.0002967250141163185,
+      "loss": 0.6191,
       "step": 10800
     },
     {
       "epoch": 2.034632034632035,
-      "grad_norm": 8.403626441955566,
-      "learning_rate": 1.7792207792207792e-05,
-      "loss": 0.856,
+      "grad_norm": 2.585139751434326,
+      "learning_rate": 0.00029653679653679656,
+      "loss": 0.8831,
       "step": 10810
     },
     {
       "epoch": 2.036514210427254,
-      "grad_norm": 3.540233850479126,
-      "learning_rate": 1.778091473743648e-05,
-      "loss": 0.5674,
+      "grad_norm": 2.129727363586426,
+      "learning_rate": 0.00029634857895727463,
+      "loss": 0.5341,
       "step": 10820
     },
     {
       "epoch": 2.038396386222473,
-      "grad_norm": 13.208251953125,
-      "learning_rate": 1.776962168266516e-05,
-      "loss": 0.829,
+      "grad_norm": 3.551943063735962,
+      "learning_rate": 0.0002961603613777527,
+      "loss": 0.8215,
       "step": 10830
     },
     {
       "epoch": 2.0402785620176926,
-      "grad_norm": 3.715716600418091,
-      "learning_rate": 1.7758328627893847e-05,
-      "loss": 0.7736,
+      "grad_norm": 0.25661563873291016,
+      "learning_rate": 0.0002959721437982308,
+      "loss": 0.6864,
       "step": 10840
     },
     {
       "epoch": 2.0421607378129116,
-      "grad_norm": 14.193865776062012,
-      "learning_rate": 1.774703557312253e-05,
-      "loss": 0.4945,
+      "grad_norm": 2.998699903488159,
+      "learning_rate": 0.0002957839262187088,
+      "loss": 0.512,
       "step": 10850
     },
     {
       "epoch": 2.044042913608131,
-      "grad_norm": 0.2950342893600464,
-      "learning_rate": 1.7735742518351213e-05,
-      "loss": 0.7292,
+      "grad_norm": 0.43398717045783997,
+      "learning_rate": 0.0002955957086391869,
+      "loss": 0.5836,
       "step": 10860
     },
     {
       "epoch": 2.04592508940335,
-      "grad_norm": 6.488024711608887,
-      "learning_rate": 1.77244494635799e-05,
-      "loss": 0.7146,
+      "grad_norm": 1.4403588771820068,
+      "learning_rate": 0.00029540749105966496,
+      "loss": 0.6644,
       "step": 10870
     },
     {
       "epoch": 2.0478072651985695,
-      "grad_norm": 19.793678283691406,
-      "learning_rate": 1.771315640880858e-05,
-      "loss": 0.6736,
+      "grad_norm": 2.820168972015381,
+      "learning_rate": 0.00029521927348014304,
+      "loss": 0.6328,
       "step": 10880
     },
     {
       "epoch": 2.049689440993789,
-      "grad_norm": 12.970588684082031,
-      "learning_rate": 1.7701863354037267e-05,
-      "loss": 0.8303,
+      "grad_norm": 3.101900339126587,
+      "learning_rate": 0.0002950310559006211,
+      "loss": 0.6874,
       "step": 10890
     },
     {
       "epoch": 2.051571616789008,
-      "grad_norm": 8.980803489685059,
-      "learning_rate": 1.7690570299265954e-05,
-      "loss": 0.9072,
+      "grad_norm": 3.710576295852661,
+      "learning_rate": 0.0002948428383210992,
+      "loss": 0.9401,
       "step": 10900
     },
     {
       "epoch": 2.0534537925842273,
-      "grad_norm": 3.427503824234009,
-      "learning_rate": 1.7679277244494636e-05,
-      "loss": 0.6361,
+      "grad_norm": 2.558265447616577,
+      "learning_rate": 0.00029465462074157727,
+      "loss": 0.6474,
       "step": 10910
     },
     {
       "epoch": 2.0553359683794468,
-      "grad_norm": 9.372367858886719,
-      "learning_rate": 1.7667984189723322e-05,
-      "loss": 0.601,
+      "grad_norm": 1.6897790431976318,
+      "learning_rate": 0.00029446640316205534,
+      "loss": 0.5211,
       "step": 10920
     },
     {
       "epoch": 2.0572181441746658,
-      "grad_norm": 31.908103942871094,
-      "learning_rate": 1.7656691134952005e-05,
-      "loss": 0.4396,
+      "grad_norm": 3.507096529006958,
+      "learning_rate": 0.0002942781855825334,
+      "loss": 0.5326,
       "step": 10930
     },
     {
       "epoch": 2.059100319969885,
-      "grad_norm": 10.805863380432129,
-      "learning_rate": 1.7645398080180688e-05,
-      "loss": 0.6055,
+      "grad_norm": 3.556894540786743,
+      "learning_rate": 0.00029408996800301144,
+      "loss": 0.6389,
       "step": 10940
     },
     {
       "epoch": 2.0609824957651046,
-      "grad_norm": 9.006129264831543,
-      "learning_rate": 1.7634105025409374e-05,
-      "loss": 0.7486,
+      "grad_norm": 2.366298198699951,
+      "learning_rate": 0.0002939017504234895,
+      "loss": 0.5868,
       "step": 10950
     },
     {
       "epoch": 2.0628646715603236,
-      "grad_norm": 12.054622650146484,
-      "learning_rate": 1.7622811970638056e-05,
-      "loss": 0.5213,
+      "grad_norm": 2.686265707015991,
+      "learning_rate": 0.00029371353284396765,
+      "loss": 0.5434,
       "step": 10960
     },
     {
       "epoch": 2.064746847355543,
-      "grad_norm": 4.848597526550293,
-      "learning_rate": 1.7611518915866743e-05,
-      "loss": 0.4883,
+      "grad_norm": 1.7855459451675415,
+      "learning_rate": 0.0002935253152644457,
+      "loss": 0.5321,
       "step": 10970
     },
     {
       "epoch": 2.0666290231507625,
-      "grad_norm": 0.7347248792648315,
-      "learning_rate": 1.760022586109543e-05,
-      "loss": 0.6849,
+      "grad_norm": 2.971705436706543,
+      "learning_rate": 0.0002933370976849238,
+      "loss": 0.6643,
       "step": 10980
     },
     {
       "epoch": 2.0685111989459815,
-      "grad_norm": 3.6684467792510986,
-      "learning_rate": 1.758893280632411e-05,
-      "loss": 0.7872,
+      "grad_norm": 0.7849436402320862,
+      "learning_rate": 0.0002931488801054019,
+      "loss": 0.6436,
       "step": 10990
     },
     {
       "epoch": 2.070393374741201,
-      "grad_norm": 1.874008059501648,
-      "learning_rate": 1.7577639751552797e-05,
-      "loss": 0.8974,
+      "grad_norm": 0.2096141129732132,
+      "learning_rate": 0.00029296066252587996,
+      "loss": 0.8931,
       "step": 11000
     },
     {
       "epoch": 2.0722755505364203,
-      "grad_norm": 8.359776496887207,
-      "learning_rate": 1.756634669678148e-05,
-      "loss": 0.9378,
+      "grad_norm": 2.518460273742676,
+      "learning_rate": 0.00029277244494635803,
+      "loss": 0.9559,
       "step": 11010
     },
     {
       "epoch": 2.0741577263316393,
-      "grad_norm": 8.968391418457031,
-      "learning_rate": 1.7555053642010163e-05,
-      "loss": 0.404,
+      "grad_norm": 4.189047813415527,
+      "learning_rate": 0.00029258422736683605,
+      "loss": 0.4586,
       "step": 11020
     },
     {
       "epoch": 2.0760399021268587,
-      "grad_norm": 40.07794952392578,
-      "learning_rate": 1.754376058723885e-05,
-      "loss": 0.5055,
+      "grad_norm": 7.893125057220459,
+      "learning_rate": 0.00029239600978731413,
+      "loss": 0.4236,
       "step": 11030
     },
     {
       "epoch": 2.0779220779220777,
-      "grad_norm": 45.67797088623047,
-      "learning_rate": 1.753246753246753e-05,
-      "loss": 0.5616,
+      "grad_norm": 19.647254943847656,
+      "learning_rate": 0.0002922077922077922,
+      "loss": 0.5715,
       "step": 11040
     },
     {
       "epoch": 2.079804253717297,
-      "grad_norm": 11.17763900756836,
-      "learning_rate": 1.7521174477696218e-05,
-      "loss": 0.6244,
+      "grad_norm": 2.632347822189331,
+      "learning_rate": 0.0002920195746282703,
+      "loss": 0.5478,
       "step": 11050
     },
     {
       "epoch": 2.0816864295125166,
-      "grad_norm": 11.508498191833496,
-      "learning_rate": 1.75098814229249e-05,
-      "loss": 0.6413,
+      "grad_norm": 5.197866439819336,
+      "learning_rate": 0.00029183135704874836,
+      "loss": 0.7797,
       "step": 11060
     },
     {
       "epoch": 2.0835686053077356,
-      "grad_norm": 7.054535865783691,
-      "learning_rate": 1.7498588368153586e-05,
-      "loss": 1.0319,
+      "grad_norm": 2.2426016330718994,
+      "learning_rate": 0.00029164313946922644,
+      "loss": 0.9397,
       "step": 11070
     },
     {
       "epoch": 2.085450781102955,
-      "grad_norm": 1.5207102298736572,
-      "learning_rate": 1.7487295313382272e-05,
-      "loss": 0.6158,
+      "grad_norm": 1.2567468881607056,
+      "learning_rate": 0.0002914549218897045,
+      "loss": 0.6406,
       "step": 11080
     },
     {
       "epoch": 2.0873329568981744,
-      "grad_norm": 6.750110149383545,
-      "learning_rate": 1.7476002258610955e-05,
-      "loss": 0.5906,
+      "grad_norm": 3.520487070083618,
+      "learning_rate": 0.0002912667043101826,
+      "loss": 0.5526,
       "step": 11090
     },
     {
       "epoch": 2.0892151326933934,
-      "grad_norm": 5.071254730224609,
-      "learning_rate": 1.746470920383964e-05,
-      "loss": 0.7611,
+      "grad_norm": 3.5984339714050293,
+      "learning_rate": 0.00029107848673066067,
+      "loss": 0.6277,
       "step": 11100
     },
     {
       "epoch": 2.091097308488613,
-      "grad_norm": 8.039938926696777,
-      "learning_rate": 1.745341614906832e-05,
-      "loss": 0.7571,
+      "grad_norm": 3.593083381652832,
+      "learning_rate": 0.0002908902691511387,
+      "loss": 0.8077,
       "step": 11110
     },
     {
       "epoch": 2.0929794842838323,
-      "grad_norm": 9.071053504943848,
-      "learning_rate": 1.7442123094297007e-05,
-      "loss": 0.6185,
+      "grad_norm": 4.32984733581543,
+      "learning_rate": 0.00029070205157161676,
+      "loss": 0.5985,
       "step": 11120
     },
     {
       "epoch": 2.0948616600790513,
-      "grad_norm": 7.90115213394165,
-      "learning_rate": 1.7430830039525693e-05,
-      "loss": 0.8708,
+      "grad_norm": 4.792181968688965,
+      "learning_rate": 0.00029051383399209484,
+      "loss": 0.91,
       "step": 11130
     },
     {
       "epoch": 2.0967438358742707,
-      "grad_norm": 1.540299415588379,
-      "learning_rate": 1.7419536984754375e-05,
-      "loss": 0.6717,
+      "grad_norm": 0.8018267154693604,
+      "learning_rate": 0.0002903256164125729,
+      "loss": 0.8573,
       "step": 11140
     },
     {
       "epoch": 2.09862601166949,
-      "grad_norm": 5.848174095153809,
-      "learning_rate": 1.740824392998306e-05,
-      "loss": 0.785,
+      "grad_norm": 1.1662284135818481,
+      "learning_rate": 0.000290137398833051,
+      "loss": 0.8669,
       "step": 11150
     },
     {
       "epoch": 2.100508187464709,
-      "grad_norm": 12.827478408813477,
-      "learning_rate": 1.7396950875211744e-05,
-      "loss": 0.6168,
+      "grad_norm": 3.500774621963501,
+      "learning_rate": 0.00028994918125352907,
+      "loss": 0.6141,
       "step": 11160
     },
     {
       "epoch": 2.1023903632599286,
-      "grad_norm": 2.8471341133117676,
-      "learning_rate": 1.738565782044043e-05,
-      "loss": 0.6963,
+      "grad_norm": 0.8624693751335144,
+      "learning_rate": 0.00028976096367400715,
+      "loss": 0.6528,
       "step": 11170
     },
     {
       "epoch": 2.1042725390551476,
-      "grad_norm": 12.310620307922363,
-      "learning_rate": 1.7374364765669116e-05,
-      "loss": 0.6679,
+      "grad_norm": 4.650200843811035,
+      "learning_rate": 0.0002895727460944853,
+      "loss": 0.5764,
       "step": 11180
     },
     {
       "epoch": 2.106154714850367,
-      "grad_norm": 2.6820571422576904,
-      "learning_rate": 1.73630717108978e-05,
-      "loss": 0.7162,
+      "grad_norm": 0.9694834351539612,
+      "learning_rate": 0.00028938452851496335,
+      "loss": 0.7189,
       "step": 11190
     },
     {
       "epoch": 2.1080368906455864,
-      "grad_norm": 28.728363037109375,
-      "learning_rate": 1.735177865612648e-05,
-      "loss": 1.1478,
+      "grad_norm": 3.7187447547912598,
+      "learning_rate": 0.0002891963109354414,
+      "loss": 1.0254,
       "step": 11200
     },
     {
       "epoch": 2.1099190664408054,
-      "grad_norm": 13.076791763305664,
-      "learning_rate": 1.7340485601355168e-05,
-      "loss": 0.9537,
+      "grad_norm": 0.849228024482727,
+      "learning_rate": 0.00028900809335591945,
+      "loss": 0.8398,
       "step": 11210
     },
     {
       "epoch": 2.111801242236025,
-      "grad_norm": 22.759828567504883,
-      "learning_rate": 1.732919254658385e-05,
-      "loss": 0.8617,
+      "grad_norm": 3.1720340251922607,
+      "learning_rate": 0.00028881987577639753,
+      "loss": 0.7663,
       "step": 11220
     },
     {
       "epoch": 2.1136834180312443,
-      "grad_norm": 9.096440315246582,
-      "learning_rate": 1.7317899491812536e-05,
-      "loss": 0.5249,
+      "grad_norm": 0.4519988000392914,
+      "learning_rate": 0.0002886316581968756,
+      "loss": 0.3329,
       "step": 11230
     },
     {
       "epoch": 2.1155655938264633,
-      "grad_norm": 14.975255012512207,
-      "learning_rate": 1.730660643704122e-05,
-      "loss": 0.6067,
+      "grad_norm": 3.3207151889801025,
+      "learning_rate": 0.0002884434406173537,
+      "loss": 0.5899,
       "step": 11240
     },
     {
       "epoch": 2.1174477696216827,
-      "grad_norm": 16.731473922729492,
-      "learning_rate": 1.7295313382269905e-05,
-      "loss": 0.588,
+      "grad_norm": 2.0980584621429443,
+      "learning_rate": 0.00028825522303783176,
+      "loss": 0.4889,
       "step": 11250
     },
     {
       "epoch": 2.119329945416902,
-      "grad_norm": 37.46254348754883,
-      "learning_rate": 1.728402032749859e-05,
-      "loss": 0.6994,
+      "grad_norm": 6.887931823730469,
+      "learning_rate": 0.00028806700545830983,
+      "loss": 0.7055,
       "step": 11260
     },
     {
       "epoch": 2.121212121212121,
-      "grad_norm": 4.683788299560547,
-      "learning_rate": 1.7272727272727274e-05,
-      "loss": 0.5429,
+      "grad_norm": 0.9553453326225281,
+      "learning_rate": 0.0002878787878787879,
+      "loss": 0.5491,
       "step": 11270
     },
     {
       "epoch": 2.1230942970073405,
-      "grad_norm": 1.7957135438919067,
-      "learning_rate": 1.7261434217955957e-05,
-      "loss": 0.4845,
+      "grad_norm": 0.46205490827560425,
+      "learning_rate": 0.00028769057029926593,
+      "loss": 0.4709,
       "step": 11280
     },
     {
       "epoch": 2.12497647280256,
-      "grad_norm": 13.624578475952148,
-      "learning_rate": 1.725014116318464e-05,
-      "loss": 0.892,
+      "grad_norm": 2.981776475906372,
+      "learning_rate": 0.000287502352719744,
+      "loss": 0.7086,
       "step": 11290
     },
     {
       "epoch": 2.126858648597779,
-      "grad_norm": 40.41400909423828,
-      "learning_rate": 1.7238848108413325e-05,
-      "loss": 0.8509,
+      "grad_norm": 6.404483318328857,
+      "learning_rate": 0.0002873141351402221,
+      "loss": 0.6254,
       "step": 11300
     },
     {
       "epoch": 2.1287408243929984,
-      "grad_norm": 16.918149948120117,
-      "learning_rate": 1.722755505364201e-05,
-      "loss": 0.9205,
+      "grad_norm": 2.772292375564575,
+      "learning_rate": 0.00028712591756070016,
+      "loss": 0.7199,
       "step": 11310
     },
     {
       "epoch": 2.1306230001882174,
-      "grad_norm": 35.62562561035156,
-      "learning_rate": 1.7216261998870694e-05,
-      "loss": 1.0122,
+      "grad_norm": 3.2327826023101807,
+      "learning_rate": 0.00028693769998117824,
+      "loss": 0.8967,
       "step": 11320
     },
     {
       "epoch": 2.132505175983437,
-      "grad_norm": 10.24116325378418,
-      "learning_rate": 1.720496894409938e-05,
-      "loss": 0.7243,
+      "grad_norm": 2.0082974433898926,
+      "learning_rate": 0.0002867494824016563,
+      "loss": 0.7676,
       "step": 11330
     },
     {
       "epoch": 2.1343873517786562,
-      "grad_norm": 20.230663299560547,
-      "learning_rate": 1.7193675889328063e-05,
-      "loss": 0.8808,
+      "grad_norm": 3.831946611404419,
+      "learning_rate": 0.0002865612648221344,
+      "loss": 0.8358,
       "step": 11340
     },
     {
       "epoch": 2.1362695275738752,
-      "grad_norm": 11.159217834472656,
-      "learning_rate": 1.718238283455675e-05,
-      "loss": 0.9238,
+      "grad_norm": 1.170330286026001,
+      "learning_rate": 0.00028637304724261247,
+      "loss": 0.9855,
       "step": 11350
     },
     {
       "epoch": 2.1381517033690947,
-      "grad_norm": 15.367325782775879,
-      "learning_rate": 1.7171089779785435e-05,
-      "loss": 0.837,
+      "grad_norm": 2.1947906017303467,
+      "learning_rate": 0.00028618482966309054,
+      "loss": 0.7599,
       "step": 11360
     },
     {
       "epoch": 2.140033879164314,
-      "grad_norm": 6.264622688293457,
-      "learning_rate": 1.7159796725014114e-05,
-      "loss": 0.7584,
+      "grad_norm": 0.594887375831604,
+      "learning_rate": 0.00028599661208356857,
+      "loss": 0.6236,
       "step": 11370
     },
     {
       "epoch": 2.141916054959533,
-      "grad_norm": 7.990375518798828,
-      "learning_rate": 1.71485036702428e-05,
-      "loss": 0.5522,
+      "grad_norm": 2.23850417137146,
+      "learning_rate": 0.00028580839450404664,
+      "loss": 0.5686,
       "step": 11380
     },
     {
       "epoch": 2.1437982307547525,
-      "grad_norm": 13.449748039245605,
-      "learning_rate": 1.7137210615471483e-05,
-      "loss": 0.9335,
+      "grad_norm": 2.4512314796447754,
+      "learning_rate": 0.0002856201769245247,
+      "loss": 0.8437,
       "step": 11390
     },
     {
       "epoch": 2.145680406549972,
-      "grad_norm": 16.00578498840332,
-      "learning_rate": 1.712591756070017e-05,
-      "loss": 0.6726,
+      "grad_norm": 2.53229022026062,
+      "learning_rate": 0.0002854319593450028,
+      "loss": 0.7272,
       "step": 11400
     },
     {
       "epoch": 2.147562582345191,
-      "grad_norm": 6.167961120605469,
-      "learning_rate": 1.7114624505928855e-05,
-      "loss": 0.9023,
+      "grad_norm": 2.3535914421081543,
+      "learning_rate": 0.0002852437417654809,
+      "loss": 0.9296,
       "step": 11410
     },
     {
       "epoch": 2.1494447581404104,
-      "grad_norm": 16.115100860595703,
-      "learning_rate": 1.7103331451157538e-05,
-      "loss": 0.9606,
+      "grad_norm": 3.0220978260040283,
+      "learning_rate": 0.000285055524185959,
+      "loss": 0.7514,
       "step": 11420
     },
     {
       "epoch": 2.15132693393563,
-      "grad_norm": 30.645904541015625,
-      "learning_rate": 1.7092038396386224e-05,
-      "loss": 0.6028,
+      "grad_norm": 4.237234592437744,
+      "learning_rate": 0.0002848673066064371,
+      "loss": 0.5762,
       "step": 11430
     },
     {
       "epoch": 2.153209109730849,
-      "grad_norm": 4.448369026184082,
-      "learning_rate": 1.7080745341614907e-05,
-      "loss": 0.9826,
+      "grad_norm": 1.3424307107925415,
+      "learning_rate": 0.00028467908902691516,
+      "loss": 0.9281,
       "step": 11440
     },
     {
       "epoch": 2.155091285526068,
-      "grad_norm": 13.802371978759766,
-      "learning_rate": 1.7069452286843593e-05,
-      "loss": 1.0515,
+      "grad_norm": 3.5801596641540527,
+      "learning_rate": 0.00028449087144739323,
+      "loss": 1.062,
       "step": 11450
     },
     {
       "epoch": 2.1569734613212876,
-      "grad_norm": 18.508493423461914,
-      "learning_rate": 1.7058159232072276e-05,
-      "loss": 0.5478,
+      "grad_norm": 3.8475425243377686,
+      "learning_rate": 0.00028430265386787125,
+      "loss": 0.5018,
       "step": 11460
     },
     {
       "epoch": 2.1588556371165066,
-      "grad_norm": 18.791305541992188,
-      "learning_rate": 1.7046866177300958e-05,
-      "loss": 0.6688,
+      "grad_norm": 5.038808822631836,
+      "learning_rate": 0.00028411443628834933,
+      "loss": 0.4989,
       "step": 11470
     },
     {
       "epoch": 2.160737812911726,
-      "grad_norm": 13.487887382507324,
-      "learning_rate": 1.7035573122529644e-05,
-      "loss": 0.7173,
+      "grad_norm": 4.923741817474365,
+      "learning_rate": 0.0002839262187088274,
+      "loss": 0.6093,
       "step": 11480
     },
     {
       "epoch": 2.162619988706945,
-      "grad_norm": 7.567160129547119,
-      "learning_rate": 1.702428006775833e-05,
-      "loss": 0.5889,
+      "grad_norm": 1.457595705986023,
+      "learning_rate": 0.0002837380011293055,
+      "loss": 0.6103,
       "step": 11490
     },
     {
       "epoch": 2.1645021645021645,
-      "grad_norm": 14.993462562561035,
-      "learning_rate": 1.7012987012987013e-05,
-      "loss": 0.6631,
+      "grad_norm": 3.0185911655426025,
+      "learning_rate": 0.00028354978354978356,
+      "loss": 0.7228,
       "step": 11500
     },
     {
       "epoch": 2.166384340297384,
-      "grad_norm": 2.703808069229126,
-      "learning_rate": 1.70016939582157e-05,
-      "loss": 0.8147,
+      "grad_norm": 1.2738662958145142,
+      "learning_rate": 0.00028336156597026164,
+      "loss": 0.6904,
       "step": 11510
     },
     {
       "epoch": 2.168266516092603,
-      "grad_norm": 13.842448234558105,
-      "learning_rate": 1.6990400903444382e-05,
-      "loss": 0.6228,
+      "grad_norm": 1.419976830482483,
+      "learning_rate": 0.0002831733483907397,
+      "loss": 0.7167,
       "step": 11520
     },
     {
       "epoch": 2.1701486918878223,
-      "grad_norm": 14.048020362854004,
-      "learning_rate": 1.6979107848673068e-05,
-      "loss": 0.6811,
+      "grad_norm": 3.3569860458374023,
+      "learning_rate": 0.0002829851308112178,
+      "loss": 0.5419,
       "step": 11530
     },
     {
       "epoch": 2.1720308676830418,
-      "grad_norm": 9.474082946777344,
-      "learning_rate": 1.6967814793901754e-05,
-      "loss": 1.0404,
+      "grad_norm": 1.6389613151550293,
+      "learning_rate": 0.00028279691323169587,
+      "loss": 0.8041,
       "step": 11540
     },
     {
       "epoch": 2.1739130434782608,
-      "grad_norm": 7.542745113372803,
-      "learning_rate": 1.6956521739130433e-05,
-      "loss": 0.7817,
+      "grad_norm": 1.0149704217910767,
+      "learning_rate": 0.0002826086956521739,
+      "loss": 0.6128,
       "step": 11550
     },
     {
       "epoch": 2.17579521927348,
-      "grad_norm": 14.078716278076172,
-      "learning_rate": 1.694522868435912e-05,
-      "loss": 0.9704,
+      "grad_norm": 3.377378225326538,
+      "learning_rate": 0.00028242047807265196,
+      "loss": 0.8804,
       "step": 11560
     },
     {
       "epoch": 2.1776773950686996,
-      "grad_norm": 19.203222274780273,
-      "learning_rate": 1.6933935629587802e-05,
-      "loss": 0.6797,
+      "grad_norm": 1.8844712972640991,
+      "learning_rate": 0.00028223226049313004,
+      "loss": 0.8288,
       "step": 11570
     },
     {
       "epoch": 2.1795595708639186,
-      "grad_norm": 10.718687057495117,
-      "learning_rate": 1.6922642574816488e-05,
-      "loss": 0.7273,
+      "grad_norm": 2.894857168197632,
+      "learning_rate": 0.0002820440429136081,
+      "loss": 0.7565,
       "step": 11580
     },
     {
       "epoch": 2.181441746659138,
-      "grad_norm": 7.656870365142822,
-      "learning_rate": 1.6911349520045174e-05,
-      "loss": 0.4612,
+      "grad_norm": 4.635049343109131,
+      "learning_rate": 0.0002818558253340862,
+      "loss": 0.6153,
       "step": 11590
     },
     {
       "epoch": 2.183323922454357,
-      "grad_norm": 9.046762466430664,
-      "learning_rate": 1.6900056465273857e-05,
-      "loss": 0.815,
+      "grad_norm": 2.9711811542510986,
+      "learning_rate": 0.00028166760775456427,
+      "loss": 0.6096,
       "step": 11600
     },
     {
       "epoch": 2.1852060982495765,
-      "grad_norm": 16.06904411315918,
-      "learning_rate": 1.6888763410502543e-05,
-      "loss": 0.7,
+      "grad_norm": 2.8086323738098145,
+      "learning_rate": 0.00028147939017504235,
+      "loss": 0.6508,
       "step": 11610
     },
     {
       "epoch": 2.187088274044796,
-      "grad_norm": 17.846696853637695,
-      "learning_rate": 1.6877470355731226e-05,
-      "loss": 0.8311,
+      "grad_norm": 2.190927743911743,
+      "learning_rate": 0.0002812911725955205,
+      "loss": 0.9356,
       "step": 11620
     },
     {
       "epoch": 2.188970449840015,
-      "grad_norm": 0.43426594138145447,
-      "learning_rate": 1.686617730095991e-05,
-      "loss": 0.901,
+      "grad_norm": 0.3786638379096985,
+      "learning_rate": 0.0002811029550159985,
+      "loss": 1.0633,
       "step": 11630
     },
     {
       "epoch": 2.1908526256352343,
-      "grad_norm": 21.463172912597656,
-      "learning_rate": 1.6854884246188594e-05,
-      "loss": 1.1355,
+      "grad_norm": 4.207404136657715,
+      "learning_rate": 0.0002809147374364766,
+      "loss": 0.9465,
       "step": 11640
     },
     {
       "epoch": 2.1927348014304537,
-      "grad_norm": 4.620266914367676,
-      "learning_rate": 1.6843591191417277e-05,
-      "loss": 0.6449,
+      "grad_norm": 1.1538981199264526,
+      "learning_rate": 0.00028072651985695465,
+      "loss": 0.6115,
       "step": 11650
     },
     {
       "epoch": 2.1946169772256727,
-      "grad_norm": 1.4968249797821045,
-      "learning_rate": 1.6832298136645963e-05,
-      "loss": 0.6416,
+      "grad_norm": 0.5456836223602295,
+      "learning_rate": 0.00028053830227743273,
+      "loss": 0.5041,
       "step": 11660
     },
     {
       "epoch": 2.196499153020892,
-      "grad_norm": 5.055369853973389,
-      "learning_rate": 1.6821005081874646e-05,
-      "loss": 0.6069,
+      "grad_norm": 2.0955114364624023,
+      "learning_rate": 0.0002803500846979108,
+      "loss": 0.6258,
       "step": 11670
     },
     {
       "epoch": 2.1983813288161116,
-      "grad_norm": 17.229318618774414,
-      "learning_rate": 1.6809712027103332e-05,
-      "loss": 0.7881,
+      "grad_norm": 2.7322144508361816,
+      "learning_rate": 0.0002801618671183889,
+      "loss": 0.8496,
       "step": 11680
     },
     {
       "epoch": 2.2002635046113306,
-      "grad_norm": 26.358625411987305,
-      "learning_rate": 1.6798418972332018e-05,
-      "loss": 0.9306,
+      "grad_norm": 6.970944404602051,
+      "learning_rate": 0.00027997364953886696,
+      "loss": 0.8897,
       "step": 11690
     },
     {
       "epoch": 2.20214568040655,
-      "grad_norm": 14.272749900817871,
-      "learning_rate": 1.67871259175607e-05,
-      "loss": 0.7933,
+      "grad_norm": 0.6811609864234924,
+      "learning_rate": 0.00027978543195934503,
+      "loss": 0.8129,
       "step": 11700
     },
     {
       "epoch": 2.2040278562017694,
-      "grad_norm": 17.075315475463867,
-      "learning_rate": 1.6775832862789387e-05,
-      "loss": 0.9922,
+      "grad_norm": 5.177496433258057,
+      "learning_rate": 0.0002795972143798231,
+      "loss": 0.8563,
       "step": 11710
     },
     {
       "epoch": 2.2059100319969884,
-      "grad_norm": 18.468908309936523,
-      "learning_rate": 1.676453980801807e-05,
-      "loss": 0.549,
+      "grad_norm": 3.5530927181243896,
+      "learning_rate": 0.00027940899680030113,
+      "loss": 0.5076,
       "step": 11720
     },
     {
       "epoch": 2.207792207792208,
-      "grad_norm": 7.144861698150635,
-      "learning_rate": 1.6753246753246752e-05,
-      "loss": 0.7335,
+      "grad_norm": 2.238463878631592,
+      "learning_rate": 0.0002792207792207792,
+      "loss": 0.7625,
       "step": 11730
     },
     {
       "epoch": 2.2096743835874273,
-      "grad_norm": 20.43572235107422,
-      "learning_rate": 1.6741953698475438e-05,
-      "loss": 0.4813,
+      "grad_norm": 3.0056402683258057,
+      "learning_rate": 0.0002790325616412573,
+      "loss": 0.5276,
       "step": 11740
     },
     {
       "epoch": 2.2115565593826463,
-      "grad_norm": 8.59123420715332,
-      "learning_rate": 1.673066064370412e-05,
-      "loss": 0.8275,
+      "grad_norm": 2.8394482135772705,
+      "learning_rate": 0.00027884434406173536,
+      "loss": 0.8038,
       "step": 11750
     },
     {
       "epoch": 2.2134387351778657,
-      "grad_norm": 17.360458374023438,
-      "learning_rate": 1.6719367588932807e-05,
-      "loss": 0.7514,
+      "grad_norm": 1.930206298828125,
+      "learning_rate": 0.00027865612648221344,
+      "loss": 0.7018,
       "step": 11760
     },
     {
       "epoch": 2.2153209109730847,
-      "grad_norm": 10.042230606079102,
-      "learning_rate": 1.6708074534161493e-05,
-      "loss": 1.0016,
+      "grad_norm": 2.354780912399292,
+      "learning_rate": 0.0002784679089026915,
+      "loss": 0.987,
       "step": 11770
     },
     {
       "epoch": 2.217203086768304,
-      "grad_norm": 19.570775985717773,
-      "learning_rate": 1.6696781479390176e-05,
-      "loss": 0.8327,
+      "grad_norm": 5.227718830108643,
+      "learning_rate": 0.0002782796913231696,
+      "loss": 0.7865,
       "step": 11780
     },
     {
       "epoch": 2.2190852625635236,
-      "grad_norm": 14.94032096862793,
-      "learning_rate": 1.6685488424618862e-05,
-      "loss": 0.6058,
+      "grad_norm": 4.314568042755127,
+      "learning_rate": 0.00027809147374364767,
+      "loss": 0.6195,
       "step": 11790
     },
     {
       "epoch": 2.2209674383587426,
-      "grad_norm": 15.301405906677246,
-      "learning_rate": 1.6674195369847545e-05,
-      "loss": 0.4892,
+      "grad_norm": 2.7393124103546143,
+      "learning_rate": 0.00027790325616412574,
+      "loss": 0.4084,
       "step": 11800
     },
     {
       "epoch": 2.222849614153962,
-      "grad_norm": 18.80755615234375,
-      "learning_rate": 1.6662902315076227e-05,
-      "loss": 0.7236,
+      "grad_norm": 2.8917291164398193,
+      "learning_rate": 0.00027771503858460377,
+      "loss": 0.6465,
       "step": 11810
     },
     {
       "epoch": 2.2247317899491814,
-      "grad_norm": 15.558100700378418,
-      "learning_rate": 1.6651609260304913e-05,
-      "loss": 0.7063,
+      "grad_norm": 1.8676915168762207,
+      "learning_rate": 0.00027752682100508184,
+      "loss": 0.776,
       "step": 11820
     },
     {
       "epoch": 2.2266139657444004,
-      "grad_norm": 13.449424743652344,
-      "learning_rate": 1.6640316205533596e-05,
-      "loss": 0.5049,
+      "grad_norm": 4.098349094390869,
+      "learning_rate": 0.0002773386034255599,
+      "loss": 0.4904,
       "step": 11830
     },
     {
       "epoch": 2.22849614153962,
-      "grad_norm": 10.576664924621582,
-      "learning_rate": 1.6629023150762282e-05,
-      "loss": 1.2078,
+      "grad_norm": 3.7217068672180176,
+      "learning_rate": 0.000277150385846038,
+      "loss": 0.9678,
       "step": 11840
     },
     {
       "epoch": 2.2303783173348393,
-      "grad_norm": 4.189072608947754,
-      "learning_rate": 1.6617730095990965e-05,
-      "loss": 0.7146,
+      "grad_norm": 3.343392848968506,
+      "learning_rate": 0.0002769621682665161,
+      "loss": 0.7159,
       "step": 11850
     },
     {
       "epoch": 2.2322604931300583,
-      "grad_norm": 10.161325454711914,
-      "learning_rate": 1.660643704121965e-05,
-      "loss": 0.6925,
+      "grad_norm": 3.8852624893188477,
+      "learning_rate": 0.0002767739506869942,
+      "loss": 0.6171,
       "step": 11860
     },
     {
       "epoch": 2.2341426689252777,
-      "grad_norm": 29.488740921020508,
-      "learning_rate": 1.6595143986448337e-05,
-      "loss": 0.6859,
+      "grad_norm": 3.444432258605957,
+      "learning_rate": 0.0002765857331074723,
+      "loss": 0.4322,
       "step": 11870
     },
     {
       "epoch": 2.2360248447204967,
-      "grad_norm": 6.638284206390381,
-      "learning_rate": 1.658385093167702e-05,
-      "loss": 0.7455,
+      "grad_norm": 4.453119277954102,
+      "learning_rate": 0.00027639751552795036,
+      "loss": 0.7776,
       "step": 11880
     },
     {
       "epoch": 2.237907020515716,
-      "grad_norm": 14.819140434265137,
-      "learning_rate": 1.6572557876905702e-05,
-      "loss": 0.9682,
+      "grad_norm": 1.7033886909484863,
+      "learning_rate": 0.0002762092979484284,
+      "loss": 0.7152,
       "step": 11890
     },
     {
       "epoch": 2.2397891963109355,
-      "grad_norm": 16.32223129272461,
-      "learning_rate": 1.6561264822134385e-05,
-      "loss": 0.8409,
+      "grad_norm": 3.5321342945098877,
+      "learning_rate": 0.00027602108036890645,
+      "loss": 0.7527,
       "step": 11900
     },
     {
       "epoch": 2.2416713721061545,
-      "grad_norm": 33.57245635986328,
-      "learning_rate": 1.654997176736307e-05,
-      "loss": 0.8191,
+      "grad_norm": 4.50179386138916,
+      "learning_rate": 0.00027583286278938453,
+      "loss": 0.7161,
       "step": 11910
     },
     {
       "epoch": 2.243553547901374,
-      "grad_norm": 11.583626747131348,
-      "learning_rate": 1.6538678712591757e-05,
-      "loss": 0.7672,
+      "grad_norm": 4.627577304840088,
+      "learning_rate": 0.0002756446452098626,
+      "loss": 0.7929,
       "step": 11920
     },
     {
       "epoch": 2.2454357236965934,
-      "grad_norm": 19.42552947998047,
-      "learning_rate": 1.652738565782044e-05,
-      "loss": 0.6141,
+      "grad_norm": 1.926927089691162,
+      "learning_rate": 0.0002754564276303407,
+      "loss": 0.5007,
       "step": 11930
     },
     {
       "epoch": 2.2473178994918124,
-      "grad_norm": 8.740943908691406,
-      "learning_rate": 1.6516092603049126e-05,
-      "loss": 0.5423,
+      "grad_norm": 7.196998119354248,
+      "learning_rate": 0.00027526821005081876,
+      "loss": 0.4507,
       "step": 11940
     },
     {
       "epoch": 2.249200075287032,
-      "grad_norm": 9.475835800170898,
-      "learning_rate": 1.650479954827781e-05,
-      "loss": 0.9297,
+      "grad_norm": 1.706291675567627,
+      "learning_rate": 0.00027507999247129684,
+      "loss": 1.0266,
       "step": 11950
     },
     {
       "epoch": 2.2510822510822512,
-      "grad_norm": 10.173590660095215,
-      "learning_rate": 1.6493506493506495e-05,
-      "loss": 0.6418,
+      "grad_norm": 2.996903657913208,
+      "learning_rate": 0.0002748917748917749,
+      "loss": 0.5698,
       "step": 11960
     },
     {
       "epoch": 2.2529644268774702,
-      "grad_norm": 5.159122467041016,
-      "learning_rate": 1.648221343873518e-05,
-      "loss": 0.4525,
+      "grad_norm": 1.0445246696472168,
+      "learning_rate": 0.000274703557312253,
+      "loss": 0.382,
       "step": 11970
     },
     {
       "epoch": 2.2548466026726897,
-      "grad_norm": 20.905466079711914,
-      "learning_rate": 1.647092038396386e-05,
-      "loss": 0.4453,
+      "grad_norm": 3.8929078578948975,
+      "learning_rate": 0.000274515339732731,
+      "loss": 0.3706,
       "step": 11980
     },
     {
       "epoch": 2.256728778467909,
-      "grad_norm": 14.282583236694336,
-      "learning_rate": 1.6459627329192546e-05,
-      "loss": 0.6531,
+      "grad_norm": 3.6355106830596924,
+      "learning_rate": 0.0002743271221532091,
+      "loss": 0.6808,
       "step": 11990
     },
     {
       "epoch": 2.258610954263128,
-      "grad_norm": 4.331676006317139,
-      "learning_rate": 1.6448334274421232e-05,
-      "loss": 0.6144,
+      "grad_norm": 0.9284248352050781,
+      "learning_rate": 0.00027413890457368716,
+      "loss": 0.5733,
       "step": 12000
     },
     {
       "epoch": 2.2604931300583475,
-      "grad_norm": 18.63337516784668,
-      "learning_rate": 1.6437041219649915e-05,
-      "loss": 0.9553,
+      "grad_norm": 4.699124336242676,
+      "learning_rate": 0.00027395068699416524,
+      "loss": 0.8723,
       "step": 12010
     },
     {
       "epoch": 2.262375305853567,
-      "grad_norm": 8.563776016235352,
-      "learning_rate": 1.64257481648786e-05,
-      "loss": 0.5918,
+      "grad_norm": 2.852851152420044,
+      "learning_rate": 0.0002737624694146433,
+      "loss": 0.6141,
       "step": 12020
     },
     {
       "epoch": 2.264257481648786,
-      "grad_norm": 26.9168758392334,
-      "learning_rate": 1.6414455110107284e-05,
-      "loss": 0.9051,
+      "grad_norm": 3.81124210357666,
+      "learning_rate": 0.0002735742518351214,
+      "loss": 0.8832,
       "step": 12030
     },
     {
       "epoch": 2.2661396574440054,
-      "grad_norm": 20.256752014160156,
-      "learning_rate": 1.640316205533597e-05,
-      "loss": 0.6102,
+      "grad_norm": 2.0278055667877197,
+      "learning_rate": 0.00027338603425559947,
+      "loss": 0.49,
       "step": 12040
     },
     {
       "epoch": 2.2680218332392243,
-      "grad_norm": 0.3493950366973877,
-      "learning_rate": 1.6391869000564656e-05,
-      "loss": 0.5125,
+      "grad_norm": 0.07482793182134628,
+      "learning_rate": 0.00027319781667607755,
+      "loss": 0.5725,
       "step": 12050
     },
     {
       "epoch": 2.269904009034444,
-      "grad_norm": 2.2542970180511475,
-      "learning_rate": 1.638057594579334e-05,
-      "loss": 0.7501,
+      "grad_norm": 1.8518965244293213,
+      "learning_rate": 0.0002730095990965556,
+      "loss": 0.6832,
       "step": 12060
     },
     {
       "epoch": 2.271786184829663,
-      "grad_norm": 8.313566207885742,
-      "learning_rate": 1.636928289102202e-05,
-      "loss": 0.6996,
+      "grad_norm": 1.8055180311203003,
+      "learning_rate": 0.0002728213815170337,
+      "loss": 0.6431,
       "step": 12070
     },
     {
       "epoch": 2.273668360624882,
-      "grad_norm": 7.590768337249756,
-      "learning_rate": 1.6357989836250704e-05,
-      "loss": 0.6913,
+      "grad_norm": 1.2969673871994019,
+      "learning_rate": 0.0002726331639375118,
+      "loss": 0.5825,
       "step": 12080
     },
     {
       "epoch": 2.2755505364201016,
-      "grad_norm": 17.979164123535156,
-      "learning_rate": 1.634669678147939e-05,
-      "loss": 0.5275,
+      "grad_norm": 4.328277587890625,
+      "learning_rate": 0.00027244494635798985,
+      "loss": 0.4851,
       "step": 12090
     },
     {
       "epoch": 2.277432712215321,
-      "grad_norm": 30.397066116333008,
-      "learning_rate": 1.6335403726708076e-05,
-      "loss": 0.8555,
+      "grad_norm": 4.203436851501465,
+      "learning_rate": 0.00027225672877846793,
+      "loss": 0.7723,
       "step": 12100
     },
     {
       "epoch": 2.27931488801054,
-      "grad_norm": 8.03955364227295,
-      "learning_rate": 1.632411067193676e-05,
-      "loss": 0.6765,
+      "grad_norm": 2.0930275917053223,
+      "learning_rate": 0.000272068511198946,
+      "loss": 0.6907,
       "step": 12110
     },
     {
       "epoch": 2.2811970638057595,
-      "grad_norm": 8.310486793518066,
-      "learning_rate": 1.6312817617165445e-05,
-      "loss": 0.8106,
+      "grad_norm": 1.692587971687317,
+      "learning_rate": 0.0002718802936194241,
+      "loss": 0.6536,
       "step": 12120
     },
     {
       "epoch": 2.283079239600979,
-      "grad_norm": 5.86186408996582,
-      "learning_rate": 1.6301524562394127e-05,
-      "loss": 0.4987,
+      "grad_norm": 1.9245163202285767,
+      "learning_rate": 0.00027169207603990216,
+      "loss": 0.4055,
       "step": 12130
     },
     {
       "epoch": 2.284961415396198,
-      "grad_norm": 19.821392059326172,
-      "learning_rate": 1.6290231507622814e-05,
-      "loss": 0.5728,
+      "grad_norm": 4.446379661560059,
+      "learning_rate": 0.00027150385846038023,
+      "loss": 0.4683,
       "step": 12140
     },
     {
       "epoch": 2.2868435911914173,
-      "grad_norm": 12.544803619384766,
-      "learning_rate": 1.62789384528515e-05,
-      "loss": 0.7435,
+      "grad_norm": 3.3666815757751465,
+      "learning_rate": 0.0002713156408808583,
+      "loss": 0.81,
       "step": 12150
     },
     {
       "epoch": 2.2887257669866363,
-      "grad_norm": 13.145238876342773,
-      "learning_rate": 1.626764539808018e-05,
-      "loss": 0.4866,
+      "grad_norm": 6.301428318023682,
+      "learning_rate": 0.00027112742330133633,
+      "loss": 0.5008,
       "step": 12160
     },
     {
       "epoch": 2.2906079427818558,
-      "grad_norm": 15.785924911499023,
-      "learning_rate": 1.6256352343308865e-05,
-      "loss": 0.8673,
+      "grad_norm": 4.091559410095215,
+      "learning_rate": 0.0002709392057218144,
+      "loss": 0.8201,
       "step": 12170
     },
     {
       "epoch": 2.292490118577075,
-      "grad_norm": 9.898990631103516,
-      "learning_rate": 1.6245059288537548e-05,
-      "loss": 0.7766,
+      "grad_norm": 2.8951735496520996,
+      "learning_rate": 0.0002707509881422925,
+      "loss": 0.6976,
       "step": 12180
     },
     {
       "epoch": 2.2943722943722946,
-      "grad_norm": 14.689043998718262,
-      "learning_rate": 1.6233766233766234e-05,
-      "loss": 0.9227,
+      "grad_norm": 2.0975356101989746,
+      "learning_rate": 0.00027056277056277056,
+      "loss": 0.7667,
       "step": 12190
     },
     {
       "epoch": 2.2962544701675136,
-      "grad_norm": 6.211693286895752,
-      "learning_rate": 1.622247317899492e-05,
-      "loss": 0.5012,
+      "grad_norm": 2.5957348346710205,
+      "learning_rate": 0.00027037455298324864,
+      "loss": 0.4919,
       "step": 12200
     },
     {
       "epoch": 2.298136645962733,
-      "grad_norm": 6.014758586883545,
-      "learning_rate": 1.6211180124223603e-05,
-      "loss": 0.8023,
+      "grad_norm": 1.4557297229766846,
+      "learning_rate": 0.0002701863354037267,
+      "loss": 0.6835,
       "step": 12210
     },
     {
       "epoch": 2.300018821757952,
-      "grad_norm": 32.03548812866211,
-      "learning_rate": 1.619988706945229e-05,
-      "loss": 0.86,
+      "grad_norm": 2.45815110206604,
+      "learning_rate": 0.0002699981178242048,
+      "loss": 0.8599,
       "step": 12220
     },
     {
       "epoch": 2.3019009975531715,
-      "grad_norm": 25.236791610717773,
-      "learning_rate": 1.6188594014680975e-05,
-      "loss": 0.7572,
+      "grad_norm": 3.9854960441589355,
+      "learning_rate": 0.00026980990024468287,
+      "loss": 0.6801,
       "step": 12230
     },
     {
       "epoch": 2.303783173348391,
-      "grad_norm": 7.06341552734375,
-      "learning_rate": 1.6177300959909654e-05,
-      "loss": 0.6443,
+      "grad_norm": 1.293655514717102,
+      "learning_rate": 0.0002696216826651609,
+      "loss": 0.6667,
       "step": 12240
     },
     {
       "epoch": 2.30566534914361,
-      "grad_norm": 3.819842576980591,
-      "learning_rate": 1.616600790513834e-05,
-      "loss": 0.6632,
+      "grad_norm": 2.157909870147705,
+      "learning_rate": 0.00026943346508563897,
+      "loss": 0.7323,
       "step": 12250
     },
     {
       "epoch": 2.3075475249388293,
-      "grad_norm": 11.308211326599121,
-      "learning_rate": 1.6154714850367023e-05,
-      "loss": 0.7541,
+      "grad_norm": 2.9319350719451904,
+      "learning_rate": 0.00026924524750611704,
+      "loss": 0.7194,
       "step": 12260
     },
     {
       "epoch": 2.3094297007340487,
-      "grad_norm": 9.683237075805664,
-      "learning_rate": 1.614342179559571e-05,
-      "loss": 0.9261,
+      "grad_norm": 3.461108922958374,
+      "learning_rate": 0.0002690570299265951,
+      "loss": 0.8909,
       "step": 12270
     },
     {
       "epoch": 2.3113118765292677,
-      "grad_norm": 21.996740341186523,
-      "learning_rate": 1.6132128740824395e-05,
-      "loss": 0.6662,
+      "grad_norm": 2.8008790016174316,
+      "learning_rate": 0.0002688688123470732,
+      "loss": 0.6946,
       "step": 12280
     },
     {
       "epoch": 2.313194052324487,
-      "grad_norm": 8.742904663085938,
-      "learning_rate": 1.6120835686053078e-05,
-      "loss": 0.7921,
+      "grad_norm": 2.3989953994750977,
+      "learning_rate": 0.0002686805947675513,
+      "loss": 0.9312,
       "step": 12290
     },
     {
       "epoch": 2.3150762281197066,
-      "grad_norm": 1.3880571126937866,
-      "learning_rate": 1.6109542631281764e-05,
-      "loss": 0.8751,
+      "grad_norm": 0.9723449945449829,
+      "learning_rate": 0.0002684923771880294,
+      "loss": 0.8079,
       "step": 12300
     },
     {
       "epoch": 2.3169584039149256,
-      "grad_norm": 26.320383071899414,
-      "learning_rate": 1.6098249576510446e-05,
-      "loss": 0.6595,
+      "grad_norm": 2.027454376220703,
+      "learning_rate": 0.0002683041596085075,
+      "loss": 0.5137,
       "step": 12310
     },
     {
       "epoch": 2.318840579710145,
-      "grad_norm": 34.07208251953125,
-      "learning_rate": 1.6086956521739132e-05,
-      "loss": 0.7587,
+      "grad_norm": 3.510030746459961,
+      "learning_rate": 0.00026811594202898556,
+      "loss": 0.6373,
       "step": 12320
     },
     {
       "epoch": 2.320722755505364,
-      "grad_norm": 7.523997783660889,
-      "learning_rate": 1.6075663466967815e-05,
-      "loss": 0.6733,
+      "grad_norm": 2.559351682662964,
+      "learning_rate": 0.0002679277244494636,
+      "loss": 0.6808,
       "step": 12330
     },
     {
       "epoch": 2.3226049313005834,
-      "grad_norm": 18.36374855041504,
-      "learning_rate": 1.6064370412196498e-05,
-      "loss": 0.8033,
+      "grad_norm": 3.51894474029541,
+      "learning_rate": 0.00026773950686994165,
+      "loss": 0.7996,
       "step": 12340
     },
     {
       "epoch": 2.324487107095803,
-      "grad_norm": 13.632719039916992,
-      "learning_rate": 1.6053077357425184e-05,
-      "loss": 0.9909,
+      "grad_norm": 3.97428822517395,
+      "learning_rate": 0.00026755128929041973,
+      "loss": 1.1201,
       "step": 12350
     },
     {
       "epoch": 2.326369282891022,
-      "grad_norm": 13.183788299560547,
-      "learning_rate": 1.6041784302653867e-05,
-      "loss": 0.8184,
+      "grad_norm": 2.7532434463500977,
+      "learning_rate": 0.0002673630717108978,
+      "loss": 0.8154,
       "step": 12360
     },
     {
       "epoch": 2.3282514586862413,
-      "grad_norm": 2.855896472930908,
-      "learning_rate": 1.6030491247882553e-05,
-      "loss": 0.5908,
+      "grad_norm": 4.514427185058594,
+      "learning_rate": 0.0002671748541313759,
+      "loss": 0.6268,
       "step": 12370
     },
     {
       "epoch": 2.3301336344814607,
-      "grad_norm": 2.077853202819824,
-      "learning_rate": 1.601919819311124e-05,
-      "loss": 0.5688,
+      "grad_norm": 1.4542497396469116,
+      "learning_rate": 0.00026698663655185396,
+      "loss": 0.6466,
       "step": 12380
     },
     {
       "epoch": 2.3320158102766797,
-      "grad_norm": 10.118651390075684,
-      "learning_rate": 1.600790513833992e-05,
-      "loss": 0.7142,
+      "grad_norm": 1.8610076904296875,
+      "learning_rate": 0.00026679841897233204,
+      "loss": 0.6654,
       "step": 12390
     },
     {
       "epoch": 2.333897986071899,
-      "grad_norm": 11.305803298950195,
-      "learning_rate": 1.5996612083568607e-05,
-      "loss": 1.1076,
+      "grad_norm": 2.503138303756714,
+      "learning_rate": 0.0002666102013928101,
+      "loss": 1.1184,
       "step": 12400
     },
     {
       "epoch": 2.3357801618671186,
-      "grad_norm": 10.361531257629395,
-      "learning_rate": 1.598531902879729e-05,
-      "loss": 0.7602,
+      "grad_norm": 4.176797389984131,
+      "learning_rate": 0.0002664219838132882,
+      "loss": 0.7134,
       "step": 12410
     },
     {
       "epoch": 2.3376623376623376,
-      "grad_norm": 20.7995662689209,
-      "learning_rate": 1.5974025974025973e-05,
-      "loss": 0.6195,
+      "grad_norm": 3.3237433433532715,
+      "learning_rate": 0.0002662337662337662,
+      "loss": 0.6421,
       "step": 12420
     },
     {
       "epoch": 2.339544513457557,
-      "grad_norm": 8.287638664245605,
-      "learning_rate": 1.596273291925466e-05,
-      "loss": 0.8501,
+      "grad_norm": 1.97113037109375,
+      "learning_rate": 0.0002660455486542443,
+      "loss": 0.8516,
       "step": 12430
     },
     {
       "epoch": 2.341426689252776,
-      "grad_norm": 20.09409523010254,
-      "learning_rate": 1.595143986448334e-05,
-      "loss": 0.8378,
+      "grad_norm": 3.358541488647461,
+      "learning_rate": 0.00026585733107472236,
+      "loss": 0.8102,
       "step": 12440
     },
     {
       "epoch": 2.3433088650479954,
-      "grad_norm": 26.83869171142578,
-      "learning_rate": 1.5940146809712028e-05,
-      "loss": 0.9183,
+      "grad_norm": 7.149372100830078,
+      "learning_rate": 0.00026566911349520044,
+      "loss": 0.9773,
       "step": 12450
     },
     {
       "epoch": 2.345191040843215,
-      "grad_norm": 0.30550795793533325,
-      "learning_rate": 1.592885375494071e-05,
-      "loss": 1.0637,
+      "grad_norm": 0.12019229680299759,
+      "learning_rate": 0.0002654808959156785,
+      "loss": 0.9357,
       "step": 12460
     },
     {
       "epoch": 2.3470732166384343,
-      "grad_norm": 2.1787874698638916,
-      "learning_rate": 1.5917560700169396e-05,
-      "loss": 0.7166,
+      "grad_norm": 2.3379759788513184,
+      "learning_rate": 0.0002652926783361566,
+      "loss": 0.5609,
       "step": 12470
     },
     {
       "epoch": 2.3489553924336533,
-      "grad_norm": 9.78334903717041,
-      "learning_rate": 1.5906267645398083e-05,
-      "loss": 1.08,
+      "grad_norm": 2.5689191818237305,
+      "learning_rate": 0.00026510446075663467,
+      "loss": 1.118,
       "step": 12480
     },
     {
       "epoch": 2.3508375682288727,
-      "grad_norm": 26.472990036010742,
-      "learning_rate": 1.5894974590626765e-05,
-      "loss": 0.5742,
+      "grad_norm": 4.990429401397705,
+      "learning_rate": 0.00026491624317711275,
+      "loss": 0.4498,
       "step": 12490
     },
     {
       "epoch": 2.3527197440240917,
-      "grad_norm": 7.039992332458496,
-      "learning_rate": 1.5883681535855448e-05,
-      "loss": 0.4755,
+      "grad_norm": 5.412718772888184,
+      "learning_rate": 0.00026472802559759077,
+      "loss": 0.458,
       "step": 12500
     },
     {
       "epoch": 2.354601919819311,
-      "grad_norm": 12.98085880279541,
-      "learning_rate": 1.5872388481084134e-05,
-      "loss": 0.9198,
+      "grad_norm": 4.076390743255615,
+      "learning_rate": 0.0002645398080180689,
+      "loss": 0.9594,
       "step": 12510
     },
     {
       "epoch": 2.3564840956145305,
-      "grad_norm": 19.38686752319336,
-      "learning_rate": 1.5861095426312817e-05,
-      "loss": 0.512,
+      "grad_norm": 5.14693546295166,
+      "learning_rate": 0.000264351590438547,
+      "loss": 0.6583,
       "step": 12520
     },
     {
       "epoch": 2.3583662714097495,
-      "grad_norm": 16.51802635192871,
-      "learning_rate": 1.5849802371541503e-05,
-      "loss": 0.7799,
+      "grad_norm": 2.6487724781036377,
+      "learning_rate": 0.00026416337285902505,
+      "loss": 0.8533,
       "step": 12530
     },
     {
       "epoch": 2.360248447204969,
-      "grad_norm": 5.628615379333496,
-      "learning_rate": 1.5838509316770185e-05,
-      "loss": 0.8057,
+      "grad_norm": 1.8451364040374756,
+      "learning_rate": 0.00026397515527950313,
+      "loss": 0.7168,
       "step": 12540
     },
     {
       "epoch": 2.3621306230001884,
-      "grad_norm": 25.433387756347656,
-      "learning_rate": 1.582721626199887e-05,
-      "loss": 0.6807,
+      "grad_norm": 7.2082977294921875,
+      "learning_rate": 0.0002637869376999812,
+      "loss": 0.7608,
       "step": 12550
     },
     {
       "epoch": 2.3640127987954074,
-      "grad_norm": 11.819149017333984,
-      "learning_rate": 1.5815923207227558e-05,
-      "loss": 0.6426,
+      "grad_norm": 3.3436646461486816,
+      "learning_rate": 0.0002635987201204593,
+      "loss": 0.8196,
       "step": 12560
     },
     {
       "epoch": 2.365894974590627,
-      "grad_norm": 18.320758819580078,
-      "learning_rate": 1.580463015245624e-05,
-      "loss": 0.5381,
+      "grad_norm": 6.613883018493652,
+      "learning_rate": 0.00026341050254093736,
+      "loss": 0.5313,
       "step": 12570
     },
     {
       "epoch": 2.3677771503858462,
-      "grad_norm": 21.337982177734375,
-      "learning_rate": 1.5793337097684926e-05,
-      "loss": 0.6222,
+      "grad_norm": 5.031750679016113,
+      "learning_rate": 0.00026322228496141543,
+      "loss": 0.505,
       "step": 12580
     },
     {
       "epoch": 2.3696593261810652,
-      "grad_norm": 30.114587783813477,
-      "learning_rate": 1.5782044042913606e-05,
-      "loss": 0.8209,
+      "grad_norm": 3.989000082015991,
+      "learning_rate": 0.00026303406738189346,
+      "loss": 0.82,
       "step": 12590
     },
     {
       "epoch": 2.3715415019762847,
-      "grad_norm": 13.773634910583496,
-      "learning_rate": 1.5770750988142292e-05,
-      "loss": 0.7724,
+      "grad_norm": 3.471708297729492,
+      "learning_rate": 0.00026284584980237153,
+      "loss": 0.7288,
       "step": 12600
     },
     {
       "epoch": 2.3734236777715036,
-      "grad_norm": 7.200165748596191,
-      "learning_rate": 1.5759457933370978e-05,
-      "loss": 0.9503,
+      "grad_norm": 2.664614200592041,
+      "learning_rate": 0.0002626576322228496,
+      "loss": 0.8995,
       "step": 12610
     },
     {
       "epoch": 2.375305853566723,
-      "grad_norm": 16.634714126586914,
-      "learning_rate": 1.574816487859966e-05,
-      "loss": 0.5145,
+      "grad_norm": 2.755415439605713,
+      "learning_rate": 0.0002624694146433277,
+      "loss": 0.6285,
       "step": 12620
     },
     {
       "epoch": 2.3771880293619425,
-      "grad_norm": 13.062280654907227,
-      "learning_rate": 1.5736871823828347e-05,
-      "loss": 0.584,
+      "grad_norm": 4.312176704406738,
+      "learning_rate": 0.00026228119706380576,
+      "loss": 0.6126,
       "step": 12630
     },
     {
       "epoch": 2.3790702051571615,
-      "grad_norm": 15.320236206054688,
-      "learning_rate": 1.572557876905703e-05,
-      "loss": 1.109,
+      "grad_norm": 4.8685736656188965,
+      "learning_rate": 0.00026209297948428384,
+      "loss": 1.2171,
       "step": 12640
     },
     {
       "epoch": 2.380952380952381,
-      "grad_norm": 18.621957778930664,
-      "learning_rate": 1.5714285714285715e-05,
-      "loss": 0.5475,
+      "grad_norm": 1.3688360452651978,
+      "learning_rate": 0.0002619047619047619,
+      "loss": 0.6181,
       "step": 12650
     },
     {
       "epoch": 2.3828345567476004,
-      "grad_norm": 21.138986587524414,
-      "learning_rate": 1.57029926595144e-05,
-      "loss": 0.6148,
+      "grad_norm": 5.439986705780029,
+      "learning_rate": 0.00026171654432524,
+      "loss": 0.4511,
       "step": 12660
     },
     {
       "epoch": 2.3847167325428194,
-      "grad_norm": 7.1945929527282715,
-      "learning_rate": 1.5691699604743084e-05,
-      "loss": 0.6836,
+      "grad_norm": 2.772137403488159,
+      "learning_rate": 0.00026152832674571807,
+      "loss": 0.6025,
       "step": 12670
     },
     {
       "epoch": 2.386598908338039,
-      "grad_norm": 7.824986457824707,
-      "learning_rate": 1.5680406549971767e-05,
-      "loss": 1.0433,
+      "grad_norm": 3.3040308952331543,
+      "learning_rate": 0.0002613401091661961,
+      "loss": 1.046,
       "step": 12680
     },
     {
       "epoch": 2.388481084133258,
-      "grad_norm": 26.856182098388672,
-      "learning_rate": 1.566911349520045e-05,
-      "loss": 0.9668,
+      "grad_norm": 1.7851004600524902,
+      "learning_rate": 0.00026115189158667417,
+      "loss": 0.6861,
       "step": 12690
     },
     {
       "epoch": 2.390363259928477,
-      "grad_norm": 1.3294514417648315,
-      "learning_rate": 1.5657820440429136e-05,
-      "loss": 0.6016,
+      "grad_norm": 1.1720726490020752,
+      "learning_rate": 0.00026096367400715224,
+      "loss": 0.5638,
       "step": 12700
     },
     {
       "epoch": 2.3922454357236966,
-      "grad_norm": 25.570005416870117,
-      "learning_rate": 1.564652738565782e-05,
-      "loss": 0.6145,
+      "grad_norm": 2.6076951026916504,
+      "learning_rate": 0.0002607754564276303,
+      "loss": 0.8071,
       "step": 12710
     },
     {
       "epoch": 2.3941276115189156,
-      "grad_norm": 22.4234676361084,
-      "learning_rate": 1.5635234330886504e-05,
-      "loss": 0.7482,
+      "grad_norm": 2.853635549545288,
+      "learning_rate": 0.0002605872388481084,
+      "loss": 0.7192,
       "step": 12720
     },
     {
       "epoch": 2.396009787314135,
-      "grad_norm": 14.264887809753418,
-      "learning_rate": 1.562394127611519e-05,
-      "loss": 0.9606,
+      "grad_norm": 3.179858922958374,
+      "learning_rate": 0.0002603990212685865,
+      "loss": 0.8151,
       "step": 12730
     },
     {
       "epoch": 2.3978919631093545,
-      "grad_norm": 23.162593841552734,
-      "learning_rate": 1.5612648221343876e-05,
-      "loss": 0.6843,
+      "grad_norm": 3.948190689086914,
+      "learning_rate": 0.0002602108036890646,
+      "loss": 0.5294,
       "step": 12740
     },
     {
       "epoch": 2.399774138904574,
-      "grad_norm": 30.500329971313477,
-      "learning_rate": 1.560135516657256e-05,
-      "loss": 0.8691,
+      "grad_norm": 3.1576452255249023,
+      "learning_rate": 0.0002600225861095427,
+      "loss": 0.7788,
       "step": 12750
     },
     {
       "epoch": 2.401656314699793,
-      "grad_norm": 10.762563705444336,
-      "learning_rate": 1.5590062111801242e-05,
-      "loss": 0.7518,
+      "grad_norm": 4.19622278213501,
+      "learning_rate": 0.0002598343685300207,
+      "loss": 0.7703,
       "step": 12760
     },
     {
       "epoch": 2.4035384904950123,
-      "grad_norm": 38.12343978881836,
-      "learning_rate": 1.5578769057029925e-05,
-      "loss": 0.8324,
+      "grad_norm": 5.953280925750732,
+      "learning_rate": 0.0002596461509504988,
+      "loss": 0.8527,
       "step": 12770
     },
     {
       "epoch": 2.4054206662902313,
-      "grad_norm": 26.032785415649414,
-      "learning_rate": 1.556747600225861e-05,
-      "loss": 0.9027,
+      "grad_norm": 5.18197774887085,
+      "learning_rate": 0.00025945793337097685,
+      "loss": 0.9497,
       "step": 12780
     },
     {
       "epoch": 2.4073028420854508,
-      "grad_norm": 12.26636028289795,
-      "learning_rate": 1.5556182947487297e-05,
-      "loss": 0.6673,
+      "grad_norm": 2.9982385635375977,
+      "learning_rate": 0.00025926971579145493,
+      "loss": 0.8016,
       "step": 12790
     },
     {
       "epoch": 2.40918501788067,
-      "grad_norm": 22.655187606811523,
-      "learning_rate": 1.554488989271598e-05,
-      "loss": 0.9456,
+      "grad_norm": 3.964134931564331,
+      "learning_rate": 0.000259081498211933,
+      "loss": 1.0195,
       "step": 12800
     },
     {
       "epoch": 2.411067193675889,
-      "grad_norm": 8.51685619354248,
-      "learning_rate": 1.5533596837944665e-05,
-      "loss": 0.6657,
+      "grad_norm": 1.3622925281524658,
+      "learning_rate": 0.0002588932806324111,
+      "loss": 0.7713,
       "step": 12810
     },
     {
       "epoch": 2.4129493694711086,
-      "grad_norm": 8.790712356567383,
-      "learning_rate": 1.5522303783173348e-05,
-      "loss": 0.6555,
+      "grad_norm": 0.5836087465286255,
+      "learning_rate": 0.00025870506305288916,
+      "loss": 0.5852,
       "step": 12820
     },
     {
       "epoch": 2.414831545266328,
-      "grad_norm": 24.235885620117188,
-      "learning_rate": 1.5511010728402034e-05,
-      "loss": 0.5839,
+      "grad_norm": 0.42106154561042786,
+      "learning_rate": 0.00025851684547336724,
+      "loss": 0.744,
       "step": 12830
     },
     {
       "epoch": 2.416713721061547,
-      "grad_norm": 11.751230239868164,
-      "learning_rate": 1.549971767363072e-05,
-      "loss": 0.53,
+      "grad_norm": 4.040215492248535,
+      "learning_rate": 0.0002583286278938453,
+      "loss": 0.6945,
       "step": 12840
     },
     {
       "epoch": 2.4185958968567665,
-      "grad_norm": 11.033337593078613,
-      "learning_rate": 1.54884246188594e-05,
-      "loss": 0.7369,
+      "grad_norm": 1.9589953422546387,
+      "learning_rate": 0.00025814041031432334,
+      "loss": 0.7177,
       "step": 12850
     },
     {
       "epoch": 2.420478072651986,
-      "grad_norm": 11.11131763458252,
-      "learning_rate": 1.5477131564088086e-05,
-      "loss": 0.5856,
+      "grad_norm": 13.924626350402832,
+      "learning_rate": 0.0002579521927348014,
+      "loss": 0.5441,
       "step": 12860
     },
     {
       "epoch": 2.422360248447205,
-      "grad_norm": 31.728572845458984,
-      "learning_rate": 1.546583850931677e-05,
-      "loss": 0.8794,
+      "grad_norm": 4.234344959259033,
+      "learning_rate": 0.0002577639751552795,
+      "loss": 0.7613,
       "step": 12870
     },
     {
       "epoch": 2.4242424242424243,
-      "grad_norm": 15.746371269226074,
-      "learning_rate": 1.5454545454545454e-05,
-      "loss": 0.6741,
+      "grad_norm": 4.341315746307373,
+      "learning_rate": 0.00025757575757575756,
+      "loss": 0.6245,
       "step": 12880
     },
     {
       "epoch": 2.4261246000376433,
-      "grad_norm": 7.750115394592285,
-      "learning_rate": 1.544325239977414e-05,
-      "loss": 0.9826,
+      "grad_norm": 4.634925365447998,
+      "learning_rate": 0.00025738753999623564,
+      "loss": 0.9347,
       "step": 12890
     },
     {
       "epoch": 2.4280067758328627,
-      "grad_norm": 12.74660873413086,
-      "learning_rate": 1.5431959345002823e-05,
-      "loss": 0.7174,
+      "grad_norm": 4.451136589050293,
+      "learning_rate": 0.0002571993224167137,
+      "loss": 0.7098,
       "step": 12900
     },
     {
       "epoch": 2.429888951628082,
-      "grad_norm": 10.540325164794922,
-      "learning_rate": 1.542066629023151e-05,
-      "loss": 0.6789,
+      "grad_norm": 2.306314468383789,
+      "learning_rate": 0.0002570111048371918,
+      "loss": 0.8713,
       "step": 12910
     },
     {
       "epoch": 2.431771127423301,
-      "grad_norm": 16.440242767333984,
-      "learning_rate": 1.5409373235460192e-05,
-      "loss": 0.5718,
+      "grad_norm": 2.34293270111084,
+      "learning_rate": 0.00025682288725766987,
+      "loss": 0.5413,
       "step": 12920
     },
     {
       "epoch": 2.4336533032185206,
-      "grad_norm": 6.608902931213379,
-      "learning_rate": 1.5398080180688878e-05,
-      "loss": 0.7068,
+      "grad_norm": 2.1198337078094482,
+      "learning_rate": 0.00025663466967814795,
+      "loss": 0.7821,
       "step": 12930
     },
     {
       "epoch": 2.43553547901374,
-      "grad_norm": 9.328181266784668,
-      "learning_rate": 1.538678712591756e-05,
-      "loss": 0.5848,
+      "grad_norm": 2.277205228805542,
+      "learning_rate": 0.00025644645209862597,
+      "loss": 0.6455,
       "step": 12940
     },
     {
       "epoch": 2.437417654808959,
-      "grad_norm": 3.718935966491699,
-      "learning_rate": 1.5375494071146243e-05,
-      "loss": 0.7668,
+      "grad_norm": 1.2600597143173218,
+      "learning_rate": 0.00025625823451910405,
+      "loss": 0.8161,
       "step": 12950
     },
     {
       "epoch": 2.4392998306041784,
-      "grad_norm": 12.87177848815918,
-      "learning_rate": 1.536420101637493e-05,
-      "loss": 0.5857,
+      "grad_norm": 2.1574079990386963,
+      "learning_rate": 0.0002560700169395822,
+      "loss": 0.4708,
       "step": 12960
     },
     {
       "epoch": 2.441182006399398,
-      "grad_norm": 2.911306142807007,
-      "learning_rate": 1.5352907961603616e-05,
-      "loss": 0.8994,
+      "grad_norm": 1.071172833442688,
+      "learning_rate": 0.00025588179936006025,
+      "loss": 0.7987,
       "step": 12970
     },
     {
       "epoch": 2.443064182194617,
-      "grad_norm": 10.852980613708496,
-      "learning_rate": 1.5341614906832298e-05,
-      "loss": 0.8624,
+      "grad_norm": 3.36719012260437,
+      "learning_rate": 0.00025569358178053833,
+      "loss": 0.79,
       "step": 12980
     },
     {
       "epoch": 2.4449463579898363,
-      "grad_norm": 5.891432285308838,
-      "learning_rate": 1.5330321852060984e-05,
-      "loss": 0.7049,
+      "grad_norm": 4.198141098022461,
+      "learning_rate": 0.0002555053642010164,
+      "loss": 0.764,
       "step": 12990
     },
     {
       "epoch": 2.4468285337850557,
-      "grad_norm": 7.124139785766602,
-      "learning_rate": 1.5319028797289667e-05,
-      "loss": 0.7183,
+      "grad_norm": 2.2598376274108887,
+      "learning_rate": 0.0002553171466214945,
+      "loss": 0.6441,
       "step": 13000
     },
     {
       "epoch": 2.4487107095802747,
-      "grad_norm": 11.209503173828125,
-      "learning_rate": 1.5307735742518353e-05,
-      "loss": 0.5121,
+      "grad_norm": 4.295562267303467,
+      "learning_rate": 0.00025512892904197256,
+      "loss": 0.5728,
       "step": 13010
     },
     {
       "epoch": 2.450592885375494,
-      "grad_norm": 12.793279647827148,
-      "learning_rate": 1.529644268774704e-05,
-      "loss": 0.6083,
+      "grad_norm": 5.111675262451172,
+      "learning_rate": 0.00025494071146245063,
+      "loss": 0.6935,
       "step": 13020
     },
     {
       "epoch": 2.4524750611707136,
-      "grad_norm": 16.64006996154785,
-      "learning_rate": 1.528514963297572e-05,
-      "loss": 0.3401,
+      "grad_norm": 4.394181251525879,
+      "learning_rate": 0.00025475249388292866,
+      "loss": 0.4368,
       "step": 13030
     },
     {
       "epoch": 2.4543572369659326,
-      "grad_norm": 3.3853352069854736,
-      "learning_rate": 1.5273856578204405e-05,
-      "loss": 0.5073,
+      "grad_norm": 1.7228174209594727,
+      "learning_rate": 0.00025456427630340673,
+      "loss": 0.5746,
       "step": 13040
     },
     {
       "epoch": 2.456239412761152,
-      "grad_norm": 11.800168991088867,
-      "learning_rate": 1.5262563523433087e-05,
-      "loss": 0.4107,
+      "grad_norm": 1.8490511178970337,
+      "learning_rate": 0.0002543760587238848,
+      "loss": 0.432,
       "step": 13050
     },
     {
       "epoch": 2.458121588556371,
-      "grad_norm": 17.713973999023438,
-      "learning_rate": 1.5251270468661773e-05,
-      "loss": 0.4237,
+      "grad_norm": 2.733003854751587,
+      "learning_rate": 0.0002541878411443629,
+      "loss": 0.3822,
       "step": 13060
     },
     {
       "epoch": 2.4600037643515904,
-      "grad_norm": 14.999812126159668,
-      "learning_rate": 1.5239977413890458e-05,
-      "loss": 0.8685,
+      "grad_norm": 2.048168420791626,
+      "learning_rate": 0.00025399962356484096,
+      "loss": 0.758,
       "step": 13070
     },
     {
       "epoch": 2.46188594014681,
-      "grad_norm": 10.52674674987793,
-      "learning_rate": 1.5228684359119142e-05,
-      "loss": 0.8917,
+      "grad_norm": 1.9236838817596436,
+      "learning_rate": 0.00025381140598531904,
+      "loss": 0.8686,
       "step": 13080
     },
     {
       "epoch": 2.463768115942029,
-      "grad_norm": 21.042428970336914,
-      "learning_rate": 1.5217391304347828e-05,
-      "loss": 0.7128,
+      "grad_norm": 3.5483946800231934,
+      "learning_rate": 0.0002536231884057971,
+      "loss": 0.7579,
       "step": 13090
     },
     {
       "epoch": 2.4656502917372483,
-      "grad_norm": 0.1580163836479187,
-      "learning_rate": 1.5206098249576513e-05,
-      "loss": 0.9413,
+      "grad_norm": 0.027952812612056732,
+      "learning_rate": 0.0002534349708262752,
+      "loss": 0.8328,
       "step": 13100
     },
     {
       "epoch": 2.4675324675324677,
-      "grad_norm": 7.958024501800537,
-      "learning_rate": 1.5194805194805194e-05,
-      "loss": 0.4504,
+      "grad_norm": 2.3775055408477783,
+      "learning_rate": 0.0002532467532467532,
+      "loss": 0.4102,
       "step": 13110
     },
     {
       "epoch": 2.4694146433276867,
-      "grad_norm": 21.001428604125977,
-      "learning_rate": 1.5183512140033878e-05,
-      "loss": 0.9076,
+      "grad_norm": 3.9588053226470947,
+      "learning_rate": 0.0002530585356672313,
+      "loss": 0.9381,
       "step": 13120
     },
     {
       "epoch": 2.471296819122906,
-      "grad_norm": 14.668991088867188,
-      "learning_rate": 1.5172219085262562e-05,
-      "loss": 0.7698,
+      "grad_norm": 5.909026145935059,
+      "learning_rate": 0.00025287031808770937,
+      "loss": 0.7946,
       "step": 13130
     },
     {
       "epoch": 2.4731789949181255,
-      "grad_norm": 9.113917350769043,
-      "learning_rate": 1.5160926030491248e-05,
-      "loss": 0.8785,
+      "grad_norm": 2.5847971439361572,
+      "learning_rate": 0.00025268210050818744,
+      "loss": 0.7612,
       "step": 13140
     },
     {
       "epoch": 2.4750611707133445,
-      "grad_norm": 16.353412628173828,
-      "learning_rate": 1.5149632975719933e-05,
-      "loss": 0.7005,
+      "grad_norm": 3.247206211090088,
+      "learning_rate": 0.0002524938829286655,
+      "loss": 0.6497,
       "step": 13150
     },
     {
       "epoch": 2.476943346508564,
-      "grad_norm": 7.100589275360107,
-      "learning_rate": 1.5138339920948617e-05,
-      "loss": 0.5536,
+      "grad_norm": 2.7777774333953857,
+      "learning_rate": 0.0002523056653491436,
+      "loss": 0.7623,
       "step": 13160
     },
     {
       "epoch": 2.478825522303783,
-      "grad_norm": 10.338568687438965,
-      "learning_rate": 1.5127046866177302e-05,
-      "loss": 0.794,
+      "grad_norm": 3.336306095123291,
+      "learning_rate": 0.00025211744776962167,
+      "loss": 0.7382,
       "step": 13170
     },
     {
       "epoch": 2.4807076980990024,
-      "grad_norm": 19.2764949798584,
-      "learning_rate": 1.5115753811405988e-05,
-      "loss": 0.7741,
+      "grad_norm": 2.404705286026001,
+      "learning_rate": 0.0002519292301900998,
+      "loss": 0.7361,
       "step": 13180
     },
     {
       "epoch": 2.482589873894222,
-      "grad_norm": 41.238563537597656,
-      "learning_rate": 1.5104460756634672e-05,
-      "loss": 0.5174,
+      "grad_norm": 4.079376220703125,
+      "learning_rate": 0.0002517410126105779,
+      "loss": 0.5151,
       "step": 13190
     },
     {
       "epoch": 2.4844720496894412,
-      "grad_norm": 14.756248474121094,
-      "learning_rate": 1.5093167701863353e-05,
-      "loss": 0.6969,
+      "grad_norm": 2.1398980617523193,
+      "learning_rate": 0.0002515527950310559,
+      "loss": 0.6292,
       "step": 13200
     },
     {
       "epoch": 2.4863542254846602,
-      "grad_norm": 25.224035263061523,
-      "learning_rate": 1.5081874647092037e-05,
-      "loss": 0.7373,
+      "grad_norm": 4.041257858276367,
+      "learning_rate": 0.000251364577451534,
+      "loss": 0.6575,
       "step": 13210
     },
     {
       "epoch": 2.4882364012798797,
-      "grad_norm": 27.549787521362305,
-      "learning_rate": 1.5070581592320722e-05,
-      "loss": 0.7657,
+      "grad_norm": 3.372771739959717,
+      "learning_rate": 0.00025117635987201205,
+      "loss": 0.591,
       "step": 13220
     },
     {
       "epoch": 2.4901185770750986,
-      "grad_norm": 13.31575870513916,
-      "learning_rate": 1.5059288537549408e-05,
-      "loss": 0.8013,
+      "grad_norm": 3.360886573791504,
+      "learning_rate": 0.00025098814229249013,
+      "loss": 0.728,
       "step": 13230
     },
     {
       "epoch": 2.492000752870318,
-      "grad_norm": 15.506152153015137,
-      "learning_rate": 1.5047995482778092e-05,
-      "loss": 0.8362,
+      "grad_norm": 3.187817335128784,
+      "learning_rate": 0.0002507999247129682,
+      "loss": 0.7468,
       "step": 13240
     },
     {
       "epoch": 2.4938829286655375,
-      "grad_norm": 12.64037036895752,
-      "learning_rate": 1.5036702428006777e-05,
-      "loss": 0.7907,
+      "grad_norm": 1.4808297157287598,
+      "learning_rate": 0.0002506117071334463,
+      "loss": 0.6668,
       "step": 13250
     },
     {
       "epoch": 2.4957651044607565,
-      "grad_norm": 7.721663475036621,
-      "learning_rate": 1.5025409373235461e-05,
-      "loss": 0.6027,
+      "grad_norm": 1.4633982181549072,
+      "learning_rate": 0.00025042348955392436,
+      "loss": 0.6514,
       "step": 13260
     },
     {
       "epoch": 2.497647280255976,
-      "grad_norm": 20.326704025268555,
-      "learning_rate": 1.5014116318464145e-05,
-      "loss": 0.7932,
+      "grad_norm": 4.78045129776001,
+      "learning_rate": 0.00025023527197440244,
+      "loss": 0.7101,
       "step": 13270
     },
     {
       "epoch": 2.4995294560511954,
-      "grad_norm": 3.6941041946411133,
-      "learning_rate": 1.5002823263692831e-05,
-      "loss": 0.6615,
+      "grad_norm": 2.0362279415130615,
+      "learning_rate": 0.0002500470543948805,
+      "loss": 0.7201,
       "step": 13280
     },
     {
       "epoch": 2.5014116318464144,
-      "grad_norm": 23.14988899230957,
-      "learning_rate": 1.4991530208921514e-05,
-      "loss": 0.6778,
+      "grad_norm": 2.6918833255767822,
+      "learning_rate": 0.0002498588368153586,
+      "loss": 0.6822,
       "step": 13290
     },
     {
       "epoch": 2.503293807641634,
-      "grad_norm": 8.961627960205078,
-      "learning_rate": 1.4980237154150198e-05,
-      "loss": 0.5089,
+      "grad_norm": 5.8409504890441895,
+      "learning_rate": 0.00024967061923583667,
+      "loss": 0.4258,
       "step": 13300
     },
     {
       "epoch": 2.505175983436853,
-      "grad_norm": 17.82870864868164,
-      "learning_rate": 1.4968944099378881e-05,
-      "loss": 0.8967,
+      "grad_norm": 3.082871437072754,
+      "learning_rate": 0.0002494824016563147,
+      "loss": 0.9643,
       "step": 13310
     },
     {
       "epoch": 2.507058159232072,
-      "grad_norm": 3.7372190952301025,
-      "learning_rate": 1.4957651044607567e-05,
-      "loss": 0.6563,
+      "grad_norm": 0.6219594478607178,
+      "learning_rate": 0.00024929418407679276,
+      "loss": 0.5603,
       "step": 13320
     },
     {
       "epoch": 2.5089403350272916,
-      "grad_norm": 17.692962646484375,
-      "learning_rate": 1.4946357989836252e-05,
-      "loss": 0.9301,
+      "grad_norm": 3.4000182151794434,
+      "learning_rate": 0.00024910596649727084,
+      "loss": 0.7337,
       "step": 13330
     },
     {
       "epoch": 2.5108225108225106,
-      "grad_norm": 11.346606254577637,
-      "learning_rate": 1.4935064935064936e-05,
-      "loss": 0.689,
+      "grad_norm": 2.250074625015259,
+      "learning_rate": 0.0002489177489177489,
+      "loss": 0.6619,
       "step": 13340
     },
     {
       "epoch": 2.51270468661773,
-      "grad_norm": 25.42557144165039,
-      "learning_rate": 1.4923771880293619e-05,
-      "loss": 0.6755,
+      "grad_norm": 8.501276969909668,
+      "learning_rate": 0.000248729531338227,
+      "loss": 0.5692,
       "step": 13350
     },
     {
       "epoch": 2.5145868624129495,
-      "grad_norm": 16.371252059936523,
-      "learning_rate": 1.4912478825522303e-05,
-      "loss": 0.6872,
+      "grad_norm": 2.2752902507781982,
+      "learning_rate": 0.00024854131375870507,
+      "loss": 0.6721,
       "step": 13360
     },
     {
       "epoch": 2.516469038208169,
-      "grad_norm": 10.551392555236816,
-      "learning_rate": 1.4901185770750989e-05,
-      "loss": 0.3674,
+      "grad_norm": 3.7881362438201904,
+      "learning_rate": 0.00024835309617918315,
+      "loss": 0.5059,
       "step": 13370
     },
     {
       "epoch": 2.518351214003388,
-      "grad_norm": 7.954850673675537,
-      "learning_rate": 1.4889892715979674e-05,
-      "loss": 0.445,
+      "grad_norm": 5.592350006103516,
+      "learning_rate": 0.0002481648785996612,
+      "loss": 0.4467,
       "step": 13380
     },
     {
       "epoch": 2.5202333897986073,
-      "grad_norm": 6.890119552612305,
-      "learning_rate": 1.4878599661208358e-05,
-      "loss": 0.5636,
+      "grad_norm": 2.011734962463379,
+      "learning_rate": 0.0002479766610201393,
+      "loss": 0.6093,
       "step": 13390
     },
     {
       "epoch": 2.5221155655938263,
-      "grad_norm": 11.65666389465332,
-      "learning_rate": 1.486730660643704e-05,
-      "loss": 0.6063,
+      "grad_norm": 2.0985467433929443,
+      "learning_rate": 0.0002477884434406174,
+      "loss": 0.8551,
       "step": 13400
     },
     {
       "epoch": 2.5239977413890458,
-      "grad_norm": 17.123275756835938,
-      "learning_rate": 1.4856013551665727e-05,
-      "loss": 0.615,
+      "grad_norm": 3.176607608795166,
+      "learning_rate": 0.00024760022586109545,
+      "loss": 0.7844,
       "step": 13410
     },
     {
       "epoch": 2.525879917184265,
-      "grad_norm": 24.541301727294922,
-      "learning_rate": 1.4844720496894411e-05,
-      "loss": 0.9433,
+      "grad_norm": 5.991983890533447,
+      "learning_rate": 0.00024741200828157353,
+      "loss": 1.1296,
       "step": 13420
     },
     {
       "epoch": 2.527762092979484,
-      "grad_norm": 11.410900115966797,
-      "learning_rate": 1.4833427442123095e-05,
-      "loss": 0.5644,
+      "grad_norm": 2.146676540374756,
+      "learning_rate": 0.0002472237907020516,
+      "loss": 0.5867,
       "step": 13430
     },
     {
       "epoch": 2.5296442687747036,
-      "grad_norm": 7.17426872253418,
-      "learning_rate": 1.4822134387351778e-05,
-      "loss": 0.5348,
+      "grad_norm": 1.7108885049819946,
+      "learning_rate": 0.00024703557312252963,
+      "loss": 0.5653,
       "step": 13440
     },
     {
       "epoch": 2.5315264445699226,
-      "grad_norm": 19.93268585205078,
-      "learning_rate": 1.4810841332580463e-05,
-      "loss": 0.9548,
+      "grad_norm": 4.368906497955322,
+      "learning_rate": 0.0002468473555430077,
+      "loss": 0.9203,
       "step": 13450
     },
     {
       "epoch": 2.533408620365142,
-      "grad_norm": 12.935050964355469,
-      "learning_rate": 1.4799548277809149e-05,
-      "loss": 0.6989,
+      "grad_norm": 1.6300593614578247,
+      "learning_rate": 0.0002466591379634858,
+      "loss": 0.5578,
       "step": 13460
     },
     {
       "epoch": 2.5352907961603615,
-      "grad_norm": 19.485132217407227,
-      "learning_rate": 1.4788255223037833e-05,
-      "loss": 0.7934,
+      "grad_norm": 6.861873149871826,
+      "learning_rate": 0.00024647092038396386,
+      "loss": 0.7893,
       "step": 13470
     },
     {
       "epoch": 2.537172971955581,
-      "grad_norm": 28.300479888916016,
-      "learning_rate": 1.4776962168266516e-05,
-      "loss": 0.6659,
+      "grad_norm": 3.013028860092163,
+      "learning_rate": 0.00024628270280444193,
+      "loss": 0.4753,
       "step": 13480
     },
     {
       "epoch": 2.5390551477508,
-      "grad_norm": 27.766523361206055,
-      "learning_rate": 1.47656691134952e-05,
-      "loss": 1.0873,
+      "grad_norm": 3.5351712703704834,
+      "learning_rate": 0.00024609448522492,
+      "loss": 1.182,
       "step": 13490
     },
     {
       "epoch": 2.5409373235460193,
-      "grad_norm": 23.600576400756836,
-      "learning_rate": 1.4754376058723884e-05,
-      "loss": 0.5181,
+      "grad_norm": 2.7321786880493164,
+      "learning_rate": 0.0002459062676453981,
+      "loss": 0.6641,
       "step": 13500
     },
     {
       "epoch": 2.5428194993412383,
-      "grad_norm": 12.327131271362305,
-      "learning_rate": 1.474308300395257e-05,
-      "loss": 0.7655,
+      "grad_norm": 2.764721393585205,
+      "learning_rate": 0.00024571805006587616,
+      "loss": 0.6495,
       "step": 13510
     },
     {
       "epoch": 2.5447016751364577,
-      "grad_norm": 19.353673934936523,
-      "learning_rate": 1.4731789949181255e-05,
-      "loss": 0.8238,
+      "grad_norm": 4.903502941131592,
+      "learning_rate": 0.00024552983248635424,
+      "loss": 0.6116,
       "step": 13520
     },
     {
       "epoch": 2.546583850931677,
-      "grad_norm": 23.07147979736328,
-      "learning_rate": 1.4720496894409938e-05,
-      "loss": 0.8695,
+      "grad_norm": 5.111001014709473,
+      "learning_rate": 0.0002453416149068323,
+      "loss": 0.9473,
       "step": 13530
     },
     {
       "epoch": 2.548466026726896,
-      "grad_norm": 19.188295364379883,
-      "learning_rate": 1.4709203839638622e-05,
-      "loss": 0.7139,
+      "grad_norm": 1.7780566215515137,
+      "learning_rate": 0.0002451533973273104,
+      "loss": 0.7633,
       "step": 13540
     },
     {
       "epoch": 2.5503482025221156,
-      "grad_norm": 8.434711456298828,
-      "learning_rate": 1.4697910784867308e-05,
-      "loss": 0.8626,
+      "grad_norm": 5.7975850105285645,
+      "learning_rate": 0.00024496517974778847,
+      "loss": 0.8065,
       "step": 13550
     },
     {
       "epoch": 2.5522303783173346,
-      "grad_norm": 4.494293212890625,
-      "learning_rate": 1.4686617730095992e-05,
-      "loss": 0.7623,
+      "grad_norm": 2.282766819000244,
+      "learning_rate": 0.00024477696216826654,
+      "loss": 0.7373,
       "step": 13560
     },
     {
       "epoch": 2.554112554112554,
-      "grad_norm": 17.76078224182129,
-      "learning_rate": 1.4675324675324675e-05,
-      "loss": 0.7771,
+      "grad_norm": 3.5103821754455566,
+      "learning_rate": 0.00024458874458874457,
+      "loss": 0.7865,
       "step": 13570
     },
     {
       "epoch": 2.5559947299077734,
-      "grad_norm": 18.283334732055664,
-      "learning_rate": 1.466403162055336e-05,
-      "loss": 0.615,
+      "grad_norm": 4.03853178024292,
+      "learning_rate": 0.00024440052700922264,
+      "loss": 0.8197,
       "step": 13580
     },
     {
       "epoch": 2.557876905702993,
-      "grad_norm": 15.662129402160645,
-      "learning_rate": 1.4652738565782044e-05,
-      "loss": 1.0202,
+      "grad_norm": 2.961564540863037,
+      "learning_rate": 0.0002442123094297007,
+      "loss": 1.087,
       "step": 13590
     },
     {
       "epoch": 2.559759081498212,
-      "grad_norm": 11.736997604370117,
-      "learning_rate": 1.464144551101073e-05,
-      "loss": 0.6311,
+      "grad_norm": 4.064123153686523,
+      "learning_rate": 0.00024402409185017882,
+      "loss": 0.5747,
       "step": 13600
     },
     {
       "epoch": 2.5616412572934313,
-      "grad_norm": 26.242910385131836,
-      "learning_rate": 1.4630152456239413e-05,
-      "loss": 0.5408,
+      "grad_norm": 2.7881264686584473,
+      "learning_rate": 0.00024383587427065687,
+      "loss": 0.6466,
       "step": 13610
     },
     {
       "epoch": 2.5635234330886503,
-      "grad_norm": 34.35280227661133,
-      "learning_rate": 1.4618859401468097e-05,
-      "loss": 0.5723,
+      "grad_norm": 3.5973377227783203,
+      "learning_rate": 0.00024364765669113495,
+      "loss": 0.5321,
       "step": 13620
     },
     {
       "epoch": 2.5654056088838697,
-      "grad_norm": 12.662686347961426,
-      "learning_rate": 1.4607566346696781e-05,
-      "loss": 0.9724,
+      "grad_norm": 1.8289278745651245,
+      "learning_rate": 0.00024345943911161303,
+      "loss": 1.026,
       "step": 13630
     },
     {
       "epoch": 2.567287784679089,
-      "grad_norm": 7.174187660217285,
-      "learning_rate": 1.4596273291925466e-05,
-      "loss": 0.7687,
+      "grad_norm": 3.8798322677612305,
+      "learning_rate": 0.0002432712215320911,
+      "loss": 0.8,
       "step": 13640
     },
     {
       "epoch": 2.5691699604743086,
-      "grad_norm": 13.642120361328125,
-      "learning_rate": 1.4584980237154152e-05,
-      "loss": 0.5166,
+      "grad_norm": 6.2854461669921875,
+      "learning_rate": 0.00024308300395256918,
+      "loss": 0.5575,
       "step": 13650
     },
     {
       "epoch": 2.5710521362695276,
-      "grad_norm": 18.49991798400879,
-      "learning_rate": 1.4573687182382835e-05,
-      "loss": 0.889,
+      "grad_norm": 2.248408794403076,
+      "learning_rate": 0.00024289478637304723,
+      "loss": 0.8251,
       "step": 13660
     },
     {
       "epoch": 2.572934312064747,
-      "grad_norm": 15.515763282775879,
-      "learning_rate": 1.4562394127611519e-05,
-      "loss": 0.7712,
+      "grad_norm": 3.2871415615081787,
+      "learning_rate": 0.0002427065687935253,
+      "loss": 0.8645,
       "step": 13670
     },
     {
       "epoch": 2.574816487859966,
-      "grad_norm": 9.801226615905762,
-      "learning_rate": 1.4551101072840203e-05,
-      "loss": 1.0405,
+      "grad_norm": 2.0668184757232666,
+      "learning_rate": 0.0002425183512140034,
+      "loss": 1.0611,
       "step": 13680
     },
     {
       "epoch": 2.5766986636551854,
-      "grad_norm": 8.876042366027832,
-      "learning_rate": 1.453980801806889e-05,
-      "loss": 0.3575,
+      "grad_norm": 2.037912368774414,
+      "learning_rate": 0.00024233013363448148,
+      "loss": 0.3269,
       "step": 13690
     },
     {
       "epoch": 2.578580839450405,
-      "grad_norm": 23.405811309814453,
-      "learning_rate": 1.4528514963297572e-05,
-      "loss": 1.0442,
+      "grad_norm": 4.442802906036377,
+      "learning_rate": 0.00024214191605495953,
+      "loss": 0.9673,
       "step": 13700
     },
     {
       "epoch": 2.580463015245624,
-      "grad_norm": 19.530874252319336,
-      "learning_rate": 1.4517221908526256e-05,
-      "loss": 0.7402,
+      "grad_norm": 5.721139430999756,
+      "learning_rate": 0.0002419536984754376,
+      "loss": 0.627,
       "step": 13710
     },
     {
       "epoch": 2.5823451910408433,
-      "grad_norm": 14.73330020904541,
-      "learning_rate": 1.450592885375494e-05,
-      "loss": 0.6519,
+      "grad_norm": 12.355080604553223,
+      "learning_rate": 0.00024176548089591569,
+      "loss": 0.6522,
       "step": 13720
     },
     {
       "epoch": 2.5842273668360622,
-      "grad_norm": 31.803998947143555,
-      "learning_rate": 1.4494635798983625e-05,
-      "loss": 1.0237,
+      "grad_norm": 4.2488813400268555,
+      "learning_rate": 0.00024157726331639376,
+      "loss": 0.8524,
       "step": 13730
     },
     {
       "epoch": 2.5861095426312817,
-      "grad_norm": 25.37823486328125,
-      "learning_rate": 1.448334274421231e-05,
-      "loss": 0.9164,
+      "grad_norm": 5.772805690765381,
+      "learning_rate": 0.0002413890457368718,
+      "loss": 0.8406,
       "step": 13740
     },
     {
       "epoch": 2.587991718426501,
-      "grad_norm": 25.509199142456055,
-      "learning_rate": 1.4472049689440994e-05,
-      "loss": 0.76,
+      "grad_norm": 4.153681755065918,
+      "learning_rate": 0.0002412008281573499,
+      "loss": 0.6873,
       "step": 13750
     },
     {
       "epoch": 2.5898738942217205,
-      "grad_norm": 4.799928665161133,
-      "learning_rate": 1.4460756634669678e-05,
-      "loss": 0.2943,
+      "grad_norm": 0.7333030104637146,
+      "learning_rate": 0.00024101261057782796,
+      "loss": 0.2909,
       "step": 13760
     },
     {
       "epoch": 2.5917560700169395,
-      "grad_norm": 12.444110870361328,
-      "learning_rate": 1.4449463579898363e-05,
-      "loss": 0.8041,
+      "grad_norm": 0.3361065983772278,
+      "learning_rate": 0.00024082439299830604,
+      "loss": 0.8215,
       "step": 13770
     },
     {
       "epoch": 2.593638245812159,
-      "grad_norm": 16.842971801757812,
-      "learning_rate": 1.4438170525127047e-05,
-      "loss": 0.6848,
+      "grad_norm": 3.8252694606781006,
+      "learning_rate": 0.00024063617541878414,
+      "loss": 0.521,
       "step": 13780
     },
     {
       "epoch": 2.595520421607378,
-      "grad_norm": 21.784448623657227,
-      "learning_rate": 1.4426877470355732e-05,
-      "loss": 0.7807,
+      "grad_norm": 4.091780185699463,
+      "learning_rate": 0.0002404479578392622,
+      "loss": 0.8096,
       "step": 13790
     },
     {
       "epoch": 2.5974025974025974,
-      "grad_norm": 1.9101155996322632,
-      "learning_rate": 1.4415584415584416e-05,
-      "loss": 0.4079,
+      "grad_norm": 1.5982844829559326,
+      "learning_rate": 0.00024025974025974027,
+      "loss": 0.4528,
       "step": 13800
     },
     {
       "epoch": 2.599284773197817,
-      "grad_norm": 10.34659194946289,
-      "learning_rate": 1.44042913608131e-05,
-      "loss": 0.9427,
+      "grad_norm": 3.969874620437622,
+      "learning_rate": 0.00024007152268021835,
+      "loss": 0.9265,
       "step": 13810
     },
     {
       "epoch": 2.601166948993036,
-      "grad_norm": 32.64616394042969,
-      "learning_rate": 1.4392998306041785e-05,
-      "loss": 0.968,
+      "grad_norm": 3.207587242126465,
+      "learning_rate": 0.00023988330510069642,
+      "loss": 1.0132,
       "step": 13820
     },
     {
       "epoch": 2.6030491247882552,
-      "grad_norm": 23.128334045410156,
-      "learning_rate": 1.4381705251270469e-05,
-      "loss": 0.8594,
+      "grad_norm": 4.083827495574951,
+      "learning_rate": 0.00023969508752117447,
+      "loss": 0.7304,
       "step": 13830
     },
     {
       "epoch": 2.6049313005834747,
-      "grad_norm": 13.279285430908203,
-      "learning_rate": 1.4370412196499153e-05,
-      "loss": 0.7133,
+      "grad_norm": 3.4876480102539062,
+      "learning_rate": 0.00023950686994165255,
+      "loss": 0.618,
       "step": 13840
     },
     {
       "epoch": 2.6068134763786937,
-      "grad_norm": 3.632760763168335,
-      "learning_rate": 1.4359119141727838e-05,
-      "loss": 0.4146,
+      "grad_norm": 0.7380732297897339,
+      "learning_rate": 0.00023931865236213063,
+      "loss": 0.4644,
       "step": 13850
     },
     {
       "epoch": 2.608695652173913,
-      "grad_norm": 5.402322769165039,
-      "learning_rate": 1.4347826086956522e-05,
-      "loss": 0.4458,
+      "grad_norm": 1.9918617010116577,
+      "learning_rate": 0.0002391304347826087,
+      "loss": 0.3434,
       "step": 13860
     },
     {
       "epoch": 2.6105778279691325,
-      "grad_norm": 18.91129493713379,
-      "learning_rate": 1.4336533032185207e-05,
-      "loss": 0.7429,
+      "grad_norm": 0.4870288074016571,
+      "learning_rate": 0.00023894221720308678,
+      "loss": 0.7492,
       "step": 13870
     },
     {
       "epoch": 2.6124600037643515,
-      "grad_norm": 29.642852783203125,
-      "learning_rate": 1.4325239977413891e-05,
-      "loss": 0.6789,
+      "grad_norm": 3.0908610820770264,
+      "learning_rate": 0.00023875399962356483,
+      "loss": 0.7168,
       "step": 13880
     },
     {
       "epoch": 2.614342179559571,
-      "grad_norm": 26.60822868347168,
-      "learning_rate": 1.4313946922642575e-05,
-      "loss": 0.708,
+      "grad_norm": 2.242732524871826,
+      "learning_rate": 0.0002385657820440429,
+      "loss": 0.7139,
       "step": 13890
     },
     {
       "epoch": 2.61622435535479,
-      "grad_norm": 8.151140213012695,
-      "learning_rate": 1.430265386787126e-05,
-      "loss": 0.9136,
+      "grad_norm": 4.407373905181885,
+      "learning_rate": 0.000238377564464521,
+      "loss": 1.0584,
       "step": 13900
     },
     {
       "epoch": 2.6181065311500094,
-      "grad_norm": 9.463515281677246,
-      "learning_rate": 1.4291360813099944e-05,
-      "loss": 0.7923,
+      "grad_norm": 4.035942077636719,
+      "learning_rate": 0.00023818934688499908,
+      "loss": 0.7392,
       "step": 13910
     },
     {
       "epoch": 2.619988706945229,
-      "grad_norm": 21.169139862060547,
-      "learning_rate": 1.4280067758328628e-05,
-      "loss": 0.7719,
+      "grad_norm": 5.146146774291992,
+      "learning_rate": 0.00023800112930547713,
+      "loss": 0.7454,
       "step": 13920
     },
     {
       "epoch": 2.621870882740448,
-      "grad_norm": 2.78265643119812,
-      "learning_rate": 1.4268774703557313e-05,
-      "loss": 0.564,
+      "grad_norm": 2.152848482131958,
+      "learning_rate": 0.0002378129117259552,
+      "loss": 0.5348,
       "step": 13930
     },
     {
       "epoch": 2.623753058535667,
-      "grad_norm": 10.70055103302002,
-      "learning_rate": 1.4257481648785997e-05,
-      "loss": 0.7906,
+      "grad_norm": 4.042505741119385,
+      "learning_rate": 0.00023762469414643329,
+      "loss": 0.673,
       "step": 13940
     },
     {
       "epoch": 2.6256352343308866,
-      "grad_norm": 6.613348007202148,
-      "learning_rate": 1.4246188594014682e-05,
-      "loss": 0.4984,
+      "grad_norm": 4.227035999298096,
+      "learning_rate": 0.00023743647656691136,
+      "loss": 0.4952,
       "step": 13950
     },
     {
       "epoch": 2.6275174101261056,
-      "grad_norm": 13.339488983154297,
-      "learning_rate": 1.4234895539243364e-05,
-      "loss": 0.9332,
+      "grad_norm": 3.608710289001465,
+      "learning_rate": 0.0002372482589873894,
+      "loss": 0.9144,
       "step": 13960
     },
     {
       "epoch": 2.629399585921325,
-      "grad_norm": 11.467644691467285,
-      "learning_rate": 1.422360248447205e-05,
-      "loss": 0.5156,
+      "grad_norm": 3.100811004638672,
+      "learning_rate": 0.0002370600414078675,
+      "loss": 0.6548,
       "step": 13970
     },
     {
       "epoch": 2.6312817617165445,
-      "grad_norm": 29.80975914001465,
-      "learning_rate": 1.4212309429700735e-05,
-      "loss": 0.5243,
+      "grad_norm": 3.055030584335327,
+      "learning_rate": 0.00023687182382834556,
+      "loss": 0.4532,
       "step": 13980
     },
     {
       "epoch": 2.6331639375117635,
-      "grad_norm": 18.506032943725586,
-      "learning_rate": 1.4201016374929419e-05,
-      "loss": 0.9703,
+      "grad_norm": 3.4377622604370117,
+      "learning_rate": 0.00023668360624882364,
+      "loss": 0.9787,
       "step": 13990
     },
     {
       "epoch": 2.635046113306983,
-      "grad_norm": 12.98060131072998,
-      "learning_rate": 1.4189723320158104e-05,
-      "loss": 0.9162,
+      "grad_norm": 5.330870628356934,
+      "learning_rate": 0.00023649538866930172,
+      "loss": 0.8944,
       "step": 14000
     },
     {
       "epoch": 2.636928289102202,
-      "grad_norm": 3.217790365219116,
-      "learning_rate": 1.4178430265386786e-05,
-      "loss": 0.6168,
+      "grad_norm": 1.1406224966049194,
+      "learning_rate": 0.0002363071710897798,
+      "loss": 0.5997,
       "step": 14010
     },
     {
       "epoch": 2.6388104648974213,
-      "grad_norm": 12.240312576293945,
-      "learning_rate": 1.4167137210615472e-05,
-      "loss": 0.587,
+      "grad_norm": 2.4622602462768555,
+      "learning_rate": 0.00023611895351025787,
+      "loss": 0.6162,
       "step": 14020
     },
     {
       "epoch": 2.6406926406926408,
-      "grad_norm": 17.046951293945312,
-      "learning_rate": 1.4155844155844157e-05,
-      "loss": 0.4854,
+      "grad_norm": 0.7391192317008972,
+      "learning_rate": 0.00023593073593073595,
+      "loss": 0.4323,
       "step": 14030
     },
     {
       "epoch": 2.64257481648786,
-      "grad_norm": 20.683822631835938,
-      "learning_rate": 1.4144551101072841e-05,
-      "loss": 0.5354,
+      "grad_norm": 4.043608665466309,
+      "learning_rate": 0.00023574251835121402,
+      "loss": 0.6287,
       "step": 14040
     },
     {
       "epoch": 2.644456992283079,
-      "grad_norm": 19.47835350036621,
-      "learning_rate": 1.4133258046301524e-05,
-      "loss": 0.5272,
+      "grad_norm": 4.61051082611084,
+      "learning_rate": 0.00023555430077169207,
+      "loss": 0.691,
       "step": 14050
     },
     {
       "epoch": 2.6463391680782986,
-      "grad_norm": 5.546383380889893,
-      "learning_rate": 1.412196499153021e-05,
-      "loss": 0.7365,
+      "grad_norm": 0.22076202929019928,
+      "learning_rate": 0.00023536608319217015,
+      "loss": 0.8374,
       "step": 14060
     },
     {
       "epoch": 2.6482213438735176,
-      "grad_norm": 19.25607681274414,
-      "learning_rate": 1.4110671936758894e-05,
-      "loss": 0.4252,
+      "grad_norm": 5.464056968688965,
+      "learning_rate": 0.00023517786561264823,
+      "loss": 0.3803,
       "step": 14070
     },
     {
       "epoch": 2.650103519668737,
-      "grad_norm": 5.1752448081970215,
-      "learning_rate": 1.4099378881987579e-05,
-      "loss": 0.7042,
+      "grad_norm": 0.7410657405853271,
+      "learning_rate": 0.0002349896480331263,
+      "loss": 0.6079,
       "step": 14080
     },
     {
       "epoch": 2.6519856954639565,
-      "grad_norm": 1.2071647644042969,
-      "learning_rate": 1.4088085827216261e-05,
-      "loss": 0.5272,
+      "grad_norm": 3.5357539653778076,
+      "learning_rate": 0.00023480143045360435,
+      "loss": 0.5994,
       "step": 14090
     },
     {
       "epoch": 2.6538678712591754,
-      "grad_norm": 33.30137252807617,
-      "learning_rate": 1.4076792772444946e-05,
-      "loss": 0.7419,
+      "grad_norm": 1.7675532102584839,
+      "learning_rate": 0.00023461321287408243,
+      "loss": 0.7059,
       "step": 14100
     },
     {
       "epoch": 2.655750047054395,
-      "grad_norm": 2.014209508895874,
-      "learning_rate": 1.4065499717673632e-05,
-      "loss": 0.7304,
+      "grad_norm": 2.598212480545044,
+      "learning_rate": 0.0002344249952945605,
+      "loss": 0.6269,
       "step": 14110
     },
     {
       "epoch": 2.6576322228496143,
-      "grad_norm": 24.12761116027832,
-      "learning_rate": 1.4054206662902316e-05,
-      "loss": 0.859,
+      "grad_norm": 5.439962387084961,
+      "learning_rate": 0.0002342367777150386,
+      "loss": 0.9013,
       "step": 14120
     },
     {
       "epoch": 2.6595143986448333,
-      "grad_norm": 13.404014587402344,
-      "learning_rate": 1.4042913608131e-05,
-      "loss": 0.7206,
+      "grad_norm": 2.2769856452941895,
+      "learning_rate": 0.00023404856013551668,
+      "loss": 0.7557,
       "step": 14130
     },
     {
       "epoch": 2.6613965744400527,
-      "grad_norm": 28.344799041748047,
-      "learning_rate": 1.4031620553359683e-05,
-      "loss": 0.7208,
+      "grad_norm": 4.408846855163574,
+      "learning_rate": 0.00023386034255599473,
+      "loss": 0.6591,
       "step": 14140
     },
     {
       "epoch": 2.663278750235272,
-      "grad_norm": 23.86800765991211,
-      "learning_rate": 1.4020327498588368e-05,
-      "loss": 0.8126,
+      "grad_norm": 5.050843715667725,
+      "learning_rate": 0.0002336721249764728,
+      "loss": 0.8656,
       "step": 14150
     },
     {
       "epoch": 2.665160926030491,
-      "grad_norm": 14.190437316894531,
-      "learning_rate": 1.4009034443817054e-05,
-      "loss": 0.6967,
+      "grad_norm": 3.493145704269409,
+      "learning_rate": 0.00023348390739695089,
+      "loss": 0.5883,
       "step": 14160
     },
     {
       "epoch": 2.6670431018257106,
-      "grad_norm": 0.3645530045032501,
-      "learning_rate": 1.3997741389045738e-05,
-      "loss": 0.4742,
+      "grad_norm": 0.21144308149814606,
+      "learning_rate": 0.00023329568981742896,
+      "loss": 0.4608,
       "step": 14170
     },
     {
       "epoch": 2.6689252776209296,
-      "grad_norm": 15.038154602050781,
-      "learning_rate": 1.398644833427442e-05,
-      "loss": 1.0225,
+      "grad_norm": 5.041484832763672,
+      "learning_rate": 0.000233107472237907,
+      "loss": 0.9009,
       "step": 14180
     },
     {
       "epoch": 2.670807453416149,
-      "grad_norm": 12.963705062866211,
-      "learning_rate": 1.3975155279503105e-05,
-      "loss": 0.6174,
+      "grad_norm": 2.556727409362793,
+      "learning_rate": 0.0002329192546583851,
+      "loss": 0.6786,
       "step": 14190
     },
     {
       "epoch": 2.6726896292113684,
-      "grad_norm": 15.526269912719727,
-      "learning_rate": 1.3963862224731791e-05,
-      "loss": 0.6071,
+      "grad_norm": 3.09854793548584,
+      "learning_rate": 0.00023273103707886316,
+      "loss": 0.6226,
       "step": 14200
     },
     {
       "epoch": 2.674571805006588,
-      "grad_norm": 12.310412406921387,
-      "learning_rate": 1.3952569169960476e-05,
-      "loss": 1.0341,
+      "grad_norm": 2.4376440048217773,
+      "learning_rate": 0.00023254281949934124,
+      "loss": 0.9534,
       "step": 14210
     },
     {
       "epoch": 2.676453980801807,
-      "grad_norm": 17.386167526245117,
-      "learning_rate": 1.3941276115189158e-05,
-      "loss": 0.9449,
+      "grad_norm": 1.3566420078277588,
+      "learning_rate": 0.0002323546019198193,
+      "loss": 0.9804,
       "step": 14220
     },
     {
       "epoch": 2.6783361565970263,
-      "grad_norm": 3.0282480716705322,
-      "learning_rate": 1.3929983060417843e-05,
-      "loss": 0.4089,
+      "grad_norm": 0.6485238671302795,
+      "learning_rate": 0.0002321663843402974,
+      "loss": 0.4259,
       "step": 14230
     },
     {
       "epoch": 2.6802183323922453,
-      "grad_norm": 1.6062228679656982,
-      "learning_rate": 1.3918690005646527e-05,
-      "loss": 0.4825,
+      "grad_norm": 0.4050973653793335,
+      "learning_rate": 0.00023197816676077547,
+      "loss": 0.5302,
       "step": 14240
     },
     {
       "epoch": 2.6821005081874647,
-      "grad_norm": 1.4682273864746094,
-      "learning_rate": 1.3907396950875213e-05,
-      "loss": 0.597,
+      "grad_norm": 2.0743534564971924,
+      "learning_rate": 0.00023178994918125355,
+      "loss": 0.6506,
       "step": 14250
     },
     {
       "epoch": 2.683982683982684,
-      "grad_norm": 19.706525802612305,
-      "learning_rate": 1.3896103896103897e-05,
-      "loss": 0.9402,
+      "grad_norm": 3.749516487121582,
+      "learning_rate": 0.00023160173160173162,
+      "loss": 0.9767,
       "step": 14260
     },
     {
       "epoch": 2.685864859777903,
-      "grad_norm": 13.49492073059082,
-      "learning_rate": 1.388481084133258e-05,
-      "loss": 1.1029,
+      "grad_norm": 2.801708936691284,
+      "learning_rate": 0.00023141351402220967,
+      "loss": 1.0014,
       "step": 14270
     },
     {
       "epoch": 2.6877470355731226,
-      "grad_norm": 27.556337356567383,
-      "learning_rate": 1.3873517786561265e-05,
-      "loss": 0.789,
+      "grad_norm": 6.906372547149658,
+      "learning_rate": 0.00023122529644268775,
+      "loss": 0.7968,
       "step": 14280
     },
     {
       "epoch": 2.6896292113683415,
-      "grad_norm": 16.97081756591797,
-      "learning_rate": 1.3862224731789949e-05,
-      "loss": 0.5639,
+      "grad_norm": 3.7884345054626465,
+      "learning_rate": 0.00023103707886316583,
+      "loss": 0.7028,
       "step": 14290
     },
     {
       "epoch": 2.691511387163561,
-      "grad_norm": 24.679975509643555,
-      "learning_rate": 1.3850931677018635e-05,
-      "loss": 0.7003,
+      "grad_norm": 1.527142882347107,
+      "learning_rate": 0.0002308488612836439,
+      "loss": 0.6148,
       "step": 14300
     },
     {
       "epoch": 2.6933935629587804,
-      "grad_norm": 24.730314254760742,
-      "learning_rate": 1.3839638622247318e-05,
-      "loss": 0.6968,
+      "grad_norm": 2.611942768096924,
+      "learning_rate": 0.00023066064370412195,
+      "loss": 0.6652,
       "step": 14310
     },
     {
       "epoch": 2.695275738754,
-      "grad_norm": 13.802581787109375,
-      "learning_rate": 1.3828345567476002e-05,
-      "loss": 0.7957,
+      "grad_norm": 3.685335636138916,
+      "learning_rate": 0.00023047242612460003,
+      "loss": 0.7482,
       "step": 14320
     },
     {
       "epoch": 2.697157914549219,
-      "grad_norm": 21.049495697021484,
-      "learning_rate": 1.3817052512704686e-05,
-      "loss": 0.3892,
+      "grad_norm": 1.5061359405517578,
+      "learning_rate": 0.0002302842085450781,
+      "loss": 0.4571,
       "step": 14330
     },
     {
       "epoch": 2.6990400903444383,
-      "grad_norm": 4.196599006652832,
-      "learning_rate": 1.3805759457933373e-05,
-      "loss": 0.6734,
+      "grad_norm": 1.478553056716919,
+      "learning_rate": 0.0002300959909655562,
+      "loss": 0.6591,
       "step": 14340
     },
     {
       "epoch": 2.7009222661396572,
-      "grad_norm": 37.20542526245117,
-      "learning_rate": 1.3794466403162055e-05,
-      "loss": 0.9126,
+      "grad_norm": 5.025265693664551,
+      "learning_rate": 0.00022990777338603426,
+      "loss": 0.8795,
       "step": 14350
     },
     {
       "epoch": 2.7028044419348767,
-      "grad_norm": 11.342184066772461,
-      "learning_rate": 1.378317334839074e-05,
-      "loss": 0.5825,
+      "grad_norm": 4.221109390258789,
+      "learning_rate": 0.00022971955580651233,
+      "loss": 0.7459,
       "step": 14360
     },
     {
       "epoch": 2.704686617730096,
-      "grad_norm": 9.5272798538208,
-      "learning_rate": 1.3771880293619424e-05,
-      "loss": 0.6676,
+      "grad_norm": 4.284846782684326,
+      "learning_rate": 0.0002295313382269904,
+      "loss": 0.7888,
       "step": 14370
     },
     {
       "epoch": 2.7065687935253155,
-      "grad_norm": 14.55465030670166,
-      "learning_rate": 1.3760587238848108e-05,
-      "loss": 0.899,
+      "grad_norm": 3.834550619125366,
+      "learning_rate": 0.00022934312064746849,
+      "loss": 0.7785,
       "step": 14380
     },
     {
       "epoch": 2.7084509693205345,
-      "grad_norm": 16.193349838256836,
-      "learning_rate": 1.3749294184076794e-05,
-      "loss": 0.5359,
+      "grad_norm": 5.762689590454102,
+      "learning_rate": 0.00022915490306794656,
+      "loss": 0.5134,
       "step": 14390
     },
     {
       "epoch": 2.710333145115754,
-      "grad_norm": 0.4011596441268921,
-      "learning_rate": 1.3738001129305477e-05,
-      "loss": 0.6422,
+      "grad_norm": 0.5311426520347595,
+      "learning_rate": 0.0002289666854884246,
+      "loss": 0.6027,
       "step": 14400
     },
     {
       "epoch": 2.712215320910973,
-      "grad_norm": 18.894222259521484,
-      "learning_rate": 1.3726708074534161e-05,
-      "loss": 0.6957,
+      "grad_norm": 3.2206649780273438,
+      "learning_rate": 0.0002287784679089027,
+      "loss": 0.7232,
       "step": 14410
     },
     {
       "epoch": 2.7140974967061924,
-      "grad_norm": 23.98822593688965,
-      "learning_rate": 1.3715415019762846e-05,
-      "loss": 0.9018,
+      "grad_norm": 3.6478111743927,
+      "learning_rate": 0.00022859025032938076,
+      "loss": 0.893,
       "step": 14420
     },
     {
       "epoch": 2.715979672501412,
-      "grad_norm": 24.73630142211914,
-      "learning_rate": 1.3704121964991532e-05,
-      "loss": 0.6684,
+      "grad_norm": 5.927709102630615,
+      "learning_rate": 0.00022840203274985884,
+      "loss": 0.7657,
       "step": 14430
     },
     {
       "epoch": 2.717861848296631,
-      "grad_norm": 0.7143511772155762,
-      "learning_rate": 1.3692828910220215e-05,
-      "loss": 0.7074,
+      "grad_norm": 0.11305610835552216,
+      "learning_rate": 0.0002282138151703369,
+      "loss": 0.5883,
       "step": 14440
     },
     {
       "epoch": 2.7197440240918502,
-      "grad_norm": 13.2362699508667,
-      "learning_rate": 1.3681535855448899e-05,
-      "loss": 0.6953,
+      "grad_norm": 2.982926845550537,
+      "learning_rate": 0.000228025597590815,
+      "loss": 0.5712,
       "step": 14450
     },
     {
       "epoch": 2.721626199887069,
-      "grad_norm": 0.7118558287620544,
-      "learning_rate": 1.3670242800677583e-05,
-      "loss": 0.6667,
+      "grad_norm": 3.5719408988952637,
+      "learning_rate": 0.00022783738001129307,
+      "loss": 0.6554,
       "step": 14460
     },
     {
       "epoch": 2.7235083756822887,
-      "grad_norm": 10.795101165771484,
-      "learning_rate": 1.3658949745906268e-05,
-      "loss": 0.8088,
+      "grad_norm": 2.8321266174316406,
+      "learning_rate": 0.00022764916243177115,
+      "loss": 0.7021,
       "step": 14470
     },
     {
       "epoch": 2.725390551477508,
-      "grad_norm": 21.050437927246094,
-      "learning_rate": 1.3647656691134954e-05,
-      "loss": 0.6499,
+      "grad_norm": 1.245736837387085,
+      "learning_rate": 0.00022746094485224922,
+      "loss": 0.5736,
       "step": 14480
     },
     {
       "epoch": 2.7272727272727275,
-      "grad_norm": 15.647856712341309,
-      "learning_rate": 1.3636363636363637e-05,
-      "loss": 0.769,
+      "grad_norm": 3.6251332759857178,
+      "learning_rate": 0.00022727272727272727,
+      "loss": 0.7772,
       "step": 14490
     },
     {
       "epoch": 2.7291549030679465,
-      "grad_norm": 18.59046173095703,
-      "learning_rate": 1.3625070581592321e-05,
-      "loss": 0.8155,
+      "grad_norm": 3.757814407348633,
+      "learning_rate": 0.00022708450969320535,
+      "loss": 0.8901,
       "step": 14500
     },
     {
       "epoch": 2.731037078863166,
-      "grad_norm": 0.6544022560119629,
-      "learning_rate": 1.3613777526821005e-05,
-      "loss": 0.5492,
+      "grad_norm": 0.9213415384292603,
+      "learning_rate": 0.00022689629211368343,
+      "loss": 0.7134,
       "step": 14510
     },
     {
       "epoch": 2.732919254658385,
-      "grad_norm": 5.2929182052612305,
-      "learning_rate": 1.360248447204969e-05,
-      "loss": 0.6304,
+      "grad_norm": 3.748073101043701,
+      "learning_rate": 0.0002267080745341615,
+      "loss": 0.6412,
       "step": 14520
     },
     {
       "epoch": 2.7348014304536044,
-      "grad_norm": 18.719219207763672,
-      "learning_rate": 1.3591191417278374e-05,
-      "loss": 0.7405,
+      "grad_norm": 6.530447006225586,
+      "learning_rate": 0.00022651985695463955,
+      "loss": 0.7077,
       "step": 14530
     },
     {
       "epoch": 2.736683606248824,
-      "grad_norm": 14.632789611816406,
-      "learning_rate": 1.3579898362507058e-05,
-      "loss": 0.6674,
+      "grad_norm": 2.432340621948242,
+      "learning_rate": 0.00022633163937511763,
+      "loss": 0.5575,
       "step": 14540
     },
     {
       "epoch": 2.7385657820440428,
-      "grad_norm": 1.8742183446884155,
-      "learning_rate": 1.3568605307735743e-05,
-      "loss": 0.7499,
+      "grad_norm": 0.6200935244560242,
+      "learning_rate": 0.0002261434217955957,
+      "loss": 0.6548,
       "step": 14550
     },
     {
       "epoch": 2.740447957839262,
-      "grad_norm": 14.247894287109375,
-      "learning_rate": 1.3557312252964427e-05,
-      "loss": 0.8707,
+      "grad_norm": 4.548830986022949,
+      "learning_rate": 0.0002259552042160738,
+      "loss": 0.9457,
       "step": 14560
     },
     {
       "epoch": 2.742330133634481,
-      "grad_norm": 12.40461254119873,
-      "learning_rate": 1.3546019198193112e-05,
-      "loss": 0.6975,
+      "grad_norm": 3.477888822555542,
+      "learning_rate": 0.00022576698663655186,
+      "loss": 0.7322,
       "step": 14570
     },
     {
       "epoch": 2.7442123094297006,
-      "grad_norm": 3.423053503036499,
-      "learning_rate": 1.3534726143421796e-05,
-      "loss": 0.9175,
+      "grad_norm": 2.17950701713562,
+      "learning_rate": 0.00022557876905702993,
+      "loss": 0.9557,
       "step": 14580
     },
     {
       "epoch": 2.74609448522492,
-      "grad_norm": 6.041767597198486,
-      "learning_rate": 1.352343308865048e-05,
-      "loss": 0.6782,
+      "grad_norm": 0.9166530966758728,
+      "learning_rate": 0.000225390551477508,
+      "loss": 0.7076,
       "step": 14590
     },
     {
       "epoch": 2.7479766610201395,
-      "grad_norm": 29.75321388244629,
-      "learning_rate": 1.3512140033879165e-05,
-      "loss": 0.6639,
+      "grad_norm": 3.2019553184509277,
+      "learning_rate": 0.00022520233389798609,
+      "loss": 0.6551,
       "step": 14600
     },
     {
       "epoch": 2.7498588368153585,
-      "grad_norm": 18.750484466552734,
-      "learning_rate": 1.3500846979107849e-05,
-      "loss": 0.6613,
+      "grad_norm": 3.706820249557495,
+      "learning_rate": 0.00022501411631846416,
+      "loss": 0.6233,
       "step": 14610
     },
     {
       "epoch": 2.751741012610578,
-      "grad_norm": 15.438997268676758,
-      "learning_rate": 1.3489553924336534e-05,
-      "loss": 0.551,
+      "grad_norm": 3.3140599727630615,
+      "learning_rate": 0.0002248258987389422,
+      "loss": 0.6466,
       "step": 14620
     },
     {
       "epoch": 2.753623188405797,
-      "grad_norm": 14.487974166870117,
-      "learning_rate": 1.3478260869565218e-05,
-      "loss": 0.7749,
+      "grad_norm": 3.88219952583313,
+      "learning_rate": 0.0002246376811594203,
+      "loss": 0.7267,
       "step": 14630
     },
     {
       "epoch": 2.7555053642010163,
-      "grad_norm": 3.726011276245117,
-      "learning_rate": 1.3466967814793902e-05,
-      "loss": 0.4723,
+      "grad_norm": 1.0075336694717407,
+      "learning_rate": 0.00022444946357989836,
+      "loss": 0.4335,
       "step": 14640
     },
     {
       "epoch": 2.7573875399962358,
-      "grad_norm": 12.075268745422363,
-      "learning_rate": 1.3455674760022587e-05,
-      "loss": 0.7522,
+      "grad_norm": 2.2053027153015137,
+      "learning_rate": 0.00022426124600037644,
+      "loss": 0.7073,
       "step": 14650
     },
     {
       "epoch": 2.759269715791455,
-      "grad_norm": 6.138002872467041,
-      "learning_rate": 1.344438170525127e-05,
-      "loss": 0.7348,
+      "grad_norm": 6.247775554656982,
+      "learning_rate": 0.0002240730284208545,
+      "loss": 0.6382,
       "step": 14660
     },
     {
       "epoch": 2.761151891586674,
-      "grad_norm": 1.5578501224517822,
-      "learning_rate": 1.3433088650479955e-05,
-      "loss": 0.7188,
+      "grad_norm": 2.4420855045318604,
+      "learning_rate": 0.0002238848108413326,
+      "loss": 0.6792,
       "step": 14670
     },
     {
       "epoch": 2.7630340673818936,
-      "grad_norm": 24.87708282470703,
-      "learning_rate": 1.342179559570864e-05,
-      "loss": 0.8655,
+      "grad_norm": 5.8336687088012695,
+      "learning_rate": 0.00022369659326181067,
+      "loss": 0.8575,
       "step": 14680
     },
     {
       "epoch": 2.7649162431771126,
-      "grad_norm": 13.054447174072266,
-      "learning_rate": 1.3410502540937324e-05,
-      "loss": 0.6821,
+      "grad_norm": 10.038944244384766,
+      "learning_rate": 0.00022350837568228875,
+      "loss": 0.6917,
       "step": 14690
     },
     {
       "epoch": 2.766798418972332,
-      "grad_norm": 0.5486339926719666,
-      "learning_rate": 1.3399209486166007e-05,
-      "loss": 0.5678,
+      "grad_norm": 2.5733683109283447,
+      "learning_rate": 0.0002233201581027668,
+      "loss": 0.7355,
       "step": 14700
     },
     {
       "epoch": 2.7686805947675515,
-      "grad_norm": 16.782821655273438,
-      "learning_rate": 1.3387916431394693e-05,
-      "loss": 0.8178,
+      "grad_norm": 2.4171252250671387,
+      "learning_rate": 0.00022313194052324487,
+      "loss": 0.6444,
       "step": 14710
     },
     {
       "epoch": 2.7705627705627704,
-      "grad_norm": 5.81451416015625,
-      "learning_rate": 1.3376623376623377e-05,
-      "loss": 0.5588,
+      "grad_norm": 1.818503975868225,
+      "learning_rate": 0.00022294372294372295,
+      "loss": 0.4167,
       "step": 14720
     },
     {
       "epoch": 2.77244494635799,
-      "grad_norm": 3.6618595123291016,
-      "learning_rate": 1.3365330321852062e-05,
-      "loss": 0.7615,
+      "grad_norm": 1.2341984510421753,
+      "learning_rate": 0.00022275550536420103,
+      "loss": 0.8849,
       "step": 14730
     },
     {
       "epoch": 2.774327122153209,
-      "grad_norm": 8.5778169631958,
-      "learning_rate": 1.3354037267080746e-05,
-      "loss": 0.576,
+      "grad_norm": 1.8926407098770142,
+      "learning_rate": 0.0002225672877846791,
+      "loss": 0.7797,
       "step": 14740
     },
     {
       "epoch": 2.7762092979484283,
-      "grad_norm": 13.566119194030762,
-      "learning_rate": 1.3342744212309429e-05,
-      "loss": 0.5681,
+      "grad_norm": 3.853085517883301,
+      "learning_rate": 0.00022237907020515715,
+      "loss": 0.7213,
       "step": 14750
     },
     {
       "epoch": 2.7780914737436477,
-      "grad_norm": 15.843841552734375,
-      "learning_rate": 1.3331451157538115e-05,
-      "loss": 0.6232,
+      "grad_norm": 3.408320188522339,
+      "learning_rate": 0.00022219085262563523,
+      "loss": 0.7639,
       "step": 14760
     },
     {
       "epoch": 2.779973649538867,
-      "grad_norm": 8.985896110534668,
-      "learning_rate": 1.33201581027668e-05,
-      "loss": 0.538,
+      "grad_norm": 2.5825889110565186,
+      "learning_rate": 0.0002220026350461133,
+      "loss": 0.3973,
       "step": 14770
     },
     {
       "epoch": 2.781855825334086,
-      "grad_norm": 17.511743545532227,
-      "learning_rate": 1.3308865047995484e-05,
-      "loss": 0.8905,
+      "grad_norm": 3.6277236938476562,
+      "learning_rate": 0.0002218144174665914,
+      "loss": 0.7533,
       "step": 14780
     },
     {
       "epoch": 2.7837380011293056,
-      "grad_norm": 23.286775588989258,
-      "learning_rate": 1.3297571993224166e-05,
-      "loss": 0.8913,
+      "grad_norm": 2.047395706176758,
+      "learning_rate": 0.00022162619988706946,
+      "loss": 0.9952,
       "step": 14790
     },
     {
       "epoch": 2.7856201769245246,
-      "grad_norm": 23.496522903442383,
-      "learning_rate": 1.3286278938452852e-05,
-      "loss": 0.6562,
+      "grad_norm": 4.63572883605957,
+      "learning_rate": 0.00022143798230754753,
+      "loss": 0.8083,
       "step": 14800
     },
     {
       "epoch": 2.787502352719744,
-      "grad_norm": 11.973505973815918,
-      "learning_rate": 1.3274985883681537e-05,
-      "loss": 0.6814,
+      "grad_norm": 2.965135097503662,
+      "learning_rate": 0.0002212497647280256,
+      "loss": 0.7247,
       "step": 14810
     },
     {
       "epoch": 2.7893845285149634,
-      "grad_norm": 19.595155715942383,
-      "learning_rate": 1.3263692828910221e-05,
-      "loss": 0.4778,
+      "grad_norm": 6.954098701477051,
+      "learning_rate": 0.00022106154714850369,
+      "loss": 0.5122,
       "step": 14820
     },
     {
       "epoch": 2.7912667043101824,
-      "grad_norm": 10.94100284576416,
-      "learning_rate": 1.3252399774138904e-05,
-      "loss": 0.8547,
+      "grad_norm": 1.5478140115737915,
+      "learning_rate": 0.00022087332956898174,
+      "loss": 0.9916,
       "step": 14830
     },
     {
       "epoch": 2.793148880105402,
-      "grad_norm": 9.172542572021484,
-      "learning_rate": 1.3241106719367588e-05,
-      "loss": 0.8835,
+      "grad_norm": 7.099700450897217,
+      "learning_rate": 0.0002206851119894598,
+      "loss": 0.883,
       "step": 14840
     },
     {
       "epoch": 2.795031055900621,
-      "grad_norm": 49.06556701660156,
-      "learning_rate": 1.3229813664596274e-05,
-      "loss": 0.7209,
+      "grad_norm": 1.6341519355773926,
+      "learning_rate": 0.0002204968944099379,
+      "loss": 0.8133,
       "step": 14850
     },
     {
       "epoch": 2.7969132316958403,
-      "grad_norm": 2.8272016048431396,
-      "learning_rate": 1.3218520609824959e-05,
-      "loss": 0.6661,
+      "grad_norm": 1.9076987504959106,
+      "learning_rate": 0.00022030867683041596,
+      "loss": 0.7103,
       "step": 14860
     },
     {
       "epoch": 2.7987954074910597,
-      "grad_norm": 0.3546334505081177,
-      "learning_rate": 1.3207227555053643e-05,
-      "loss": 0.4056,
+      "grad_norm": 2.6881508827209473,
+      "learning_rate": 0.00022012045925089404,
+      "loss": 0.5018,
       "step": 14870
     },
     {
       "epoch": 2.800677583286279,
-      "grad_norm": 3.786240816116333,
-      "learning_rate": 1.3195934500282326e-05,
-      "loss": 0.964,
+      "grad_norm": 1.636492133140564,
+      "learning_rate": 0.0002199322416713721,
+      "loss": 0.8002,
       "step": 14880
     },
     {
       "epoch": 2.802559759081498,
-      "grad_norm": 5.78407096862793,
-      "learning_rate": 1.318464144551101e-05,
-      "loss": 0.6051,
+      "grad_norm": 3.2086358070373535,
+      "learning_rate": 0.0002197440240918502,
+      "loss": 0.7168,
       "step": 14890
     },
     {
       "epoch": 2.8044419348767176,
-      "grad_norm": 6.689875602722168,
-      "learning_rate": 1.3173348390739696e-05,
-      "loss": 0.6458,
+      "grad_norm": 1.55039381980896,
+      "learning_rate": 0.00021955580651232827,
+      "loss": 0.5735,
       "step": 14900
     },
     {
       "epoch": 2.8063241106719365,
-      "grad_norm": 12.659087181091309,
-      "learning_rate": 1.316205533596838e-05,
-      "loss": 0.8754,
+      "grad_norm": 4.564043045043945,
+      "learning_rate": 0.00021936758893280635,
+      "loss": 0.7911,
       "step": 14910
     },
     {
       "epoch": 2.808206286467156,
-      "grad_norm": 14.06718921661377,
-      "learning_rate": 1.3150762281197063e-05,
-      "loss": 0.9916,
+      "grad_norm": 5.466314792633057,
+      "learning_rate": 0.0002191793713532844,
+      "loss": 1.0529,
       "step": 14920
     },
     {
       "epoch": 2.8100884622623754,
-      "grad_norm": 15.806565284729004,
-      "learning_rate": 1.3139469226425748e-05,
-      "loss": 0.7995,
+      "grad_norm": 3.9521117210388184,
+      "learning_rate": 0.00021899115377376247,
+      "loss": 0.7721,
       "step": 14930
     },
     {
       "epoch": 2.811970638057595,
-      "grad_norm": 1.6029272079467773,
-      "learning_rate": 1.3128176171654434e-05,
-      "loss": 0.4056,
+      "grad_norm": 2.2904415130615234,
+      "learning_rate": 0.00021880293619424055,
+      "loss": 0.3554,
       "step": 14940
     },
     {
       "epoch": 2.813852813852814,
-      "grad_norm": 11.386082649230957,
-      "learning_rate": 1.3116883116883118e-05,
-      "loss": 0.6312,
+      "grad_norm": 2.367241859436035,
+      "learning_rate": 0.00021861471861471863,
+      "loss": 0.5897,
       "step": 14950
     },
     {
       "epoch": 2.8157349896480333,
-      "grad_norm": 12.017684936523438,
-      "learning_rate": 1.31055900621118e-05,
-      "loss": 0.8528,
+      "grad_norm": 6.305996894836426,
+      "learning_rate": 0.00021842650103519667,
+      "loss": 0.7687,
       "step": 14960
     },
     {
       "epoch": 2.8176171654432522,
-      "grad_norm": 10.35236644744873,
-      "learning_rate": 1.3094297007340485e-05,
-      "loss": 0.63,
+      "grad_norm": 4.461951732635498,
+      "learning_rate": 0.00021823828345567475,
+      "loss": 0.4529,
       "step": 14970
     },
     {
       "epoch": 2.8194993412384717,
-      "grad_norm": 18.851110458374023,
-      "learning_rate": 1.308300395256917e-05,
-      "loss": 0.7506,
+      "grad_norm": 1.6320971250534058,
+      "learning_rate": 0.00021805006587615283,
+      "loss": 0.7452,
       "step": 14980
     },
     {
       "epoch": 2.821381517033691,
-      "grad_norm": 7.350698947906494,
-      "learning_rate": 1.3071710897797856e-05,
-      "loss": 0.5077,
+      "grad_norm": 1.834617018699646,
+      "learning_rate": 0.0002178618482966309,
+      "loss": 0.6039,
       "step": 14990
     },
     {
       "epoch": 2.82326369282891,
-      "grad_norm": 9.5648775100708,
-      "learning_rate": 1.306041784302654e-05,
-      "loss": 0.6446,
+      "grad_norm": 1.4062933921813965,
+      "learning_rate": 0.000217673630717109,
+      "loss": 0.6868,
       "step": 15000
     },
     {
       "epoch": 2.8251458686241295,
-      "grad_norm": 46.55702209472656,
-      "learning_rate": 1.3049124788255223e-05,
-      "loss": 0.6478,
+      "grad_norm": 3.7862212657928467,
+      "learning_rate": 0.00021748541313758706,
+      "loss": 0.5454,
       "step": 15010
     },
     {
       "epoch": 2.8270280444193485,
-      "grad_norm": 22.763042449951172,
-      "learning_rate": 1.3037831733483907e-05,
-      "loss": 0.5402,
+      "grad_norm": 6.517659664154053,
+      "learning_rate": 0.00021729719555806513,
+      "loss": 0.5488,
       "step": 15020
     },
     {
       "epoch": 2.828910220214568,
-      "grad_norm": 23.328462600708008,
-      "learning_rate": 1.3026538678712591e-05,
-      "loss": 0.7589,
+      "grad_norm": 2.6208972930908203,
+      "learning_rate": 0.0002171089779785432,
+      "loss": 0.6687,
       "step": 15030
     },
     {
       "epoch": 2.8307923960097874,
-      "grad_norm": 9.546448707580566,
-      "learning_rate": 1.3015245623941278e-05,
-      "loss": 0.4769,
+      "grad_norm": 2.6357429027557373,
+      "learning_rate": 0.00021692076039902129,
+      "loss": 0.4496,
       "step": 15040
     },
     {
       "epoch": 2.832674571805007,
-      "grad_norm": 9.505398750305176,
-      "learning_rate": 1.300395256916996e-05,
-      "loss": 0.6579,
+      "grad_norm": 0.8929344415664673,
+      "learning_rate": 0.00021673254281949934,
+      "loss": 0.6089,
       "step": 15050
     },
     {
       "epoch": 2.834556747600226,
-      "grad_norm": 10.608928680419922,
-      "learning_rate": 1.2992659514398645e-05,
-      "loss": 0.6198,
+      "grad_norm": 1.9503272771835327,
+      "learning_rate": 0.0002165443252399774,
+      "loss": 0.6058,
       "step": 15060
     },
     {
       "epoch": 2.8364389233954452,
-      "grad_norm": 13.843592643737793,
-      "learning_rate": 1.2981366459627329e-05,
-      "loss": 0.74,
+      "grad_norm": 4.708406448364258,
+      "learning_rate": 0.0002163561076604555,
+      "loss": 0.9228,
       "step": 15070
     },
     {
       "epoch": 2.838321099190664,
-      "grad_norm": 18.67850112915039,
-      "learning_rate": 1.2970073404856015e-05,
-      "loss": 0.4672,
+      "grad_norm": 3.6658589839935303,
+      "learning_rate": 0.00021616789008093356,
+      "loss": 0.4709,
       "step": 15080
     },
     {
       "epoch": 2.8402032749858837,
-      "grad_norm": 6.967514991760254,
-      "learning_rate": 1.2958780350084698e-05,
-      "loss": 0.4409,
+      "grad_norm": 2.2713499069213867,
+      "learning_rate": 0.00021597967250141161,
+      "loss": 0.5153,
       "step": 15090
     },
     {
       "epoch": 2.842085450781103,
-      "grad_norm": 6.820723533630371,
-      "learning_rate": 1.2947487295313382e-05,
-      "loss": 0.7573,
+      "grad_norm": 1.2780481576919556,
+      "learning_rate": 0.0002157914549218897,
+      "loss": 0.7618,
       "step": 15100
     },
     {
       "epoch": 2.843967626576322,
-      "grad_norm": 18.13920783996582,
-      "learning_rate": 1.2936194240542067e-05,
-      "loss": 0.7914,
+      "grad_norm": 3.080092430114746,
+      "learning_rate": 0.0002156032373423678,
+      "loss": 0.791,
       "step": 15110
     },
     {
       "epoch": 2.8458498023715415,
-      "grad_norm": 7.37209939956665,
-      "learning_rate": 1.2924901185770751e-05,
-      "loss": 0.6351,
+      "grad_norm": 1.203285813331604,
+      "learning_rate": 0.00021541501976284587,
+      "loss": 0.9003,
       "step": 15120
     },
     {
       "epoch": 2.847731978166761,
-      "grad_norm": 0.3693366050720215,
-      "learning_rate": 1.2913608130999437e-05,
-      "loss": 0.5822,
+      "grad_norm": 0.1718599647283554,
+      "learning_rate": 0.00021522680218332395,
+      "loss": 0.5085,
       "step": 15130
     },
     {
       "epoch": 2.84961415396198,
-      "grad_norm": 21.349327087402344,
-      "learning_rate": 1.290231507622812e-05,
-      "loss": 0.6001,
+      "grad_norm": 1.1150635480880737,
+      "learning_rate": 0.000215038584603802,
+      "loss": 0.5646,
       "step": 15140
     },
     {
       "epoch": 2.8514963297571994,
-      "grad_norm": 8.462204933166504,
-      "learning_rate": 1.2891022021456804e-05,
-      "loss": 0.4354,
+      "grad_norm": 3.6562788486480713,
+      "learning_rate": 0.00021485036702428007,
+      "loss": 0.4479,
       "step": 15150
     },
     {
       "epoch": 2.853378505552419,
-      "grad_norm": 9.716917991638184,
-      "learning_rate": 1.2879728966685488e-05,
-      "loss": 0.7532,
+      "grad_norm": 3.168921947479248,
+      "learning_rate": 0.00021466214944475815,
+      "loss": 0.7417,
       "step": 15160
     },
     {
       "epoch": 2.8552606813476378,
-      "grad_norm": 4.279250144958496,
-      "learning_rate": 1.2868435911914173e-05,
-      "loss": 0.7497,
+      "grad_norm": 1.4422719478607178,
+      "learning_rate": 0.00021447393186523623,
+      "loss": 0.5953,
       "step": 15170
     },
     {
       "epoch": 2.857142857142857,
-      "grad_norm": 27.534521102905273,
-      "learning_rate": 1.2857142857142857e-05,
-      "loss": 0.5156,
+      "grad_norm": 6.2547221183776855,
+      "learning_rate": 0.00021428571428571427,
+      "loss": 0.4218,
       "step": 15180
     },
     {
       "epoch": 2.859025032938076,
-      "grad_norm": 10.771012306213379,
-      "learning_rate": 1.2845849802371542e-05,
-      "loss": 0.7265,
+      "grad_norm": 2.6894936561584473,
+      "learning_rate": 0.00021409749670619235,
+      "loss": 0.7421,
       "step": 15190
     },
     {
       "epoch": 2.8609072087332956,
-      "grad_norm": 18.646312713623047,
-      "learning_rate": 1.2834556747600226e-05,
-      "loss": 0.6741,
+      "grad_norm": 3.028836488723755,
+      "learning_rate": 0.00021390927912667043,
+      "loss": 0.6819,
       "step": 15200
     },
     {
       "epoch": 2.862789384528515,
-      "grad_norm": 19.257171630859375,
-      "learning_rate": 1.282326369282891e-05,
-      "loss": 1.1365,
+      "grad_norm": 8.584593772888184,
+      "learning_rate": 0.0002137210615471485,
+      "loss": 1.0062,
       "step": 15210
     },
     {
       "epoch": 2.8646715603237345,
-      "grad_norm": 5.380772590637207,
-      "learning_rate": 1.2811970638057596e-05,
-      "loss": 0.8546,
+      "grad_norm": 1.5320723056793213,
+      "learning_rate": 0.0002135328439676266,
+      "loss": 0.7337,
       "step": 15220
     },
     {
       "epoch": 2.8665537361189535,
-      "grad_norm": 23.16881561279297,
-      "learning_rate": 1.2800677583286279e-05,
-      "loss": 0.7943,
+      "grad_norm": 4.4340033531188965,
+      "learning_rate": 0.00021334462638810466,
+      "loss": 0.7246,
       "step": 15230
     },
     {
       "epoch": 2.868435911914173,
-      "grad_norm": 3.4778006076812744,
-      "learning_rate": 1.2789384528514964e-05,
-      "loss": 0.6461,
+      "grad_norm": 2.8282470703125,
+      "learning_rate": 0.00021315640880858273,
+      "loss": 0.6766,
       "step": 15240
     },
     {
       "epoch": 2.870318087709392,
-      "grad_norm": 6.4198994636535645,
-      "learning_rate": 1.2778091473743648e-05,
-      "loss": 0.8293,
+      "grad_norm": 3.918661117553711,
+      "learning_rate": 0.0002129681912290608,
+      "loss": 0.8459,
       "step": 15250
     },
     {
       "epoch": 2.8722002635046113,
-      "grad_norm": 9.544682502746582,
-      "learning_rate": 1.2766798418972332e-05,
-      "loss": 0.7434,
+      "grad_norm": 2.269125461578369,
+      "learning_rate": 0.00021277997364953889,
+      "loss": 0.7308,
       "step": 15260
     },
     {
       "epoch": 2.8740824392998308,
-      "grad_norm": 20.545665740966797,
-      "learning_rate": 1.2755505364201017e-05,
-      "loss": 0.6885,
+      "grad_norm": 5.307600021362305,
+      "learning_rate": 0.00021259175607001694,
+      "loss": 0.7939,
       "step": 15270
     },
     {
       "epoch": 2.8759646150950497,
-      "grad_norm": 21.490854263305664,
-      "learning_rate": 1.2744212309429701e-05,
-      "loss": 0.9683,
+      "grad_norm": 5.625484466552734,
+      "learning_rate": 0.000212403538490495,
+      "loss": 0.747,
       "step": 15280
     },
     {
       "epoch": 2.877846790890269,
-      "grad_norm": 3.391572952270508,
-      "learning_rate": 1.2732919254658385e-05,
-      "loss": 0.421,
+      "grad_norm": 0.8601845502853394,
+      "learning_rate": 0.0002122153209109731,
+      "loss": 0.4009,
       "step": 15290
     },
     {
       "epoch": 2.879728966685488,
-      "grad_norm": 6.413089275360107,
-      "learning_rate": 1.272162619988707e-05,
-      "loss": 0.6439,
+      "grad_norm": 4.669708251953125,
+      "learning_rate": 0.00021202710333145116,
+      "loss": 0.6273,
       "step": 15300
     },
     {
       "epoch": 2.8816111424807076,
-      "grad_norm": 9.638285636901855,
-      "learning_rate": 1.2710333145115754e-05,
-      "loss": 0.6487,
+      "grad_norm": 2.922671318054199,
+      "learning_rate": 0.00021183888575192921,
+      "loss": 0.846,
       "step": 15310
     },
     {
       "epoch": 2.883493318275927,
-      "grad_norm": 19.76523208618164,
-      "learning_rate": 1.2699040090344439e-05,
-      "loss": 0.7171,
+      "grad_norm": 3.249382734298706,
+      "learning_rate": 0.0002116506681724073,
+      "loss": 0.8076,
       "step": 15320
     },
     {
       "epoch": 2.8853754940711465,
-      "grad_norm": 18.670621871948242,
-      "learning_rate": 1.2687747035573123e-05,
-      "loss": 0.4666,
+      "grad_norm": 2.7542879581451416,
+      "learning_rate": 0.00021146245059288537,
+      "loss": 0.5032,
       "step": 15330
     },
     {
       "epoch": 2.8872576698663655,
-      "grad_norm": 11.9641695022583,
-      "learning_rate": 1.2676453980801807e-05,
-      "loss": 0.5461,
+      "grad_norm": 4.564001083374023,
+      "learning_rate": 0.00021127423301336347,
+      "loss": 0.6309,
       "step": 15340
     },
     {
       "epoch": 2.889139845661585,
-      "grad_norm": 1.4184321165084839,
-      "learning_rate": 1.2665160926030492e-05,
-      "loss": 0.407,
+      "grad_norm": 0.7549136877059937,
+      "learning_rate": 0.00021108601543384155,
+      "loss": 0.2515,
       "step": 15350
     },
     {
       "epoch": 2.891022021456804,
-      "grad_norm": 19.386415481567383,
-      "learning_rate": 1.2653867871259176e-05,
-      "loss": 0.4194,
+      "grad_norm": 2.5547895431518555,
+      "learning_rate": 0.0002108977978543196,
+      "loss": 0.5358,
       "step": 15360
     },
     {
       "epoch": 2.8929041972520233,
-      "grad_norm": 25.078582763671875,
-      "learning_rate": 1.264257481648786e-05,
-      "loss": 0.8741,
+      "grad_norm": 5.23292875289917,
+      "learning_rate": 0.00021070958027479767,
+      "loss": 0.8811,
       "step": 15370
     },
     {
       "epoch": 2.8947863730472427,
-      "grad_norm": 18.054174423217773,
-      "learning_rate": 1.2631281761716545e-05,
-      "loss": 0.8567,
+      "grad_norm": 5.233101844787598,
+      "learning_rate": 0.00021052136269527575,
+      "loss": 0.8276,
       "step": 15380
     },
     {
       "epoch": 2.8966685488424617,
-      "grad_norm": 9.56291675567627,
-      "learning_rate": 1.261998870694523e-05,
-      "loss": 1.0115,
+      "grad_norm": 4.63691520690918,
+      "learning_rate": 0.00021033314511575383,
+      "loss": 0.8278,
       "step": 15390
     },
     {
       "epoch": 2.898550724637681,
-      "grad_norm": 10.484170913696289,
-      "learning_rate": 1.2608695652173912e-05,
-      "loss": 0.5493,
+      "grad_norm": 5.9823479652404785,
+      "learning_rate": 0.00021014492753623187,
+      "loss": 0.5929,
       "step": 15400
     },
     {
       "epoch": 2.9004329004329006,
-      "grad_norm": 17.50212287902832,
-      "learning_rate": 1.2597402597402598e-05,
-      "loss": 0.8972,
+      "grad_norm": 5.181570529937744,
+      "learning_rate": 0.00020995670995670995,
+      "loss": 0.9555,
       "step": 15410
     },
     {
       "epoch": 2.9023150762281196,
-      "grad_norm": 11.050121307373047,
-      "learning_rate": 1.2586109542631282e-05,
-      "loss": 0.572,
+      "grad_norm": 1.9936301708221436,
+      "learning_rate": 0.00020976849237718803,
+      "loss": 0.4135,
       "step": 15420
     },
     {
       "epoch": 2.904197252023339,
-      "grad_norm": 19.2404727935791,
-      "learning_rate": 1.2574816487859967e-05,
-      "loss": 0.6457,
+      "grad_norm": 5.654709815979004,
+      "learning_rate": 0.0002095802747976661,
+      "loss": 0.7488,
       "step": 15430
     },
     {
       "epoch": 2.9060794278185584,
-      "grad_norm": 7.877418518066406,
-      "learning_rate": 1.256352343308865e-05,
-      "loss": 0.5262,
+      "grad_norm": 3.6621265411376953,
+      "learning_rate": 0.00020939205721814415,
+      "loss": 0.4353,
       "step": 15440
     },
     {
       "epoch": 2.9079616036137774,
-      "grad_norm": 25.962966918945312,
-      "learning_rate": 1.2552230378317336e-05,
-      "loss": 1.0387,
+      "grad_norm": 3.0909690856933594,
+      "learning_rate": 0.00020920383963862226,
+      "loss": 0.8345,
       "step": 15450
     },
     {
       "epoch": 2.909843779408997,
-      "grad_norm": 18.10149383544922,
-      "learning_rate": 1.254093732354602e-05,
-      "loss": 0.3264,
+      "grad_norm": 1.9756587743759155,
+      "learning_rate": 0.00020901562205910033,
+      "loss": 0.3801,
       "step": 15460
     },
     {
       "epoch": 2.911725955204216,
-      "grad_norm": 12.390528678894043,
-      "learning_rate": 1.2529644268774704e-05,
-      "loss": 0.5733,
+      "grad_norm": 1.5738675594329834,
+      "learning_rate": 0.0002088274044795784,
+      "loss": 0.4169,
       "step": 15470
     },
     {
       "epoch": 2.9136081309994353,
-      "grad_norm": 6.335510730743408,
-      "learning_rate": 1.2518351214003389e-05,
-      "loss": 0.9647,
+      "grad_norm": 1.9008299112319946,
+      "learning_rate": 0.00020863918690005649,
+      "loss": 1.0148,
       "step": 15480
     },
     {
       "epoch": 2.9154903067946547,
-      "grad_norm": 8.489152908325195,
-      "learning_rate": 1.2507058159232071e-05,
-      "loss": 0.5086,
+      "grad_norm": 1.7302570343017578,
+      "learning_rate": 0.00020845096932053454,
+      "loss": 0.554,
       "step": 15490
     },
     {
       "epoch": 2.917372482589874,
-      "grad_norm": 5.623629570007324,
-      "learning_rate": 1.2495765104460757e-05,
-      "loss": 0.8003,
+      "grad_norm": 1.9822744131088257,
+      "learning_rate": 0.0002082627517410126,
+      "loss": 0.7925,
       "step": 15500
     },
     {
       "epoch": 2.919254658385093,
-      "grad_norm": 7.3673787117004395,
-      "learning_rate": 1.2484472049689442e-05,
-      "loss": 0.5053,
+      "grad_norm": 1.7959871292114258,
+      "learning_rate": 0.0002080745341614907,
+      "loss": 0.5921,
       "step": 15510
     },
     {
       "epoch": 2.9211368341803126,
-      "grad_norm": 25.399425506591797,
-      "learning_rate": 1.2473178994918126e-05,
-      "loss": 0.6455,
+      "grad_norm": 2.998852014541626,
+      "learning_rate": 0.00020788631658196876,
+      "loss": 0.6003,
       "step": 15520
     },
     {
       "epoch": 2.9230190099755315,
-      "grad_norm": 15.276825904846191,
-      "learning_rate": 1.2461885940146809e-05,
-      "loss": 0.5071,
+      "grad_norm": 1.354353427886963,
+      "learning_rate": 0.00020769809900244681,
+      "loss": 0.52,
       "step": 15530
     },
     {
       "epoch": 2.924901185770751,
-      "grad_norm": 11.166011810302734,
-      "learning_rate": 1.2450592885375493e-05,
-      "loss": 0.8913,
+      "grad_norm": 2.2639453411102295,
+      "learning_rate": 0.0002075098814229249,
+      "loss": 0.7204,
       "step": 15540
     },
     {
       "epoch": 2.9267833615659704,
-      "grad_norm": 16.122472763061523,
-      "learning_rate": 1.243929983060418e-05,
-      "loss": 1.0398,
+      "grad_norm": 1.4416176080703735,
+      "learning_rate": 0.00020732166384340297,
+      "loss": 1.0761,
       "step": 15550
     },
     {
       "epoch": 2.9286655373611894,
-      "grad_norm": 21.32600212097168,
-      "learning_rate": 1.2428006775832864e-05,
-      "loss": 0.6833,
+      "grad_norm": 3.021568775177002,
+      "learning_rate": 0.00020713344626388107,
+      "loss": 0.6917,
       "step": 15560
     },
     {
       "epoch": 2.930547713156409,
-      "grad_norm": 3.505770206451416,
-      "learning_rate": 1.2416713721061546e-05,
-      "loss": 0.7454,
+      "grad_norm": 1.8868095874786377,
+      "learning_rate": 0.00020694522868435912,
+      "loss": 0.7583,
       "step": 15570
     },
     {
       "epoch": 2.932429888951628,
-      "grad_norm": 18.41620445251465,
-      "learning_rate": 1.240542066629023e-05,
-      "loss": 0.5258,
+      "grad_norm": 2.691807508468628,
+      "learning_rate": 0.0002067570111048372,
+      "loss": 0.5029,
       "step": 15580
     },
     {
       "epoch": 2.9343120647468472,
-      "grad_norm": 25.811620712280273,
-      "learning_rate": 1.2394127611518917e-05,
-      "loss": 0.607,
+      "grad_norm": 2.2595269680023193,
+      "learning_rate": 0.00020656879352531527,
+      "loss": 0.5884,
       "step": 15590
     },
     {
       "epoch": 2.9361942405420667,
-      "grad_norm": 40.27906036376953,
-      "learning_rate": 1.2382834556747601e-05,
-      "loss": 0.8219,
+      "grad_norm": 3.909071683883667,
+      "learning_rate": 0.00020638057594579335,
+      "loss": 0.8451,
       "step": 15600
     },
     {
       "epoch": 2.938076416337286,
-      "grad_norm": 24.22795867919922,
-      "learning_rate": 1.2371541501976286e-05,
-      "loss": 0.943,
+      "grad_norm": 6.179553985595703,
+      "learning_rate": 0.00020619235836627143,
+      "loss": 0.801,
       "step": 15610
     },
     {
       "epoch": 2.939958592132505,
-      "grad_norm": 25.92055892944336,
-      "learning_rate": 1.2360248447204968e-05,
-      "loss": 0.8276,
+      "grad_norm": 2.8360793590545654,
+      "learning_rate": 0.00020600414078674947,
+      "loss": 0.7074,
       "step": 15620
     },
     {
       "epoch": 2.9418407679277245,
-      "grad_norm": 8.324178695678711,
-      "learning_rate": 1.2348955392433653e-05,
-      "loss": 0.6895,
+      "grad_norm": 5.890757083892822,
+      "learning_rate": 0.00020581592320722755,
+      "loss": 0.686,
       "step": 15630
     },
     {
       "epoch": 2.9437229437229435,
-      "grad_norm": 22.274948120117188,
-      "learning_rate": 1.2337662337662339e-05,
-      "loss": 1.0351,
+      "grad_norm": 3.1145524978637695,
+      "learning_rate": 0.00020562770562770563,
+      "loss": 0.9267,
       "step": 15640
     },
     {
       "epoch": 2.945605119518163,
-      "grad_norm": 40.22727584838867,
-      "learning_rate": 1.2326369282891023e-05,
-      "loss": 0.5756,
+      "grad_norm": 3.2289531230926514,
+      "learning_rate": 0.0002054394880481837,
+      "loss": 0.5452,
       "step": 15650
     },
     {
       "epoch": 2.9474872953133824,
-      "grad_norm": 2.7665634155273438,
-      "learning_rate": 1.2315076228119706e-05,
-      "loss": 0.5928,
+      "grad_norm": 1.3664658069610596,
+      "learning_rate": 0.00020525127046866175,
+      "loss": 0.6293,
       "step": 15660
     },
     {
       "epoch": 2.949369471108602,
-      "grad_norm": 20.086702346801758,
-      "learning_rate": 1.230378317334839e-05,
-      "loss": 0.7117,
+      "grad_norm": 2.922199010848999,
+      "learning_rate": 0.00020506305288913986,
+      "loss": 0.8441,
       "step": 15670
     },
     {
       "epoch": 2.951251646903821,
-      "grad_norm": 13.280003547668457,
-      "learning_rate": 1.2292490118577076e-05,
-      "loss": 0.7403,
+      "grad_norm": 3.505491256713867,
+      "learning_rate": 0.00020487483530961793,
+      "loss": 0.6791,
       "step": 15680
     },
     {
       "epoch": 2.9531338226990402,
-      "grad_norm": 37.245201110839844,
-      "learning_rate": 1.228119706380576e-05,
-      "loss": 0.8726,
+      "grad_norm": 7.177020072937012,
+      "learning_rate": 0.000204686617730096,
+      "loss": 0.8528,
       "step": 15690
     },
     {
       "epoch": 2.955015998494259,
-      "grad_norm": 6.580207824707031,
-      "learning_rate": 1.2269904009034443e-05,
-      "loss": 0.6027,
+      "grad_norm": 2.837167739868164,
+      "learning_rate": 0.00020449840015057406,
+      "loss": 0.5434,
       "step": 15700
     },
     {
       "epoch": 2.9568981742894787,
-      "grad_norm": 1.7421516180038452,
-      "learning_rate": 1.2258610954263128e-05,
-      "loss": 0.6724,
+      "grad_norm": 6.894123077392578,
+      "learning_rate": 0.00020431018257105214,
+      "loss": 0.7725,
       "step": 15710
     },
     {
       "epoch": 2.958780350084698,
-      "grad_norm": 29.77842140197754,
-      "learning_rate": 1.2247317899491812e-05,
-      "loss": 0.5805,
+      "grad_norm": 3.7151124477386475,
+      "learning_rate": 0.0002041219649915302,
+      "loss": 0.5149,
       "step": 15720
     },
     {
       "epoch": 2.960662525879917,
-      "grad_norm": 26.471179962158203,
-      "learning_rate": 1.2236024844720498e-05,
-      "loss": 0.7421,
+      "grad_norm": 3.3849830627441406,
+      "learning_rate": 0.0002039337474120083,
+      "loss": 0.6201,
       "step": 15730
     },
     {
       "epoch": 2.9625447016751365,
-      "grad_norm": 8.780817031860352,
-      "learning_rate": 1.2224731789949183e-05,
-      "loss": 0.7289,
+      "grad_norm": 2.0344691276550293,
+      "learning_rate": 0.00020374552983248636,
+      "loss": 0.7283,
       "step": 15740
     },
     {
       "epoch": 2.9644268774703555,
-      "grad_norm": 14.556037902832031,
-      "learning_rate": 1.2213438735177865e-05,
-      "loss": 0.4937,
+      "grad_norm": 1.9597744941711426,
+      "learning_rate": 0.00020355731225296441,
+      "loss": 0.4933,
       "step": 15750
     },
     {
       "epoch": 2.966309053265575,
-      "grad_norm": 25.941247940063477,
-      "learning_rate": 1.220214568040655e-05,
-      "loss": 0.8835,
+      "grad_norm": 4.379768371582031,
+      "learning_rate": 0.0002033690946734425,
+      "loss": 0.8107,
       "step": 15760
     },
     {
       "epoch": 2.9681912290607944,
-      "grad_norm": 6.276753902435303,
-      "learning_rate": 1.2190852625635234e-05,
-      "loss": 0.6071,
+      "grad_norm": 2.1065194606781006,
+      "learning_rate": 0.00020318087709392057,
+      "loss": 0.5086,
       "step": 15770
     },
     {
       "epoch": 2.970073404856014,
-      "grad_norm": 24.19154930114746,
-      "learning_rate": 1.217955957086392e-05,
-      "loss": 0.907,
+      "grad_norm": 3.758819341659546,
+      "learning_rate": 0.00020299265951439867,
+      "loss": 0.9853,
       "step": 15780
     },
     {
       "epoch": 2.9719555806512328,
-      "grad_norm": 5.083404064178467,
-      "learning_rate": 1.2168266516092603e-05,
-      "loss": 0.5493,
+      "grad_norm": 0.23007777333259583,
+      "learning_rate": 0.00020280444193487672,
+      "loss": 0.5216,
       "step": 15790
     },
     {
       "epoch": 2.973837756446452,
-      "grad_norm": 9.346044540405273,
-      "learning_rate": 1.2156973461321287e-05,
-      "loss": 0.6961,
+      "grad_norm": 5.689883232116699,
+      "learning_rate": 0.0002026162243553548,
+      "loss": 0.7901,
       "step": 15800
     },
     {
       "epoch": 2.975719932241671,
-      "grad_norm": 1.4552770853042603,
-      "learning_rate": 1.2145680406549972e-05,
-      "loss": 0.6451,
+      "grad_norm": 0.3998923599720001,
+      "learning_rate": 0.00020242800677583287,
+      "loss": 0.7314,
       "step": 15810
     },
     {
       "epoch": 2.9776021080368906,
-      "grad_norm": 17.02495765686035,
-      "learning_rate": 1.2134387351778658e-05,
-      "loss": 0.5733,
+      "grad_norm": 1.291471004486084,
+      "learning_rate": 0.00020223978919631095,
+      "loss": 0.6298,
       "step": 15820
     },
     {
       "epoch": 2.97948428383211,
-      "grad_norm": 2.2290263175964355,
-      "learning_rate": 1.2123094297007342e-05,
-      "loss": 0.7652,
+      "grad_norm": 1.6297998428344727,
+      "learning_rate": 0.00020205157161678903,
+      "loss": 0.721,
       "step": 15830
     },
     {
       "epoch": 2.981366459627329,
-      "grad_norm": 20.987024307250977,
-      "learning_rate": 1.2111801242236025e-05,
-      "loss": 0.7364,
+      "grad_norm": 0.40168535709381104,
+      "learning_rate": 0.00020186335403726707,
+      "loss": 0.7562,
       "step": 15840
     },
     {
       "epoch": 2.9832486354225485,
-      "grad_norm": 18.569358825683594,
-      "learning_rate": 1.2100508187464709e-05,
-      "loss": 1.0693,
+      "grad_norm": 3.6822330951690674,
+      "learning_rate": 0.00020167513645774515,
+      "loss": 0.9433,
       "step": 15850
     },
     {
       "epoch": 2.9851308112177675,
-      "grad_norm": 22.19905662536621,
-      "learning_rate": 1.2089215132693394e-05,
-      "loss": 0.8939,
+      "grad_norm": 6.4109787940979,
+      "learning_rate": 0.00020148691887822323,
+      "loss": 0.7135,
       "step": 15860
     },
     {
       "epoch": 2.987012987012987,
-      "grad_norm": 20.058185577392578,
-      "learning_rate": 1.207792207792208e-05,
-      "loss": 0.5969,
+      "grad_norm": 1.9936270713806152,
+      "learning_rate": 0.0002012987012987013,
+      "loss": 0.7199,
       "step": 15870
     },
     {
       "epoch": 2.9888951628082063,
-      "grad_norm": 14.325201034545898,
-      "learning_rate": 1.2066629023150762e-05,
-      "loss": 0.4321,
+      "grad_norm": 2.7041404247283936,
+      "learning_rate": 0.00020111048371917935,
+      "loss": 0.2948,
       "step": 15880
     },
     {
       "epoch": 2.9907773386034258,
-      "grad_norm": 3.7948050498962402,
-      "learning_rate": 1.2055335968379447e-05,
-      "loss": 0.6696,
+      "grad_norm": 3.060731887817383,
+      "learning_rate": 0.00020092226613965746,
+      "loss": 0.7353,
       "step": 15890
     },
     {
       "epoch": 2.9926595143986447,
-      "grad_norm": 10.829690933227539,
-      "learning_rate": 1.2044042913608131e-05,
-      "loss": 0.6765,
+      "grad_norm": 2.4952120780944824,
+      "learning_rate": 0.00020073404856013553,
+      "loss": 0.6693,
       "step": 15900
     },
     {
       "epoch": 2.994541690193864,
-      "grad_norm": 16.70258903503418,
-      "learning_rate": 1.2032749858836815e-05,
-      "loss": 0.5554,
+      "grad_norm": 3.944856643676758,
+      "learning_rate": 0.0002005458309806136,
+      "loss": 0.5033,
       "step": 15910
     },
     {
       "epoch": 2.996423865989083,
-      "grad_norm": 8.688410758972168,
-      "learning_rate": 1.20214568040655e-05,
-      "loss": 0.542,
+      "grad_norm": 2.602952480316162,
+      "learning_rate": 0.00020035761340109166,
+      "loss": 0.6788,
       "step": 15920
     },
     {
       "epoch": 2.9983060417843026,
-      "grad_norm": 7.034015655517578,
-      "learning_rate": 1.2010163749294184e-05,
-      "loss": 0.5749,
+      "grad_norm": 3.6875159740448,
+      "learning_rate": 0.00020016939582156974,
+      "loss": 0.8691,
       "step": 15930
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.9153333333333333,
-      "eval_loss": 0.32141387462615967,
-      "eval_runtime": 319.8636,
-      "eval_samples_per_second": 23.447,
-      "eval_steps_per_second": 2.932,
+      "eval_accuracy": 0.9173333333333333,
+      "eval_loss": 0.27245163917541504,
+      "eval_runtime": 467.1837,
+      "eval_samples_per_second": 16.054,
+      "eval_steps_per_second": 2.008,
       "step": 15939
     },
     {
       "epoch": 3.000188217579522,
-      "grad_norm": 13.433454513549805,
-      "learning_rate": 1.1998870694522869e-05,
-      "loss": 0.4825,
+      "grad_norm": 5.088705062866211,
+      "learning_rate": 0.0001999811782420478,
+      "loss": 0.5685,
       "step": 15940
     },
     {
       "epoch": 3.002070393374741,
-      "grad_norm": 18.549455642700195,
-      "learning_rate": 1.1987577639751553e-05,
-      "loss": 0.7692,
+      "grad_norm": 4.702372074127197,
+      "learning_rate": 0.0001997929606625259,
+      "loss": 0.8557,
       "step": 15950
     },
     {
       "epoch": 3.0039525691699605,
-      "grad_norm": 22.799373626708984,
-      "learning_rate": 1.1976284584980239e-05,
-      "loss": 0.6418,
+      "grad_norm": 2.6750690937042236,
+      "learning_rate": 0.00019960474308300396,
+      "loss": 0.6402,
       "step": 15960
     },
     {
       "epoch": 3.00583474496518,
-      "grad_norm": 21.21796989440918,
-      "learning_rate": 1.1964991530208922e-05,
-      "loss": 0.5274,
+      "grad_norm": 3.8041257858276367,
+      "learning_rate": 0.00019941652550348201,
+      "loss": 0.5554,
       "step": 15970
     },
     {
       "epoch": 3.007716920760399,
-      "grad_norm": 25.831186294555664,
-      "learning_rate": 1.1953698475437606e-05,
-      "loss": 0.5339,
+      "grad_norm": 2.85681414604187,
+      "learning_rate": 0.0001992283079239601,
+      "loss": 0.617,
       "step": 15980
     },
     {
       "epoch": 3.0095990965556183,
-      "grad_norm": 9.926560401916504,
-      "learning_rate": 1.194240542066629e-05,
-      "loss": 0.8981,
+      "grad_norm": 2.0742359161376953,
+      "learning_rate": 0.00019904009034443817,
+      "loss": 0.8394,
       "step": 15990
     },
     {
       "epoch": 3.0114812723508377,
-      "grad_norm": 22.619985580444336,
-      "learning_rate": 1.1931112365894975e-05,
-      "loss": 0.7825,
+      "grad_norm": 4.4268035888671875,
+      "learning_rate": 0.00019885187276491627,
+      "loss": 0.7344,
       "step": 16000
     },
     {
       "epoch": 3.0133634481460567,
-      "grad_norm": 12.743892669677734,
-      "learning_rate": 1.191981931112366e-05,
-      "loss": 0.4295,
+      "grad_norm": 3.6508936882019043,
+      "learning_rate": 0.00019866365518539432,
+      "loss": 0.4556,
       "step": 16010
     },
     {
       "epoch": 3.015245623941276,
-      "grad_norm": 13.726835250854492,
-      "learning_rate": 1.1908526256352344e-05,
-      "loss": 0.8777,
+      "grad_norm": 2.8904614448547363,
+      "learning_rate": 0.0001984754376058724,
+      "loss": 0.8619,
       "step": 16020
     },
     {
       "epoch": 3.0171277997364956,
-      "grad_norm": 14.803107261657715,
-      "learning_rate": 1.1897233201581028e-05,
-      "loss": 0.3934,
+      "grad_norm": 1.6528865098953247,
+      "learning_rate": 0.00019828722002635047,
+      "loss": 0.3956,
       "step": 16030
     },
     {
       "epoch": 3.0190099755317146,
-      "grad_norm": 41.224464416503906,
-      "learning_rate": 1.1885940146809712e-05,
-      "loss": 0.459,
+      "grad_norm": 6.028669834136963,
+      "learning_rate": 0.00019809900244682855,
+      "loss": 0.7254,
       "step": 16040
     },
     {
       "epoch": 3.020892151326934,
-      "grad_norm": 14.260553359985352,
-      "learning_rate": 1.1874647092038397e-05,
-      "loss": 0.3349,
+      "grad_norm": 0.40972083806991577,
+      "learning_rate": 0.0001979107848673066,
+      "loss": 0.2728,
       "step": 16050
     },
     {
       "epoch": 3.022774327122153,
-      "grad_norm": 2.4585812091827393,
-      "learning_rate": 1.1863354037267081e-05,
-      "loss": 0.8261,
+      "grad_norm": 1.0837814807891846,
+      "learning_rate": 0.00019772256728778467,
+      "loss": 0.8039,
       "step": 16060
     },
     {
       "epoch": 3.0246565029173724,
-      "grad_norm": 15.067172050476074,
-      "learning_rate": 1.1852060982495766e-05,
-      "loss": 0.4319,
+      "grad_norm": 1.60385000705719,
+      "learning_rate": 0.00019753434970826275,
+      "loss": 0.4548,
       "step": 16070
     },
     {
       "epoch": 3.026538678712592,
-      "grad_norm": 16.329282760620117,
-      "learning_rate": 1.184076792772445e-05,
-      "loss": 0.7257,
+      "grad_norm": 2.654540538787842,
+      "learning_rate": 0.00019734613212874083,
+      "loss": 0.6307,
       "step": 16080
     },
     {
       "epoch": 3.028420854507811,
-      "grad_norm": 12.787527084350586,
-      "learning_rate": 1.1829474872953134e-05,
-      "loss": 0.3546,
+      "grad_norm": 2.1555159091949463,
+      "learning_rate": 0.0001971579145492189,
+      "loss": 0.4333,
       "step": 16090
     },
     {
       "epoch": 3.0303030303030303,
-      "grad_norm": 3.467725992202759,
-      "learning_rate": 1.1818181818181819e-05,
-      "loss": 0.6743,
+      "grad_norm": 1.0611692667007446,
+      "learning_rate": 0.00019696969696969695,
+      "loss": 0.5241,
       "step": 16100
     },
     {
       "epoch": 3.0321852060982497,
-      "grad_norm": 26.06414794921875,
-      "learning_rate": 1.1806888763410503e-05,
-      "loss": 0.3207,
+      "grad_norm": 1.8001810312271118,
+      "learning_rate": 0.00019678147939017506,
+      "loss": 0.3581,
       "step": 16110
     },
     {
       "epoch": 3.0340673818934687,
-      "grad_norm": 18.53022003173828,
-      "learning_rate": 1.1795595708639187e-05,
-      "loss": 0.6154,
+      "grad_norm": 2.1027791500091553,
+      "learning_rate": 0.00019659326181065313,
+      "loss": 0.5984,
       "step": 16120
     },
     {
       "epoch": 3.035949557688688,
-      "grad_norm": 7.747043132781982,
-      "learning_rate": 1.1784302653867872e-05,
-      "loss": 0.2841,
+      "grad_norm": 1.0899429321289062,
+      "learning_rate": 0.0001964050442311312,
+      "loss": 0.373,
       "step": 16130
     },
     {
       "epoch": 3.0378317334839076,
-      "grad_norm": 8.015532493591309,
-      "learning_rate": 1.1773009599096555e-05,
-      "loss": 0.5264,
+      "grad_norm": 2.035482883453369,
+      "learning_rate": 0.00019621682665160926,
+      "loss": 0.5412,
       "step": 16140
     },
     {
       "epoch": 3.0397139092791265,
-      "grad_norm": 7.965919494628906,
-      "learning_rate": 1.176171654432524e-05,
-      "loss": 0.566,
+      "grad_norm": 3.0822765827178955,
+      "learning_rate": 0.00019602860907208734,
+      "loss": 0.4806,
       "step": 16150
     },
     {
       "epoch": 3.041596085074346,
-      "grad_norm": 0.4477425813674927,
-      "learning_rate": 1.1750423489553925e-05,
-      "loss": 0.5456,
+      "grad_norm": 0.3393348157405853,
+      "learning_rate": 0.0001958403914925654,
+      "loss": 0.556,
       "step": 16160
     },
     {
       "epoch": 3.0434782608695654,
-      "grad_norm": 6.554590702056885,
-      "learning_rate": 1.173913043478261e-05,
-      "loss": 1.0002,
+      "grad_norm": 4.573878288269043,
+      "learning_rate": 0.0001956521739130435,
+      "loss": 0.7014,
       "step": 16170
     },
     {
       "epoch": 3.0453604366647844,
-      "grad_norm": 12.476119041442871,
-      "learning_rate": 1.1727837380011292e-05,
-      "loss": 0.8468,
+      "grad_norm": 4.302673816680908,
+      "learning_rate": 0.00019546395633352154,
+      "loss": 0.668,
       "step": 16180
     },
     {
       "epoch": 3.047242612460004,
-      "grad_norm": 13.832260131835938,
-      "learning_rate": 1.1716544325239978e-05,
-      "loss": 0.7469,
+      "grad_norm": 5.19127893447876,
+      "learning_rate": 0.00019527573875399961,
+      "loss": 0.7206,
       "step": 16190
     },
     {
       "epoch": 3.049124788255223,
-      "grad_norm": 1.2430438995361328,
-      "learning_rate": 1.1705251270468663e-05,
-      "loss": 0.7722,
+      "grad_norm": 0.9039764404296875,
+      "learning_rate": 0.0001950875211744777,
+      "loss": 0.7466,
       "step": 16200
     },
     {
       "epoch": 3.0510069640504422,
-      "grad_norm": 14.875078201293945,
-      "learning_rate": 1.1693958215697347e-05,
-      "loss": 0.7945,
+      "grad_norm": 3.421377182006836,
+      "learning_rate": 0.00019489930359495577,
+      "loss": 0.9628,
       "step": 16210
     },
     {
       "epoch": 3.0528891398456617,
-      "grad_norm": 5.486425399780273,
-      "learning_rate": 1.1682665160926031e-05,
-      "loss": 0.696,
+      "grad_norm": 3.118133068084717,
+      "learning_rate": 0.00019471108601543387,
+      "loss": 0.86,
       "step": 16220
     },
     {
       "epoch": 3.0547713156408807,
-      "grad_norm": 12.678677558898926,
-      "learning_rate": 1.1671372106154714e-05,
-      "loss": 0.339,
+      "grad_norm": 3.2065300941467285,
+      "learning_rate": 0.00019452286843591192,
+      "loss": 0.3279,
       "step": 16230
     },
     {
       "epoch": 3.0566534914361,
-      "grad_norm": 21.625308990478516,
-      "learning_rate": 1.16600790513834e-05,
-      "loss": 0.6191,
+      "grad_norm": 2.9220197200775146,
+      "learning_rate": 0.00019433465085639,
+      "loss": 0.7055,
       "step": 16240
     },
     {
       "epoch": 3.0585356672313195,
-      "grad_norm": 41.28392791748047,
-      "learning_rate": 1.1648785996612084e-05,
-      "loss": 0.6735,
+      "grad_norm": 7.088646411895752,
+      "learning_rate": 0.00019414643327686807,
+      "loss": 0.7472,
       "step": 16250
     },
     {
       "epoch": 3.0604178430265385,
-      "grad_norm": 2.554927349090576,
-      "learning_rate": 1.1637492941840769e-05,
-      "loss": 0.7547,
+      "grad_norm": 1.323919653892517,
+      "learning_rate": 0.00019395821569734615,
+      "loss": 0.6519,
       "step": 16260
     },
     {
       "epoch": 3.062300018821758,
-      "grad_norm": 0.5858107805252075,
-      "learning_rate": 1.1626199887069451e-05,
-      "loss": 0.6705,
+      "grad_norm": 0.42989590764045715,
+      "learning_rate": 0.0001937699981178242,
+      "loss": 0.7908,
       "step": 16270
     },
     {
       "epoch": 3.0641821946169774,
-      "grad_norm": 35.934181213378906,
-      "learning_rate": 1.1614906832298136e-05,
-      "loss": 0.5326,
+      "grad_norm": 5.152632713317871,
+      "learning_rate": 0.00019358178053830227,
+      "loss": 0.5153,
       "step": 16280
     },
     {
       "epoch": 3.0660643704121964,
-      "grad_norm": 11.104010581970215,
-      "learning_rate": 1.1603613777526822e-05,
-      "loss": 0.8268,
+      "grad_norm": 4.753070831298828,
+      "learning_rate": 0.00019339356295878035,
+      "loss": 0.8615,
       "step": 16290
     },
     {
       "epoch": 3.067946546207416,
-      "grad_norm": 6.697833061218262,
-      "learning_rate": 1.1592320722755506e-05,
-      "loss": 0.6316,
+      "grad_norm": 2.295179605484009,
+      "learning_rate": 0.00019320534537925843,
+      "loss": 0.6854,
       "step": 16300
     },
     {
       "epoch": 3.0698287220026352,
-      "grad_norm": 22.01197052001953,
-      "learning_rate": 1.1581027667984189e-05,
-      "loss": 0.5534,
+      "grad_norm": 3.444908618927002,
+      "learning_rate": 0.00019301712779973648,
+      "loss": 0.5909,
       "step": 16310
     },
     {
       "epoch": 3.0717108977978542,
-      "grad_norm": 5.453395843505859,
-      "learning_rate": 1.1569734613212873e-05,
-      "loss": 0.7422,
+      "grad_norm": 1.3395254611968994,
+      "learning_rate": 0.00019282891022021455,
+      "loss": 0.7558,
       "step": 16320
     },
     {
       "epoch": 3.0735930735930737,
-      "grad_norm": 11.562372207641602,
-      "learning_rate": 1.155844155844156e-05,
-      "loss": 0.9676,
+      "grad_norm": 2.442232131958008,
+      "learning_rate": 0.00019264069264069266,
+      "loss": 0.9028,
       "step": 16330
     },
     {
       "epoch": 3.075475249388293,
-      "grad_norm": 12.74392032623291,
-      "learning_rate": 1.1547148503670244e-05,
-      "loss": 0.6064,
+      "grad_norm": 5.411843776702881,
+      "learning_rate": 0.00019245247506117073,
+      "loss": 0.6706,
       "step": 16340
     },
     {
       "epoch": 3.077357425183512,
-      "grad_norm": 20.787565231323242,
-      "learning_rate": 1.1535855448898928e-05,
-      "loss": 0.4245,
+      "grad_norm": 4.872277736663818,
+      "learning_rate": 0.0001922642574816488,
+      "loss": 0.5234,
       "step": 16350
     },
     {
       "epoch": 3.0792396009787315,
-      "grad_norm": 27.309429168701172,
-      "learning_rate": 1.1524562394127611e-05,
-      "loss": 0.7836,
+      "grad_norm": 4.058775424957275,
+      "learning_rate": 0.00019207603990212686,
+      "loss": 0.7612,
       "step": 16360
     },
     {
       "epoch": 3.0811217767739505,
-      "grad_norm": 4.994350433349609,
-      "learning_rate": 1.1513269339356295e-05,
-      "loss": 0.7593,
+      "grad_norm": 1.6236318349838257,
+      "learning_rate": 0.00019188782232260494,
+      "loss": 0.7788,
       "step": 16370
     },
     {
       "epoch": 3.08300395256917,
-      "grad_norm": 21.30853271484375,
-      "learning_rate": 1.1501976284584981e-05,
-      "loss": 0.6651,
+      "grad_norm": 2.5879671573638916,
+      "learning_rate": 0.000191699604743083,
+      "loss": 0.7316,
       "step": 16380
     },
     {
       "epoch": 3.0848861283643894,
-      "grad_norm": 36.19512939453125,
-      "learning_rate": 1.1490683229813666e-05,
-      "loss": 0.6865,
+      "grad_norm": 3.530181407928467,
+      "learning_rate": 0.0001915113871635611,
+      "loss": 0.5907,
       "step": 16390
     },
     {
       "epoch": 3.0867683041596083,
-      "grad_norm": 0.0781698077917099,
-      "learning_rate": 1.1479390175042348e-05,
-      "loss": 0.7204,
+      "grad_norm": 0.017271628603339195,
+      "learning_rate": 0.00019132316958403914,
+      "loss": 0.7044,
       "step": 16400
     },
     {
       "epoch": 3.0886504799548278,
-      "grad_norm": 6.799276351928711,
-      "learning_rate": 1.1468097120271033e-05,
-      "loss": 0.606,
+      "grad_norm": 1.9577358961105347,
+      "learning_rate": 0.00019113495200451721,
+      "loss": 0.6627,
       "step": 16410
     },
     {
       "epoch": 3.090532655750047,
-      "grad_norm": 3.7640774250030518,
-      "learning_rate": 1.1456804065499717e-05,
-      "loss": 0.6198,
+      "grad_norm": 1.9960639476776123,
+      "learning_rate": 0.0001909467344249953,
+      "loss": 0.6884,
       "step": 16420
     },
     {
       "epoch": 3.092414831545266,
-      "grad_norm": 16.729280471801758,
-      "learning_rate": 1.1445511010728403e-05,
-      "loss": 0.6511,
+      "grad_norm": 4.5003886222839355,
+      "learning_rate": 0.00019075851684547337,
+      "loss": 0.5766,
       "step": 16430
     },
     {
       "epoch": 3.0942970073404856,
-      "grad_norm": 24.438823699951172,
-      "learning_rate": 1.1434217955957086e-05,
-      "loss": 0.4929,
+      "grad_norm": 3.9035720825195312,
+      "learning_rate": 0.00019057029926595144,
+      "loss": 0.515,
       "step": 16440
     },
     {
       "epoch": 3.096179183135705,
-      "grad_norm": 5.958948612213135,
-      "learning_rate": 1.142292490118577e-05,
-      "loss": 0.5966,
+      "grad_norm": 1.8054392337799072,
+      "learning_rate": 0.00019038208168642952,
+      "loss": 0.6085,
       "step": 16450
     },
     {
       "epoch": 3.098061358930924,
-      "grad_norm": 1.2739803791046143,
-      "learning_rate": 1.1411631846414455e-05,
-      "loss": 0.9081,
+      "grad_norm": 3.321089267730713,
+      "learning_rate": 0.0001901938641069076,
+      "loss": 0.8556,
       "step": 16460
     },
     {
       "epoch": 3.0999435347261435,
-      "grad_norm": 0.8364112377166748,
-      "learning_rate": 1.140033879164314e-05,
-      "loss": 0.4441,
+      "grad_norm": 1.0997257232666016,
+      "learning_rate": 0.00019000564652738567,
+      "loss": 0.5064,
       "step": 16470
     },
     {
       "epoch": 3.1018257105213625,
-      "grad_norm": 11.922718048095703,
-      "learning_rate": 1.1389045736871825e-05,
-      "loss": 0.669,
+      "grad_norm": 1.8963079452514648,
+      "learning_rate": 0.00018981742894786375,
+      "loss": 0.7537,
       "step": 16480
     },
     {
       "epoch": 3.103707886316582,
-      "grad_norm": 12.48450756072998,
-      "learning_rate": 1.1377752682100508e-05,
-      "loss": 0.4577,
+      "grad_norm": 5.995891571044922,
+      "learning_rate": 0.0001896292113683418,
+      "loss": 0.571,
       "step": 16490
     },
     {
       "epoch": 3.1055900621118013,
-      "grad_norm": 22.12200164794922,
-      "learning_rate": 1.1366459627329192e-05,
-      "loss": 0.4632,
+      "grad_norm": 1.4215861558914185,
+      "learning_rate": 0.00018944099378881987,
+      "loss": 0.4311,
       "step": 16500
     },
     {
       "epoch": 3.1074722379070203,
-      "grad_norm": 21.450420379638672,
-      "learning_rate": 1.1355166572557877e-05,
-      "loss": 0.798,
+      "grad_norm": 2.2452361583709717,
+      "learning_rate": 0.00018925277620929795,
+      "loss": 0.6251,
       "step": 16510
     },
     {
       "epoch": 3.1093544137022397,
-      "grad_norm": 0.42957913875579834,
-      "learning_rate": 1.1343873517786563e-05,
-      "loss": 0.5381,
+      "grad_norm": 0.5021165609359741,
+      "learning_rate": 0.00018906455862977603,
+      "loss": 0.5959,
       "step": 16520
     },
     {
       "epoch": 3.111236589497459,
-      "grad_norm": 5.7693281173706055,
-      "learning_rate": 1.1332580463015245e-05,
-      "loss": 0.5663,
+      "grad_norm": 3.0534772872924805,
+      "learning_rate": 0.00018887634105025408,
+      "loss": 0.6973,
       "step": 16530
     },
     {
       "epoch": 3.113118765292678,
-      "grad_norm": 25.847362518310547,
-      "learning_rate": 1.132128740824393e-05,
-      "loss": 0.5048,
+      "grad_norm": 3.5288970470428467,
+      "learning_rate": 0.00018868812347073215,
+      "loss": 0.5277,
       "step": 16540
     },
     {
       "epoch": 3.1150009410878976,
-      "grad_norm": 3.086296319961548,
-      "learning_rate": 1.1309994353472614e-05,
-      "loss": 0.5231,
+      "grad_norm": 0.23889213800430298,
+      "learning_rate": 0.00018849990589121026,
+      "loss": 0.5217,
       "step": 16550
     },
     {
       "epoch": 3.116883116883117,
-      "grad_norm": 6.833352088928223,
-      "learning_rate": 1.12987012987013e-05,
-      "loss": 0.6046,
+      "grad_norm": 3.701779365539551,
+      "learning_rate": 0.00018831168831168833,
+      "loss": 0.652,
       "step": 16560
     },
     {
       "epoch": 3.118765292678336,
-      "grad_norm": 11.189995765686035,
-      "learning_rate": 1.1287408243929985e-05,
-      "loss": 0.4659,
+      "grad_norm": 2.4766476154327393,
+      "learning_rate": 0.0001881234707321664,
+      "loss": 0.4487,
       "step": 16570
     },
     {
       "epoch": 3.1206474684735555,
-      "grad_norm": 17.51889419555664,
-      "learning_rate": 1.1276115189158667e-05,
-      "loss": 0.7338,
+      "grad_norm": 4.216089725494385,
+      "learning_rate": 0.00018793525315264446,
+      "loss": 0.6731,
       "step": 16580
     },
     {
       "epoch": 3.122529644268775,
-      "grad_norm": 18.669206619262695,
-      "learning_rate": 1.1264822134387352e-05,
-      "loss": 0.3663,
+      "grad_norm": 1.8510828018188477,
+      "learning_rate": 0.00018774703557312254,
+      "loss": 0.4693,
       "step": 16590
     },
     {
       "epoch": 3.124411820063994,
-      "grad_norm": 4.161768436431885,
-      "learning_rate": 1.1253529079616036e-05,
-      "loss": 0.7614,
+      "grad_norm": 3.4371228218078613,
+      "learning_rate": 0.0001875588179936006,
+      "loss": 0.7364,
       "step": 16600
     },
     {
       "epoch": 3.1262939958592133,
-      "grad_norm": 17.522029876708984,
-      "learning_rate": 1.1242236024844722e-05,
-      "loss": 0.7639,
+      "grad_norm": 4.299537181854248,
+      "learning_rate": 0.0001873706004140787,
+      "loss": 0.6797,
       "step": 16610
     },
     {
       "epoch": 3.1281761716544327,
-      "grad_norm": 20.829774856567383,
-      "learning_rate": 1.1230942970073405e-05,
-      "loss": 0.5254,
+      "grad_norm": 2.3644449710845947,
+      "learning_rate": 0.00018718238283455674,
+      "loss": 0.4436,
       "step": 16620
     },
     {
       "epoch": 3.1300583474496517,
-      "grad_norm": 12.240787506103516,
-      "learning_rate": 1.121964991530209e-05,
-      "loss": 0.6138,
+      "grad_norm": 1.7090197801589966,
+      "learning_rate": 0.00018699416525503481,
+      "loss": 0.6329,
       "step": 16630
     },
     {
       "epoch": 3.131940523244871,
-      "grad_norm": 18.72231101989746,
-      "learning_rate": 1.1208356860530774e-05,
-      "loss": 0.8026,
+      "grad_norm": 5.4652791023254395,
+      "learning_rate": 0.0001868059476755129,
+      "loss": 0.7495,
       "step": 16640
     },
     {
       "epoch": 3.13382269904009,
-      "grad_norm": 8.982686042785645,
-      "learning_rate": 1.1197063805759458e-05,
-      "loss": 0.5638,
+      "grad_norm": 2.184948444366455,
+      "learning_rate": 0.00018661773009599097,
+      "loss": 0.6987,
       "step": 16650
     },
     {
       "epoch": 3.1357048748353096,
-      "grad_norm": 0.6803452968597412,
-      "learning_rate": 1.1185770750988142e-05,
-      "loss": 0.6837,
+      "grad_norm": 0.10448634624481201,
+      "learning_rate": 0.00018642951251646904,
+      "loss": 0.7188,
       "step": 16660
     },
     {
       "epoch": 3.137587050630529,
-      "grad_norm": 2.5247535705566406,
-      "learning_rate": 1.1174477696216827e-05,
-      "loss": 0.6615,
+      "grad_norm": 1.1686161756515503,
+      "learning_rate": 0.00018624129493694712,
+      "loss": 0.5243,
       "step": 16670
     },
     {
       "epoch": 3.139469226425748,
-      "grad_norm": 7.049471378326416,
-      "learning_rate": 1.1163184641445511e-05,
-      "loss": 0.5665,
+      "grad_norm": 1.7935549020767212,
+      "learning_rate": 0.0001860530773574252,
+      "loss": 0.5573,
       "step": 16680
     },
     {
       "epoch": 3.1413514022209674,
-      "grad_norm": 22.32665252685547,
-      "learning_rate": 1.1151891586674196e-05,
-      "loss": 0.878,
+      "grad_norm": 2.2225093841552734,
+      "learning_rate": 0.00018586485977790327,
+      "loss": 0.788,
       "step": 16690
     },
     {
       "epoch": 3.143233578016187,
-      "grad_norm": 18.906843185424805,
-      "learning_rate": 1.1140598531902882e-05,
-      "loss": 0.6834,
+      "grad_norm": 4.060264587402344,
+      "learning_rate": 0.00018567664219838135,
+      "loss": 0.5902,
       "step": 16700
     },
     {
       "epoch": 3.145115753811406,
-      "grad_norm": 0.7685168385505676,
-      "learning_rate": 1.1129305477131564e-05,
-      "loss": 0.6544,
+      "grad_norm": 0.9784095287322998,
+      "learning_rate": 0.0001854884246188594,
+      "loss": 0.7869,
       "step": 16710
     },
     {
       "epoch": 3.1469979296066253,
-      "grad_norm": 0.6841329336166382,
-      "learning_rate": 1.1118012422360249e-05,
-      "loss": 0.5769,
+      "grad_norm": 0.18420638144016266,
+      "learning_rate": 0.00018530020703933747,
+      "loss": 0.638,
       "step": 16720
     },
     {
       "epoch": 3.1488801054018447,
-      "grad_norm": 9.697307586669922,
-      "learning_rate": 1.1106719367588933e-05,
-      "loss": 0.5116,
+      "grad_norm": 2.6315994262695312,
+      "learning_rate": 0.00018511198945981555,
+      "loss": 0.5666,
       "step": 16730
     },
     {
       "epoch": 3.1507622811970637,
-      "grad_norm": 16.02259635925293,
-      "learning_rate": 1.1095426312817617e-05,
-      "loss": 0.5188,
+      "grad_norm": 3.2660489082336426,
+      "learning_rate": 0.00018492377188029363,
+      "loss": 0.5412,
       "step": 16740
     },
     {
       "epoch": 3.152644456992283,
-      "grad_norm": 5.338298797607422,
-      "learning_rate": 1.1084133258046302e-05,
-      "loss": 0.8259,
+      "grad_norm": 1.4657108783721924,
+      "learning_rate": 0.00018473555430077168,
+      "loss": 0.8718,
       "step": 16750
     },
     {
       "epoch": 3.1545266327875026,
-      "grad_norm": 18.724058151245117,
-      "learning_rate": 1.1072840203274986e-05,
-      "loss": 0.6036,
+      "grad_norm": 3.5344760417938232,
+      "learning_rate": 0.00018454733672124975,
+      "loss": 0.4465,
       "step": 16760
     },
     {
       "epoch": 3.1564088085827215,
-      "grad_norm": 5.406846523284912,
-      "learning_rate": 1.106154714850367e-05,
-      "loss": 0.8246,
+      "grad_norm": 1.9240225553512573,
+      "learning_rate": 0.00018435911914172786,
+      "loss": 0.8639,
       "step": 16770
     },
     {
       "epoch": 3.158290984377941,
-      "grad_norm": 8.084793090820312,
-      "learning_rate": 1.1050254093732355e-05,
-      "loss": 0.8828,
+      "grad_norm": 3.1279828548431396,
+      "learning_rate": 0.00018417090156220593,
+      "loss": 0.8417,
       "step": 16780
     },
     {
       "epoch": 3.16017316017316,
-      "grad_norm": 24.73105812072754,
-      "learning_rate": 1.1038961038961038e-05,
-      "loss": 0.8427,
+      "grad_norm": 4.149167060852051,
+      "learning_rate": 0.00018398268398268398,
+      "loss": 0.8718,
       "step": 16790
     },
     {
       "epoch": 3.1620553359683794,
-      "grad_norm": 21.116586685180664,
-      "learning_rate": 1.1027667984189724e-05,
-      "loss": 0.6331,
+      "grad_norm": 5.092727184295654,
+      "learning_rate": 0.00018379446640316206,
+      "loss": 0.5612,
       "step": 16800
     },
     {
       "epoch": 3.163937511763599,
-      "grad_norm": 15.968790054321289,
-      "learning_rate": 1.1016374929418408e-05,
-      "loss": 0.562,
+      "grad_norm": 3.1172432899475098,
+      "learning_rate": 0.00018360624882364014,
+      "loss": 0.5667,
       "step": 16810
     },
     {
       "epoch": 3.165819687558818,
-      "grad_norm": 13.15872859954834,
-      "learning_rate": 1.1005081874647093e-05,
-      "loss": 0.4653,
+      "grad_norm": 2.793973922729492,
+      "learning_rate": 0.0001834180312441182,
+      "loss": 0.5911,
       "step": 16820
     },
     {
       "epoch": 3.1677018633540373,
-      "grad_norm": 13.242486953735352,
-      "learning_rate": 1.0993788819875777e-05,
-      "loss": 0.3991,
+      "grad_norm": 1.6922173500061035,
+      "learning_rate": 0.0001832298136645963,
+      "loss": 0.4133,
       "step": 16830
     },
     {
       "epoch": 3.1695840391492567,
-      "grad_norm": 12.819069862365723,
-      "learning_rate": 1.0982495765104461e-05,
-      "loss": 0.7846,
+      "grad_norm": 4.027005672454834,
+      "learning_rate": 0.00018304159608507434,
+      "loss": 0.7502,
       "step": 16840
     },
     {
       "epoch": 3.1714662149444757,
-      "grad_norm": 13.536721229553223,
-      "learning_rate": 1.0971202710333146e-05,
-      "loss": 0.9832,
+      "grad_norm": 2.4809811115264893,
+      "learning_rate": 0.00018285337850555241,
+      "loss": 0.8741,
       "step": 16850
     },
     {
       "epoch": 3.173348390739695,
-      "grad_norm": 2.7697112560272217,
-      "learning_rate": 1.095990965556183e-05,
-      "loss": 0.4785,
+      "grad_norm": 2.1828391551971436,
+      "learning_rate": 0.0001826651609260305,
+      "loss": 0.3988,
       "step": 16860
     },
     {
       "epoch": 3.1752305665349145,
-      "grad_norm": 5.944636821746826,
-      "learning_rate": 1.0948616600790514e-05,
-      "loss": 0.6094,
+      "grad_norm": 5.239040374755859,
+      "learning_rate": 0.00018247694334650857,
+      "loss": 0.7494,
       "step": 16870
     },
     {
       "epoch": 3.1771127423301335,
-      "grad_norm": 2.6697866916656494,
-      "learning_rate": 1.0937323546019197e-05,
-      "loss": 0.3562,
+      "grad_norm": 0.8617979288101196,
+      "learning_rate": 0.00018228872576698662,
+      "loss": 0.4216,
       "step": 16880
     },
     {
       "epoch": 3.178994918125353,
-      "grad_norm": 7.476838111877441,
-      "learning_rate": 1.0926030491247883e-05,
-      "loss": 0.8976,
+      "grad_norm": 2.9077141284942627,
+      "learning_rate": 0.00018210050818746472,
+      "loss": 1.0169,
       "step": 16890
     },
     {
       "epoch": 3.1808770939205724,
-      "grad_norm": 6.028730869293213,
-      "learning_rate": 1.0914737436476568e-05,
-      "loss": 0.4407,
+      "grad_norm": 2.365379810333252,
+      "learning_rate": 0.0001819122906079428,
+      "loss": 0.5274,
       "step": 16900
     },
     {
       "epoch": 3.1827592697157914,
-      "grad_norm": 1.0638821125030518,
-      "learning_rate": 1.0903444381705252e-05,
-      "loss": 0.6061,
+      "grad_norm": 1.3251471519470215,
+      "learning_rate": 0.00018172407302842087,
+      "loss": 0.6267,
       "step": 16910
     },
     {
       "epoch": 3.184641445511011,
-      "grad_norm": 39.25739669799805,
-      "learning_rate": 1.0892151326933935e-05,
-      "loss": 0.6129,
+      "grad_norm": 4.387365818023682,
+      "learning_rate": 0.00018153585544889892,
+      "loss": 0.612,
       "step": 16920
     },
     {
       "epoch": 3.18652362130623,
-      "grad_norm": 26.864404678344727,
-      "learning_rate": 1.0880858272162619e-05,
-      "loss": 0.6046,
+      "grad_norm": 5.2866387367248535,
+      "learning_rate": 0.000181347637869377,
+      "loss": 0.5251,
       "step": 16930
     },
     {
       "epoch": 3.1884057971014492,
-      "grad_norm": 1.348874568939209,
-      "learning_rate": 1.0869565217391305e-05,
-      "loss": 0.6278,
+      "grad_norm": 2.2605485916137695,
+      "learning_rate": 0.00018115942028985507,
+      "loss": 0.5894,
       "step": 16940
     },
     {
       "epoch": 3.1902879728966687,
-      "grad_norm": 13.918986320495605,
-      "learning_rate": 1.085827216261999e-05,
-      "loss": 0.8359,
+      "grad_norm": 2.464541435241699,
+      "learning_rate": 0.00018097120271033315,
+      "loss": 0.6772,
       "step": 16950
     },
     {
       "epoch": 3.1921701486918876,
-      "grad_norm": 12.762542724609375,
-      "learning_rate": 1.0846979107848674e-05,
-      "loss": 0.4523,
+      "grad_norm": 0.5530261993408203,
+      "learning_rate": 0.00018078298513081123,
+      "loss": 0.4976,
       "step": 16960
     },
     {
       "epoch": 3.194052324487107,
-      "grad_norm": 0.14461807906627655,
-      "learning_rate": 1.0835686053077357e-05,
-      "loss": 0.3795,
+      "grad_norm": 0.07630673050880432,
+      "learning_rate": 0.00018059476755128928,
+      "loss": 0.3104,
       "step": 16970
     },
     {
       "epoch": 3.1959345002823265,
-      "grad_norm": 12.9683256149292,
-      "learning_rate": 1.0824392998306043e-05,
-      "loss": 0.6801,
+      "grad_norm": 1.6199790239334106,
+      "learning_rate": 0.00018040654997176735,
+      "loss": 0.719,
       "step": 16980
     },
     {
       "epoch": 3.1978166760775455,
-      "grad_norm": 18.373062133789062,
-      "learning_rate": 1.0813099943534727e-05,
-      "loss": 0.5675,
+      "grad_norm": 1.9143341779708862,
+      "learning_rate": 0.00018021833239224543,
+      "loss": 0.7699,
       "step": 16990
     },
     {
       "epoch": 3.199698851872765,
-      "grad_norm": 6.4098310470581055,
-      "learning_rate": 1.0801806888763411e-05,
-      "loss": 0.7209,
+      "grad_norm": 2.9342286586761475,
+      "learning_rate": 0.00018003011481272353,
+      "loss": 0.8464,
       "step": 17000
     },
     {
       "epoch": 3.2015810276679844,
-      "grad_norm": 1.8334890604019165,
-      "learning_rate": 1.0790513833992094e-05,
-      "loss": 0.7512,
+      "grad_norm": 2.5452444553375244,
+      "learning_rate": 0.00017984189723320158,
+      "loss": 0.7062,
       "step": 17010
     },
     {
       "epoch": 3.2034632034632033,
-      "grad_norm": 12.161665916442871,
-      "learning_rate": 1.0779220779220778e-05,
-      "loss": 0.8162,
+      "grad_norm": 2.8599061965942383,
+      "learning_rate": 0.00017965367965367966,
+      "loss": 0.7402,
       "step": 17020
     },
     {
       "epoch": 3.2053453792584228,
-      "grad_norm": 14.931073188781738,
-      "learning_rate": 1.0767927724449465e-05,
-      "loss": 0.3628,
+      "grad_norm": 3.1525862216949463,
+      "learning_rate": 0.00017946546207415774,
+      "loss": 0.4111,
       "step": 17030
     },
     {
       "epoch": 3.207227555053642,
-      "grad_norm": 31.593656539916992,
-      "learning_rate": 1.0756634669678149e-05,
-      "loss": 0.8832,
+      "grad_norm": 5.667859077453613,
+      "learning_rate": 0.0001792772444946358,
+      "loss": 0.8955,
       "step": 17040
     },
     {
       "epoch": 3.209109730848861,
-      "grad_norm": 15.769742965698242,
-      "learning_rate": 1.0745341614906832e-05,
-      "loss": 0.5008,
+      "grad_norm": 4.671546936035156,
+      "learning_rate": 0.00017908902691511386,
+      "loss": 0.5498,
       "step": 17050
     },
     {
       "epoch": 3.2109919066440806,
-      "grad_norm": 14.6060152053833,
-      "learning_rate": 1.0734048560135516e-05,
-      "loss": 0.6849,
+      "grad_norm": 10.911493301391602,
+      "learning_rate": 0.00017890080933559194,
+      "loss": 0.7891,
       "step": 17060
     },
     {
       "epoch": 3.2128740824393,
-      "grad_norm": 9.041996002197266,
-      "learning_rate": 1.0722755505364202e-05,
-      "loss": 0.8141,
+      "grad_norm": 2.427689790725708,
+      "learning_rate": 0.00017871259175607001,
+      "loss": 0.9593,
       "step": 17070
     },
     {
       "epoch": 3.214756258234519,
-      "grad_norm": 17.594465255737305,
-      "learning_rate": 1.0711462450592886e-05,
-      "loss": 0.5805,
+      "grad_norm": 3.4500486850738525,
+      "learning_rate": 0.0001785243741765481,
+      "loss": 0.5289,
       "step": 17080
     },
     {
       "epoch": 3.2166384340297385,
-      "grad_norm": 26.354915618896484,
-      "learning_rate": 1.070016939582157e-05,
-      "loss": 0.6139,
+      "grad_norm": 5.073642730712891,
+      "learning_rate": 0.00017833615659702617,
+      "loss": 0.7019,
       "step": 17090
     },
     {
       "epoch": 3.2185206098249575,
-      "grad_norm": 20.24001693725586,
-      "learning_rate": 1.0688876341050254e-05,
-      "loss": 0.726,
+      "grad_norm": 4.536356449127197,
+      "learning_rate": 0.00017814793901750422,
+      "loss": 0.6364,
       "step": 17100
     },
     {
       "epoch": 3.220402785620177,
-      "grad_norm": 9.388453483581543,
-      "learning_rate": 1.0677583286278938e-05,
-      "loss": 0.837,
+      "grad_norm": 3.2975924015045166,
+      "learning_rate": 0.00017795972143798232,
+      "loss": 0.8828,
       "step": 17110
     },
     {
       "epoch": 3.2222849614153963,
-      "grad_norm": 33.33871841430664,
-      "learning_rate": 1.0666290231507624e-05,
-      "loss": 0.6729,
+      "grad_norm": 4.286025047302246,
+      "learning_rate": 0.0001777715038584604,
+      "loss": 0.5538,
       "step": 17120
     },
     {
       "epoch": 3.2241671372106153,
-      "grad_norm": 14.027726173400879,
-      "learning_rate": 1.0654997176736308e-05,
-      "loss": 0.6477,
+      "grad_norm": 2.76507830619812,
+      "learning_rate": 0.00017758328627893847,
+      "loss": 0.5498,
       "step": 17130
     },
     {
       "epoch": 3.2260493130058348,
-      "grad_norm": 10.19504451751709,
-      "learning_rate": 1.0643704121964991e-05,
-      "loss": 0.3972,
+      "grad_norm": 2.945713996887207,
+      "learning_rate": 0.00017739506869941652,
+      "loss": 0.5465,
       "step": 17140
     },
     {
       "epoch": 3.227931488801054,
-      "grad_norm": 20.70015525817871,
-      "learning_rate": 1.0632411067193675e-05,
-      "loss": 0.6082,
+      "grad_norm": 6.803636074066162,
+      "learning_rate": 0.0001772068511198946,
+      "loss": 0.6545,
       "step": 17150
     },
     {
       "epoch": 3.229813664596273,
-      "grad_norm": 6.828171730041504,
-      "learning_rate": 1.062111801242236e-05,
-      "loss": 0.7273,
+      "grad_norm": 4.420771598815918,
+      "learning_rate": 0.00017701863354037267,
+      "loss": 0.7304,
       "step": 17160
     },
     {
       "epoch": 3.2316958403914926,
-      "grad_norm": 13.635649681091309,
-      "learning_rate": 1.0609824957651046e-05,
-      "loss": 0.6924,
+      "grad_norm": 2.6977665424346924,
+      "learning_rate": 0.00017683041596085075,
+      "loss": 0.7973,
       "step": 17170
     },
     {
       "epoch": 3.233578016186712,
-      "grad_norm": 5.508321285247803,
-      "learning_rate": 1.059853190287973e-05,
-      "loss": 0.5075,
+      "grad_norm": 1.1652570962905884,
+      "learning_rate": 0.00017664219838132883,
+      "loss": 0.4759,
       "step": 17180
     },
     {
       "epoch": 3.235460191981931,
-      "grad_norm": 20.685861587524414,
-      "learning_rate": 1.0587238848108413e-05,
-      "loss": 0.5851,
+      "grad_norm": 1.8155357837677002,
+      "learning_rate": 0.00017645398080180688,
+      "loss": 0.7366,
       "step": 17190
     },
     {
       "epoch": 3.2373423677771505,
-      "grad_norm": 14.502724647521973,
-      "learning_rate": 1.0575945793337097e-05,
-      "loss": 0.4921,
+      "grad_norm": 2.0348408222198486,
+      "learning_rate": 0.00017626576322228495,
+      "loss": 0.5368,
       "step": 17200
     },
     {
       "epoch": 3.2392245435723694,
-      "grad_norm": 8.47658634185791,
-      "learning_rate": 1.0564652738565783e-05,
-      "loss": 0.6247,
+      "grad_norm": 4.619930744171143,
+      "learning_rate": 0.00017607754564276303,
+      "loss": 0.6692,
       "step": 17210
     },
     {
       "epoch": 3.241106719367589,
-      "grad_norm": 10.943485260009766,
-      "learning_rate": 1.0553359683794468e-05,
-      "loss": 0.8117,
+      "grad_norm": 2.546478271484375,
+      "learning_rate": 0.00017588932806324113,
+      "loss": 0.7912,
       "step": 17220
     },
     {
       "epoch": 3.2429888951628083,
-      "grad_norm": 10.894013404846191,
-      "learning_rate": 1.054206662902315e-05,
-      "loss": 0.494,
+      "grad_norm": 3.549199342727661,
+      "learning_rate": 0.00017570111048371918,
+      "loss": 0.5598,
       "step": 17230
     },
     {
       "epoch": 3.2448710709580273,
-      "grad_norm": 12.969403266906738,
-      "learning_rate": 1.0530773574251835e-05,
-      "loss": 0.4713,
+      "grad_norm": 3.683444023132324,
+      "learning_rate": 0.00017551289290419726,
+      "loss": 0.6625,
       "step": 17240
     },
     {
       "epoch": 3.2467532467532467,
-      "grad_norm": 13.456480979919434,
-      "learning_rate": 1.051948051948052e-05,
-      "loss": 0.6888,
+      "grad_norm": 2.0694761276245117,
+      "learning_rate": 0.00017532467532467534,
+      "loss": 0.7337,
       "step": 17250
     },
     {
       "epoch": 3.248635422548466,
-      "grad_norm": 12.490468978881836,
-      "learning_rate": 1.0508187464709205e-05,
-      "loss": 0.6876,
+      "grad_norm": 5.7573394775390625,
+      "learning_rate": 0.0001751364577451534,
+      "loss": 0.641,
       "step": 17260
     },
     {
       "epoch": 3.250517598343685,
-      "grad_norm": 13.233989715576172,
-      "learning_rate": 1.0496894409937888e-05,
-      "loss": 0.5344,
+      "grad_norm": 2.7505602836608887,
+      "learning_rate": 0.00017494824016563146,
+      "loss": 0.5127,
       "step": 17270
     },
     {
       "epoch": 3.2523997741389046,
-      "grad_norm": 26.581886291503906,
-      "learning_rate": 1.0485601355166572e-05,
-      "loss": 0.6548,
+      "grad_norm": 10.71851634979248,
+      "learning_rate": 0.00017476002258610954,
+      "loss": 0.6569,
       "step": 17280
     },
     {
       "epoch": 3.254281949934124,
-      "grad_norm": 5.6854777336120605,
-      "learning_rate": 1.0474308300395257e-05,
-      "loss": 0.3306,
+      "grad_norm": 2.259294033050537,
+      "learning_rate": 0.00017457180500658761,
+      "loss": 0.3766,
       "step": 17290
     },
     {
       "epoch": 3.256164125729343,
-      "grad_norm": 15.864883422851562,
-      "learning_rate": 1.0463015245623941e-05,
-      "loss": 0.3117,
+      "grad_norm": 5.869109630584717,
+      "learning_rate": 0.0001743835874270657,
+      "loss": 0.5871,
       "step": 17300
     },
     {
       "epoch": 3.2580463015245624,
-      "grad_norm": 1.5253409147262573,
-      "learning_rate": 1.0451722190852627e-05,
-      "loss": 0.6778,
+      "grad_norm": 4.830927848815918,
+      "learning_rate": 0.00017419536984754377,
+      "loss": 0.6989,
       "step": 17310
     },
     {
       "epoch": 3.259928477319782,
-      "grad_norm": 13.313026428222656,
-      "learning_rate": 1.044042913608131e-05,
-      "loss": 0.4523,
+      "grad_norm": 2.818700075149536,
+      "learning_rate": 0.00017400715226802182,
+      "loss": 0.4231,
       "step": 17320
     },
     {
       "epoch": 3.261810653115001,
-      "grad_norm": 6.4470343589782715,
-      "learning_rate": 1.0429136081309994e-05,
-      "loss": 0.4609,
+      "grad_norm": 3.7204909324645996,
+      "learning_rate": 0.00017381893468849992,
+      "loss": 0.5142,
       "step": 17330
     },
     {
       "epoch": 3.2636928289102203,
-      "grad_norm": 6.336509704589844,
-      "learning_rate": 1.0417843026538679e-05,
-      "loss": 0.398,
+      "grad_norm": 1.9712015390396118,
+      "learning_rate": 0.000173630717108978,
+      "loss": 0.6661,
       "step": 17340
     },
     {
       "epoch": 3.2655750047054397,
-      "grad_norm": 17.98333740234375,
-      "learning_rate": 1.0406549971767365e-05,
-      "loss": 0.8517,
+      "grad_norm": 4.433485984802246,
+      "learning_rate": 0.00017344249952945607,
+      "loss": 0.808,
       "step": 17350
     },
     {
       "epoch": 3.2674571805006587,
-      "grad_norm": 22.246747970581055,
-      "learning_rate": 1.0395256916996047e-05,
-      "loss": 0.5689,
+      "grad_norm": 3.4073405265808105,
+      "learning_rate": 0.00017325428194993412,
+      "loss": 0.5962,
       "step": 17360
     },
     {
       "epoch": 3.269339356295878,
-      "grad_norm": 16.68685531616211,
-      "learning_rate": 1.0383963862224732e-05,
-      "loss": 0.8071,
+      "grad_norm": 2.7334978580474854,
+      "learning_rate": 0.0001730660643704122,
+      "loss": 0.8815,
       "step": 17370
     },
     {
       "epoch": 3.271221532091097,
-      "grad_norm": 25.194631576538086,
-      "learning_rate": 1.0372670807453416e-05,
-      "loss": 0.6331,
+      "grad_norm": 2.573606014251709,
+      "learning_rate": 0.00017287784679089027,
+      "loss": 0.6079,
       "step": 17380
     },
     {
       "epoch": 3.2731037078863165,
-      "grad_norm": 10.572707176208496,
-      "learning_rate": 1.03613777526821e-05,
+      "grad_norm": 2.335155487060547,
+      "learning_rate": 0.00017268962921136835,
       "loss": 0.4869,
       "step": 17390
     },
     {
       "epoch": 3.274985883681536,
-      "grad_norm": 7.161264896392822,
-      "learning_rate": 1.0350084697910785e-05,
-      "loss": 0.4585,
+      "grad_norm": 4.357646942138672,
+      "learning_rate": 0.0001725014116318464,
+      "loss": 0.6921,
       "step": 17400
     },
     {
       "epoch": 3.276868059476755,
-      "grad_norm": 13.183558464050293,
-      "learning_rate": 1.033879164313947e-05,
-      "loss": 0.8198,
+      "grad_norm": 5.32853364944458,
+      "learning_rate": 0.00017231319405232448,
+      "loss": 0.8342,
       "step": 17410
     },
     {
       "epoch": 3.2787502352719744,
-      "grad_norm": 21.05937385559082,
-      "learning_rate": 1.0327498588368154e-05,
-      "loss": 0.4036,
+      "grad_norm": 2.327008008956909,
+      "learning_rate": 0.00017212497647280255,
+      "loss": 0.5999,
       "step": 17420
     },
     {
       "epoch": 3.280632411067194,
-      "grad_norm": 5.843810558319092,
-      "learning_rate": 1.0316205533596838e-05,
-      "loss": 0.5591,
+      "grad_norm": 2.867143154144287,
+      "learning_rate": 0.00017193675889328063,
+      "loss": 0.6467,
       "step": 17430
     },
     {
       "epoch": 3.282514586862413,
-      "grad_norm": 3.102691888809204,
-      "learning_rate": 1.0304912478825523e-05,
-      "loss": 0.511,
+      "grad_norm": 0.24477972090244293,
+      "learning_rate": 0.00017174854131375873,
+      "loss": 0.5012,
       "step": 17440
     },
     {
       "epoch": 3.2843967626576323,
-      "grad_norm": 22.08820915222168,
-      "learning_rate": 1.0293619424054207e-05,
-      "loss": 0.5755,
+      "grad_norm": 4.458093643188477,
+      "learning_rate": 0.00017156032373423678,
+      "loss": 0.5334,
       "step": 17450
     },
     {
       "epoch": 3.2862789384528517,
-      "grad_norm": 6.78153657913208,
-      "learning_rate": 1.0282326369282891e-05,
-      "loss": 0.3316,
+      "grad_norm": 2.012943744659424,
+      "learning_rate": 0.00017137210615471486,
+      "loss": 0.4542,
       "step": 17460
     },
     {
       "epoch": 3.2881611142480707,
-      "grad_norm": 24.446062088012695,
-      "learning_rate": 1.0271033314511576e-05,
-      "loss": 0.4758,
+      "grad_norm": 2.415400266647339,
+      "learning_rate": 0.00017118388857519294,
+      "loss": 0.572,
       "step": 17470
     },
     {
       "epoch": 3.29004329004329,
-      "grad_norm": 23.952836990356445,
-      "learning_rate": 1.025974025974026e-05,
-      "loss": 0.622,
+      "grad_norm": 1.1272308826446533,
+      "learning_rate": 0.000170995670995671,
+      "loss": 0.5705,
       "step": 17480
     },
     {
       "epoch": 3.291925465838509,
-      "grad_norm": 39.4639892578125,
-      "learning_rate": 1.0248447204968944e-05,
-      "loss": 0.5434,
+      "grad_norm": 5.979650020599365,
+      "learning_rate": 0.00017080745341614906,
+      "loss": 0.5985,
       "step": 17490
     },
     {
       "epoch": 3.2938076416337285,
-      "grad_norm": 17.225452423095703,
-      "learning_rate": 1.0237154150197629e-05,
-      "loss": 0.9825,
+      "grad_norm": 3.5249295234680176,
+      "learning_rate": 0.00017061923583662714,
+      "loss": 0.9201,
       "step": 17500
     },
     {
       "epoch": 3.295689817428948,
-      "grad_norm": 9.208147048950195,
-      "learning_rate": 1.0225861095426313e-05,
-      "loss": 0.6339,
+      "grad_norm": 4.946028709411621,
+      "learning_rate": 0.00017043101825710521,
+      "loss": 0.6594,
       "step": 17510
     },
     {
       "epoch": 3.2975719932241674,
-      "grad_norm": 30.452302932739258,
-      "learning_rate": 1.0214568040654998e-05,
-      "loss": 0.6128,
+      "grad_norm": 5.57255744934082,
+      "learning_rate": 0.0001702428006775833,
+      "loss": 0.6134,
       "step": 17520
     },
     {
       "epoch": 3.2994541690193864,
-      "grad_norm": 32.60814666748047,
-      "learning_rate": 1.020327498588368e-05,
-      "loss": 0.5782,
+      "grad_norm": 3.510176658630371,
+      "learning_rate": 0.00017005458309806134,
+      "loss": 0.6441,
       "step": 17530
     },
     {
       "epoch": 3.301336344814606,
-      "grad_norm": 16.545825958251953,
-      "learning_rate": 1.0191981931112366e-05,
-      "loss": 0.5999,
+      "grad_norm": 4.090064525604248,
+      "learning_rate": 0.00016986636551853942,
+      "loss": 0.5774,
       "step": 17540
     },
     {
       "epoch": 3.303218520609825,
-      "grad_norm": 15.124717712402344,
-      "learning_rate": 1.018068887634105e-05,
-      "loss": 0.4035,
+      "grad_norm": 1.2690385580062866,
+      "learning_rate": 0.00016967814793901752,
+      "loss": 0.4282,
       "step": 17550
     },
     {
       "epoch": 3.3051006964050442,
-      "grad_norm": 6.470951080322266,
-      "learning_rate": 1.0169395821569735e-05,
-      "loss": 0.7927,
+      "grad_norm": 1.3262484073638916,
+      "learning_rate": 0.0001694899303594956,
+      "loss": 0.8544,
       "step": 17560
     },
     {
       "epoch": 3.3069828722002637,
-      "grad_norm": 16.971155166625977,
-      "learning_rate": 1.015810276679842e-05,
-      "loss": 0.6804,
+      "grad_norm": 3.6288297176361084,
+      "learning_rate": 0.00016930171277997367,
+      "loss": 0.759,
       "step": 17570
     },
     {
       "epoch": 3.3088650479954826,
-      "grad_norm": 29.655359268188477,
-      "learning_rate": 1.0146809712027104e-05,
-      "loss": 0.7321,
+      "grad_norm": 5.26113748550415,
+      "learning_rate": 0.00016911349520045172,
+      "loss": 0.6652,
       "step": 17580
     },
     {
       "epoch": 3.310747223790702,
-      "grad_norm": 28.08059310913086,
-      "learning_rate": 1.0135516657255788e-05,
-      "loss": 0.7635,
+      "grad_norm": 4.848659992218018,
+      "learning_rate": 0.0001689252776209298,
+      "loss": 0.7512,
       "step": 17590
     },
     {
       "epoch": 3.3126293995859215,
-      "grad_norm": 0.4805198013782501,
-      "learning_rate": 1.0124223602484473e-05,
-      "loss": 0.4791,
+      "grad_norm": 0.0900026485323906,
+      "learning_rate": 0.00016873706004140787,
+      "loss": 0.3456,
       "step": 17600
     },
     {
       "epoch": 3.3145115753811405,
-      "grad_norm": 30.67386245727539,
-      "learning_rate": 1.0112930547713157e-05,
-      "loss": 0.6166,
+      "grad_norm": 4.362743377685547,
+      "learning_rate": 0.00016854884246188595,
+      "loss": 0.7094,
       "step": 17610
     },
     {
       "epoch": 3.31639375117636,
-      "grad_norm": 30.00206756591797,
-      "learning_rate": 1.010163749294184e-05,
-      "loss": 0.9969,
+      "grad_norm": 2.96457576751709,
+      "learning_rate": 0.000168360624882364,
+      "loss": 0.987,
       "step": 17620
     },
     {
       "epoch": 3.3182759269715794,
-      "grad_norm": 11.875704765319824,
-      "learning_rate": 1.0090344438170526e-05,
-      "loss": 0.4861,
+      "grad_norm": 3.4136481285095215,
+      "learning_rate": 0.00016817240730284208,
+      "loss": 0.5137,
       "step": 17630
     },
     {
       "epoch": 3.3201581027667983,
-      "grad_norm": 59.30400848388672,
-      "learning_rate": 1.007905138339921e-05,
-      "loss": 0.6634,
+      "grad_norm": 5.220181941986084,
+      "learning_rate": 0.00016798418972332015,
+      "loss": 0.8406,
       "step": 17640
     },
     {
       "epoch": 3.322040278562018,
-      "grad_norm": 7.319840908050537,
-      "learning_rate": 1.0067758328627895e-05,
-      "loss": 0.1843,
+      "grad_norm": 3.106947898864746,
+      "learning_rate": 0.00016779597214379823,
+      "loss": 0.2564,
       "step": 17650
     },
     {
       "epoch": 3.3239224543572368,
-      "grad_norm": 1.659640908241272,
-      "learning_rate": 1.0056465273856577e-05,
-      "loss": 0.8501,
+      "grad_norm": 0.1169840395450592,
+      "learning_rate": 0.0001676077545642763,
+      "loss": 0.7908,
       "step": 17660
     },
     {
       "epoch": 3.325804630152456,
-      "grad_norm": 15.587262153625488,
-      "learning_rate": 1.0045172219085262e-05,
-      "loss": 0.5065,
+      "grad_norm": 3.5396854877471924,
+      "learning_rate": 0.00016741953698475438,
+      "loss": 0.5651,
       "step": 17670
     },
     {
       "epoch": 3.3276868059476756,
-      "grad_norm": 25.570293426513672,
-      "learning_rate": 1.0033879164313948e-05,
-      "loss": 0.8362,
+      "grad_norm": 4.229373931884766,
+      "learning_rate": 0.00016723131940523246,
+      "loss": 0.6979,
       "step": 17680
     },
     {
       "epoch": 3.3295689817428946,
-      "grad_norm": 5.94968318939209,
-      "learning_rate": 1.0022586109542632e-05,
-      "loss": 0.7938,
+      "grad_norm": 2.629528045654297,
+      "learning_rate": 0.00016704310182571054,
+      "loss": 0.6957,
       "step": 17690
     },
     {
       "epoch": 3.331451157538114,
-      "grad_norm": 5.780099868774414,
-      "learning_rate": 1.0011293054771316e-05,
-      "loss": 0.6672,
+      "grad_norm": 0.5072752237319946,
+      "learning_rate": 0.0001668548842461886,
+      "loss": 0.6849,
       "step": 17700
     },
     {
       "epoch": 3.3333333333333335,
-      "grad_norm": 32.476585388183594,
-      "learning_rate": 9.999999999999999e-06,
-      "loss": 0.5259,
+      "grad_norm": 8.92314338684082,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 0.6571,
       "step": 17710
     },
     {
       "epoch": 3.3352155091285525,
-      "grad_norm": 6.785778999328613,
-      "learning_rate": 9.988706945228685e-06,
-      "loss": 0.719,
+      "grad_norm": 1.982568621635437,
+      "learning_rate": 0.00016647844908714474,
+      "loss": 0.7167,
       "step": 17720
     },
     {
       "epoch": 3.337097684923772,
-      "grad_norm": 16.405250549316406,
-      "learning_rate": 9.97741389045737e-06,
-      "loss": 0.8035,
+      "grad_norm": 5.065396308898926,
+      "learning_rate": 0.00016629023150762281,
+      "loss": 0.9074,
       "step": 17730
     },
     {
       "epoch": 3.3389798607189913,
-      "grad_norm": 8.930052757263184,
-      "learning_rate": 9.966120835686054e-06,
-      "loss": 0.7513,
+      "grad_norm": 2.799494504928589,
+      "learning_rate": 0.0001661020139281009,
+      "loss": 0.9093,
       "step": 17740
     },
     {
       "epoch": 3.3408620365142103,
-      "grad_norm": 0.5434146523475647,
-      "learning_rate": 9.954827780914737e-06,
-      "loss": 0.5323,
+      "grad_norm": 0.08490809053182602,
+      "learning_rate": 0.00016591379634857894,
+      "loss": 0.5279,
       "step": 17750
     },
     {
       "epoch": 3.3427442123094298,
-      "grad_norm": 12.213115692138672,
-      "learning_rate": 9.943534726143421e-06,
-      "loss": 0.5251,
+      "grad_norm": 5.632839679718018,
+      "learning_rate": 0.00016572557876905702,
+      "loss": 0.729,
       "step": 17760
     },
     {
       "epoch": 3.3446263881046487,
-      "grad_norm": 18.9610595703125,
-      "learning_rate": 9.932241671372107e-06,
-      "loss": 0.833,
+      "grad_norm": 3.7609736919403076,
+      "learning_rate": 0.00016553736118953512,
+      "loss": 0.8124,
       "step": 17770
     },
     {
       "epoch": 3.346508563899868,
-      "grad_norm": 8.55979061126709,
-      "learning_rate": 9.920948616600791e-06,
-      "loss": 0.9362,
+      "grad_norm": 2.617806911468506,
+      "learning_rate": 0.0001653491436100132,
+      "loss": 1.064,
       "step": 17780
     },
     {
       "epoch": 3.3483907396950876,
-      "grad_norm": 1.9761569499969482,
-      "learning_rate": 9.909655561829474e-06,
-      "loss": 0.5021,
+      "grad_norm": 1.1742140054702759,
+      "learning_rate": 0.00016516092603049125,
+      "loss": 0.4695,
       "step": 17790
     },
     {
       "epoch": 3.350272915490307,
-      "grad_norm": 16.7060546875,
-      "learning_rate": 9.898362507058159e-06,
-      "loss": 0.5859,
+      "grad_norm": 2.0559253692626953,
+      "learning_rate": 0.00016497270845096932,
+      "loss": 0.4638,
       "step": 17800
     },
     {
       "epoch": 3.352155091285526,
-      "grad_norm": 16.749752044677734,
-      "learning_rate": 9.887069452286843e-06,
-      "loss": 0.4982,
+      "grad_norm": 2.232964277267456,
+      "learning_rate": 0.0001647844908714474,
+      "loss": 0.4741,
       "step": 17810
     },
     {
       "epoch": 3.3540372670807455,
-      "grad_norm": 3.789137601852417,
-      "learning_rate": 9.875776397515529e-06,
-      "loss": 0.4616,
+      "grad_norm": 2.2919833660125732,
+      "learning_rate": 0.00016459627329192547,
+      "loss": 0.5339,
       "step": 17820
     },
     {
       "epoch": 3.3559194428759644,
-      "grad_norm": 13.295493125915527,
-      "learning_rate": 9.864483342744213e-06,
-      "loss": 0.6548,
+      "grad_norm": 2.7423524856567383,
+      "learning_rate": 0.00016440805571240355,
+      "loss": 0.4588,
       "step": 17830
     },
     {
       "epoch": 3.357801618671184,
-      "grad_norm": 25.58221435546875,
-      "learning_rate": 9.853190287972896e-06,
-      "loss": 0.7388,
+      "grad_norm": 3.7243103981018066,
+      "learning_rate": 0.0001642198381328816,
+      "loss": 0.7305,
       "step": 17840
     },
     {
       "epoch": 3.3596837944664033,
-      "grad_norm": 29.051992416381836,
-      "learning_rate": 9.84189723320158e-06,
-      "loss": 0.5679,
+      "grad_norm": 6.474857807159424,
+      "learning_rate": 0.00016403162055335968,
+      "loss": 0.5081,
       "step": 17850
     },
     {
       "epoch": 3.3615659702616223,
-      "grad_norm": 11.573369026184082,
-      "learning_rate": 9.830604178430267e-06,
-      "loss": 0.7047,
+      "grad_norm": 3.899993419647217,
+      "learning_rate": 0.00016384340297383775,
+      "loss": 0.7415,
       "step": 17860
     },
     {
       "epoch": 3.3634481460568417,
-      "grad_norm": 16.204544067382812,
-      "learning_rate": 9.819311123658951e-06,
-      "loss": 0.5335,
+      "grad_norm": 4.013248920440674,
+      "learning_rate": 0.00016365518539431583,
+      "loss": 0.5383,
       "step": 17870
     },
     {
       "epoch": 3.365330321852061,
-      "grad_norm": 18.992883682250977,
-      "learning_rate": 9.808018068887634e-06,
-      "loss": 0.5,
+      "grad_norm": 3.6923396587371826,
+      "learning_rate": 0.0001634669678147939,
+      "loss": 0.598,
       "step": 17880
     },
     {
       "epoch": 3.36721249764728,
-      "grad_norm": 17.08180809020996,
-      "learning_rate": 9.796725014116318e-06,
-      "loss": 0.6355,
+      "grad_norm": 1.4304816722869873,
+      "learning_rate": 0.00016327875023527198,
+      "loss": 0.6347,
       "step": 17890
     },
     {
       "epoch": 3.3690946734424996,
-      "grad_norm": 12.415989875793457,
-      "learning_rate": 9.785431959345002e-06,
-      "loss": 0.6227,
+      "grad_norm": 5.5415425300598145,
+      "learning_rate": 0.00016309053265575006,
+      "loss": 0.6624,
       "step": 17900
     },
     {
       "epoch": 3.370976849237719,
-      "grad_norm": 23.010984420776367,
-      "learning_rate": 9.774138904573688e-06,
-      "loss": 0.6095,
+      "grad_norm": 1.9383195638656616,
+      "learning_rate": 0.00016290231507622814,
+      "loss": 0.5608,
       "step": 17910
     },
     {
       "epoch": 3.372859025032938,
-      "grad_norm": 20.306901931762695,
-      "learning_rate": 9.762845849802373e-06,
-      "loss": 0.7475,
+      "grad_norm": 5.054727077484131,
+      "learning_rate": 0.0001627140974967062,
+      "loss": 0.8852,
       "step": 17920
     },
     {
       "epoch": 3.3747412008281574,
-      "grad_norm": 7.907451629638672,
-      "learning_rate": 9.751552795031056e-06,
-      "loss": 0.7657,
+      "grad_norm": 1.5502482652664185,
+      "learning_rate": 0.00016252587991718426,
+      "loss": 0.7796,
       "step": 17930
     },
     {
       "epoch": 3.3766233766233764,
-      "grad_norm": 10.358199119567871,
-      "learning_rate": 9.74025974025974e-06,
-      "loss": 0.8862,
+      "grad_norm": 4.015044689178467,
+      "learning_rate": 0.00016233766233766234,
+      "loss": 0.9499,
       "step": 17940
     },
     {
       "epoch": 3.378505552418596,
-      "grad_norm": 7.511529445648193,
-      "learning_rate": 9.728966685488426e-06,
-      "loss": 0.4816,
+      "grad_norm": 2.692499876022339,
+      "learning_rate": 0.00016214944475814041,
+      "loss": 0.5061,
       "step": 17950
     },
     {
       "epoch": 3.3803877282138153,
-      "grad_norm": 11.36762523651123,
-      "learning_rate": 9.71767363071711e-06,
-      "loss": 0.6414,
+      "grad_norm": 3.55739164352417,
+      "learning_rate": 0.0001619612271786185,
+      "loss": 0.7057,
       "step": 17960
     },
     {
       "epoch": 3.3822699040090343,
-      "grad_norm": 4.441726207733154,
-      "learning_rate": 9.706380575945793e-06,
-      "loss": 0.3488,
+      "grad_norm": 0.6521520614624023,
+      "learning_rate": 0.00016177300959909654,
+      "loss": 0.4082,
       "step": 17970
     },
     {
       "epoch": 3.3841520798042537,
-      "grad_norm": 19.867027282714844,
-      "learning_rate": 9.695087521174477e-06,
-      "loss": 0.7145,
+      "grad_norm": 3.7358646392822266,
+      "learning_rate": 0.00016158479201957462,
+      "loss": 0.741,
       "step": 17980
     },
     {
       "epoch": 3.386034255599473,
-      "grad_norm": 10.9057035446167,
-      "learning_rate": 9.683794466403162e-06,
-      "loss": 0.6342,
+      "grad_norm": 1.7314327955245972,
+      "learning_rate": 0.00016139657444005272,
+      "loss": 0.6929,
       "step": 17990
     },
     {
       "epoch": 3.387916431394692,
-      "grad_norm": 8.741947174072266,
-      "learning_rate": 9.672501411631848e-06,
-      "loss": 0.5844,
+      "grad_norm": 3.96738862991333,
+      "learning_rate": 0.0001612083568605308,
+      "loss": 0.5361,
       "step": 18000
     },
     {
       "epoch": 3.3897986071899116,
-      "grad_norm": 25.616764068603516,
-      "learning_rate": 9.66120835686053e-06,
-      "loss": 0.6007,
+      "grad_norm": 4.310543060302734,
+      "learning_rate": 0.00016102013928100885,
+      "loss": 0.5798,
       "step": 18010
     },
     {
       "epoch": 3.391680782985131,
-      "grad_norm": 8.889225959777832,
-      "learning_rate": 9.649915302089215e-06,
-      "loss": 0.4933,
+      "grad_norm": 1.0986257791519165,
+      "learning_rate": 0.00016083192170148692,
+      "loss": 0.4943,
       "step": 18020
     },
     {
       "epoch": 3.39356295878035,
-      "grad_norm": 9.683721542358398,
-      "learning_rate": 9.6386222473179e-06,
-      "loss": 0.5452,
+      "grad_norm": 2.5314037799835205,
+      "learning_rate": 0.000160643704121965,
+      "loss": 0.576,
       "step": 18030
     },
     {
       "epoch": 3.3954451345755694,
-      "grad_norm": 5.871509075164795,
-      "learning_rate": 9.627329192546584e-06,
-      "loss": 0.5289,
+      "grad_norm": 0.23214668035507202,
+      "learning_rate": 0.00016045548654244307,
+      "loss": 0.6748,
       "step": 18040
     },
     {
       "epoch": 3.3973273103707884,
-      "grad_norm": 12.041633605957031,
-      "learning_rate": 9.61603613777527e-06,
-      "loss": 0.6335,
+      "grad_norm": 3.713710308074951,
+      "learning_rate": 0.00016026726896292115,
+      "loss": 0.6695,
       "step": 18050
     },
     {
       "epoch": 3.399209486166008,
-      "grad_norm": 17.63007926940918,
-      "learning_rate": 9.604743083003952e-06,
-      "loss": 0.8369,
+      "grad_norm": 3.0103020668029785,
+      "learning_rate": 0.0001600790513833992,
+      "loss": 0.882,
       "step": 18060
     },
     {
       "epoch": 3.4010916619612273,
-      "grad_norm": 17.726680755615234,
-      "learning_rate": 9.593450028232637e-06,
-      "loss": 0.5378,
+      "grad_norm": 4.872165679931641,
+      "learning_rate": 0.00015989083380387728,
+      "loss": 0.7086,
       "step": 18070
     },
     {
       "epoch": 3.4029738377564467,
-      "grad_norm": 7.5816330909729,
-      "learning_rate": 9.582156973461321e-06,
-      "loss": 0.5959,
+      "grad_norm": 3.5198495388031006,
+      "learning_rate": 0.00015970261622435535,
+      "loss": 0.5763,
       "step": 18080
     },
     {
       "epoch": 3.4048560135516657,
-      "grad_norm": 12.254039764404297,
-      "learning_rate": 9.570863918690007e-06,
-      "loss": 0.5402,
+      "grad_norm": 3.0030627250671387,
+      "learning_rate": 0.00015951439864483343,
+      "loss": 0.6998,
       "step": 18090
     },
     {
       "epoch": 3.406738189346885,
-      "grad_norm": 20.039281845092773,
-      "learning_rate": 9.55957086391869e-06,
-      "loss": 0.4543,
+      "grad_norm": 5.010890960693359,
+      "learning_rate": 0.0001593261810653115,
+      "loss": 0.5333,
       "step": 18100
     },
     {
       "epoch": 3.408620365142104,
-      "grad_norm": 31.039440155029297,
-      "learning_rate": 9.548277809147374e-06,
-      "loss": 0.7242,
+      "grad_norm": 2.5112898349761963,
+      "learning_rate": 0.00015913796348578958,
+      "loss": 0.6082,
       "step": 18110
     },
     {
       "epoch": 3.4105025409373235,
-      "grad_norm": 21.17696189880371,
-      "learning_rate": 9.536984754376059e-06,
-      "loss": 0.5722,
+      "grad_norm": 2.1638071537017822,
+      "learning_rate": 0.00015894974590626766,
+      "loss": 0.6394,
       "step": 18120
     },
     {
       "epoch": 3.412384716732543,
-      "grad_norm": 37.80180740356445,
-      "learning_rate": 9.525691699604743e-06,
-      "loss": 0.6266,
+      "grad_norm": 4.111144065856934,
+      "learning_rate": 0.00015876152832674574,
+      "loss": 0.6279,
       "step": 18130
     },
     {
       "epoch": 3.414266892527762,
-      "grad_norm": 0.5937038064002991,
-      "learning_rate": 9.514398644833428e-06,
-      "loss": 0.6332,
+      "grad_norm": 0.4466768503189087,
+      "learning_rate": 0.00015857331074722378,
+      "loss": 0.5873,
       "step": 18140
     },
     {
       "epoch": 3.4161490683229814,
-      "grad_norm": 18.84284782409668,
-      "learning_rate": 9.503105590062112e-06,
-      "loss": 0.5928,
+      "grad_norm": 1.232139229774475,
+      "learning_rate": 0.00015838509316770186,
+      "loss": 0.6969,
       "step": 18150
     },
     {
       "epoch": 3.418031244118201,
-      "grad_norm": 3.241466522216797,
-      "learning_rate": 9.491812535290796e-06,
-      "loss": 0.5802,
+      "grad_norm": 4.0589985847473145,
+      "learning_rate": 0.00015819687558817994,
+      "loss": 0.7058,
       "step": 18160
     },
     {
       "epoch": 3.41991341991342,
-      "grad_norm": 4.864845275878906,
-      "learning_rate": 9.48051948051948e-06,
-      "loss": 0.4499,
+      "grad_norm": 1.1862062215805054,
+      "learning_rate": 0.00015800865800865801,
+      "loss": 0.5094,
       "step": 18170
     },
     {
       "epoch": 3.4217955957086392,
-      "grad_norm": 10.21737289428711,
-      "learning_rate": 9.469226425748165e-06,
-      "loss": 0.466,
+      "grad_norm": 1.4233732223510742,
+      "learning_rate": 0.0001578204404291361,
+      "loss": 0.4951,
       "step": 18180
     },
     {
       "epoch": 3.4236777715038587,
-      "grad_norm": 21.10309410095215,
-      "learning_rate": 9.45793337097685e-06,
-      "loss": 0.4943,
+      "grad_norm": 3.4455318450927734,
+      "learning_rate": 0.00015763222284961414,
+      "loss": 0.5394,
       "step": 18190
     },
     {
       "epoch": 3.4255599472990776,
-      "grad_norm": 20.39488983154297,
-      "learning_rate": 9.446640316205534e-06,
-      "loss": 0.4178,
+      "grad_norm": 2.995675802230835,
+      "learning_rate": 0.00015744400527009222,
+      "loss": 0.4398,
       "step": 18200
     },
     {
       "epoch": 3.427442123094297,
-      "grad_norm": 10.568327903747559,
-      "learning_rate": 9.435347261434218e-06,
-      "loss": 0.5787,
+      "grad_norm": 1.3443666696548462,
+      "learning_rate": 0.00015725578769057032,
+      "loss": 0.4524,
       "step": 18210
     },
     {
       "epoch": 3.429324298889516,
-      "grad_norm": 15.859696388244629,
-      "learning_rate": 9.424054206662903e-06,
-      "loss": 0.5257,
+      "grad_norm": 4.236741542816162,
+      "learning_rate": 0.0001570675701110484,
+      "loss": 0.5184,
       "step": 18220
     },
     {
       "epoch": 3.4312064746847355,
-      "grad_norm": 3.5136094093322754,
-      "learning_rate": 9.412761151891587e-06,
-      "loss": 0.4556,
+      "grad_norm": 0.17351055145263672,
+      "learning_rate": 0.00015687935253152645,
+      "loss": 0.3975,
       "step": 18230
     },
     {
       "epoch": 3.433088650479955,
-      "grad_norm": 43.259742736816406,
-      "learning_rate": 9.401468097120271e-06,
-      "loss": 0.2424,
+      "grad_norm": 1.4714381694793701,
+      "learning_rate": 0.00015669113495200452,
+      "loss": 0.3543,
       "step": 18240
     },
     {
       "epoch": 3.434970826275174,
-      "grad_norm": 17.20220947265625,
-      "learning_rate": 9.390175042348956e-06,
-      "loss": 0.8755,
+      "grad_norm": 3.5774731636047363,
+      "learning_rate": 0.0001565029173724826,
+      "loss": 0.8003,
       "step": 18250
     },
     {
       "epoch": 3.4368530020703933,
-      "grad_norm": 35.17365646362305,
-      "learning_rate": 9.37888198757764e-06,
-      "loss": 0.5231,
+      "grad_norm": 5.654221534729004,
+      "learning_rate": 0.00015631469979296067,
+      "loss": 0.5738,
       "step": 18260
     },
     {
       "epoch": 3.438735177865613,
-      "grad_norm": 11.22391128540039,
-      "learning_rate": 9.367588932806323e-06,
-      "loss": 0.8492,
+      "grad_norm": 3.335806369781494,
+      "learning_rate": 0.00015612648221343872,
+      "loss": 0.7048,
       "step": 18270
     },
     {
       "epoch": 3.4406173536608318,
-      "grad_norm": 15.989961624145508,
-      "learning_rate": 9.356295878035009e-06,
-      "loss": 0.6255,
+      "grad_norm": 5.399444580078125,
+      "learning_rate": 0.0001559382646339168,
+      "loss": 0.5377,
       "step": 18280
     },
     {
       "epoch": 3.442499529456051,
-      "grad_norm": 10.886322975158691,
-      "learning_rate": 9.345002823263693e-06,
-      "loss": 0.5604,
+      "grad_norm": 2.897859573364258,
+      "learning_rate": 0.00015575004705439488,
+      "loss": 0.5503,
       "step": 18290
     },
     {
       "epoch": 3.4443817052512706,
-      "grad_norm": 0.8173845410346985,
-      "learning_rate": 9.333709768492378e-06,
-      "loss": 0.5142,
+      "grad_norm": 1.5217775106430054,
+      "learning_rate": 0.00015556182947487295,
+      "loss": 0.639,
       "step": 18300
     },
     {
       "epoch": 3.4462638810464896,
-      "grad_norm": 12.035292625427246,
-      "learning_rate": 9.322416713721062e-06,
-      "loss": 0.7188,
+      "grad_norm": 1.9020847082138062,
+      "learning_rate": 0.00015537361189535103,
+      "loss": 0.7124,
       "step": 18310
     },
     {
       "epoch": 3.448146056841709,
-      "grad_norm": 24.942886352539062,
-      "learning_rate": 9.311123658949746e-06,
-      "loss": 0.5322,
+      "grad_norm": 0.4594549238681793,
+      "learning_rate": 0.0001551853943158291,
+      "loss": 0.4842,
       "step": 18320
     },
     {
       "epoch": 3.4500282326369285,
-      "grad_norm": 0.916724681854248,
-      "learning_rate": 9.29983060417843e-06,
-      "loss": 0.6067,
+      "grad_norm": 0.26805880665779114,
+      "learning_rate": 0.00015499717673630718,
+      "loss": 0.5915,
       "step": 18330
     },
     {
       "epoch": 3.4519104084321475,
-      "grad_norm": 1.716382384300232,
-      "learning_rate": 9.288537549407115e-06,
-      "loss": 0.5839,
+      "grad_norm": 0.42122676968574524,
+      "learning_rate": 0.00015480895915678526,
+      "loss": 0.6011,
       "step": 18340
     },
     {
       "epoch": 3.453792584227367,
-      "grad_norm": 5.684045314788818,
-      "learning_rate": 9.2772444946358e-06,
-      "loss": 0.6052,
+      "grad_norm": 2.4079372882843018,
+      "learning_rate": 0.00015462074157726334,
+      "loss": 0.5244,
       "step": 18350
     },
     {
       "epoch": 3.4556747600225863,
-      "grad_norm": 3.6360864639282227,
-      "learning_rate": 9.265951439864482e-06,
-      "loss": 0.6305,
+      "grad_norm": 3.0662310123443604,
+      "learning_rate": 0.00015443252399774138,
+      "loss": 0.6324,
       "step": 18360
     },
     {
       "epoch": 3.4575569358178053,
-      "grad_norm": 7.340415954589844,
-      "learning_rate": 9.254658385093168e-06,
-      "loss": 0.4902,
+      "grad_norm": 2.075991153717041,
+      "learning_rate": 0.00015424430641821946,
+      "loss": 0.5104,
       "step": 18370
     },
     {
       "epoch": 3.4594391116130248,
-      "grad_norm": 12.40035343170166,
-      "learning_rate": 9.243365330321853e-06,
-      "loss": 0.6022,
+      "grad_norm": 3.1180338859558105,
+      "learning_rate": 0.00015405608883869754,
+      "loss": 0.548,
       "step": 18380
     },
     {
       "epoch": 3.4613212874082437,
-      "grad_norm": 6.647785186767578,
-      "learning_rate": 9.232072275550537e-06,
-      "loss": 0.6062,
+      "grad_norm": 2.0586273670196533,
+      "learning_rate": 0.00015386787125917561,
+      "loss": 0.6929,
       "step": 18390
     },
     {
       "epoch": 3.463203463203463,
-      "grad_norm": 16.84423828125,
-      "learning_rate": 9.22077922077922e-06,
-      "loss": 0.5901,
+      "grad_norm": 5.819500923156738,
+      "learning_rate": 0.00015367965367965366,
+      "loss": 0.6192,
       "step": 18400
     },
     {
       "epoch": 3.4650856389986826,
-      "grad_norm": 4.819279193878174,
-      "learning_rate": 9.209486166007904e-06,
-      "loss": 0.6798,
+      "grad_norm": 1.630347490310669,
+      "learning_rate": 0.00015349143610013174,
+      "loss": 0.635,
       "step": 18410
     },
     {
       "epoch": 3.4669678147939016,
-      "grad_norm": 20.76621437072754,
-      "learning_rate": 9.19819311123659e-06,
-      "loss": 0.9342,
+      "grad_norm": 4.307313919067383,
+      "learning_rate": 0.00015330321852060982,
+      "loss": 0.9541,
       "step": 18420
     },
     {
       "epoch": 3.468849990589121,
-      "grad_norm": 18.379783630371094,
-      "learning_rate": 9.186900056465275e-06,
-      "loss": 0.7618,
+      "grad_norm": 3.076420783996582,
+      "learning_rate": 0.00015311500094108792,
+      "loss": 0.713,
       "step": 18430
     },
     {
       "epoch": 3.4707321663843405,
-      "grad_norm": 22.449026107788086,
-      "learning_rate": 9.175607001693959e-06,
-      "loss": 0.7875,
+      "grad_norm": 5.041100978851318,
+      "learning_rate": 0.000152926783361566,
+      "loss": 0.7066,
       "step": 18440
     },
     {
       "epoch": 3.4726143421795594,
-      "grad_norm": 3.216062545776367,
-      "learning_rate": 9.164313946922642e-06,
-      "loss": 0.7226,
+      "grad_norm": 2.18129301071167,
+      "learning_rate": 0.00015273856578204405,
+      "loss": 0.7679,
       "step": 18450
     },
     {
       "epoch": 3.474496517974779,
-      "grad_norm": 18.014450073242188,
-      "learning_rate": 9.153020892151328e-06,
-      "loss": 0.6888,
+      "grad_norm": 3.314056396484375,
+      "learning_rate": 0.00015255034820252212,
+      "loss": 0.7436,
       "step": 18460
     },
     {
       "epoch": 3.4763786937699983,
-      "grad_norm": 33.41580581665039,
-      "learning_rate": 9.141727837380012e-06,
-      "loss": 0.4367,
+      "grad_norm": 1.7835749387741089,
+      "learning_rate": 0.0001523621306230002,
+      "loss": 0.4808,
       "step": 18470
     },
     {
       "epoch": 3.4782608695652173,
-      "grad_norm": 0.49927645921707153,
-      "learning_rate": 9.130434782608697e-06,
-      "loss": 0.683,
+      "grad_norm": 2.1161506175994873,
+      "learning_rate": 0.00015217391304347827,
+      "loss": 0.6418,
       "step": 18480
     },
     {
       "epoch": 3.4801430453604367,
-      "grad_norm": 16.403583526611328,
-      "learning_rate": 9.11914172783738e-06,
-      "loss": 0.5037,
+      "grad_norm": 2.818833351135254,
+      "learning_rate": 0.00015198569546395632,
+      "loss": 0.633,
       "step": 18490
     },
     {
       "epoch": 3.4820252211556557,
-      "grad_norm": 19.10352325439453,
-      "learning_rate": 9.107848673066064e-06,
-      "loss": 0.8453,
+      "grad_norm": 4.812803268432617,
+      "learning_rate": 0.0001517974778844344,
+      "loss": 0.791,
       "step": 18500
     },
     {
       "epoch": 3.483907396950875,
-      "grad_norm": 42.00209426879883,
-      "learning_rate": 9.09655561829475e-06,
-      "loss": 0.4188,
+      "grad_norm": 1.8344483375549316,
+      "learning_rate": 0.00015160926030491248,
+      "loss": 0.4743,
       "step": 18510
     },
     {
       "epoch": 3.4857895727460946,
-      "grad_norm": 22.176076889038086,
-      "learning_rate": 9.085262563523434e-06,
-      "loss": 0.8926,
+      "grad_norm": 1.7361068725585938,
+      "learning_rate": 0.00015142104272539055,
+      "loss": 0.8563,
       "step": 18520
     },
     {
       "epoch": 3.4876717485413136,
-      "grad_norm": 13.631558418273926,
-      "learning_rate": 9.073969508752118e-06,
-      "loss": 0.7174,
+      "grad_norm": 2.331780195236206,
+      "learning_rate": 0.00015123282514586863,
+      "loss": 0.7446,
       "step": 18530
     },
     {
       "epoch": 3.489553924336533,
-      "grad_norm": 16.817617416381836,
-      "learning_rate": 9.062676453980801e-06,
-      "loss": 0.3922,
+      "grad_norm": 5.2538652420043945,
+      "learning_rate": 0.00015104460756634668,
+      "loss": 0.4836,
       "step": 18540
     },
     {
       "epoch": 3.4914361001317524,
-      "grad_norm": 7.2339768409729,
-      "learning_rate": 9.051383399209486e-06,
-      "loss": 0.5843,
+      "grad_norm": 1.902012825012207,
+      "learning_rate": 0.00015085638998682478,
+      "loss": 0.5584,
       "step": 18550
     },
     {
       "epoch": 3.4933182759269714,
-      "grad_norm": 7.782161712646484,
-      "learning_rate": 9.040090344438172e-06,
-      "loss": 0.6768,
+      "grad_norm": 2.5823628902435303,
+      "learning_rate": 0.00015066817240730286,
+      "loss": 0.6969,
       "step": 18560
     },
     {
       "epoch": 3.495200451722191,
-      "grad_norm": 14.207245826721191,
-      "learning_rate": 9.028797289666856e-06,
-      "loss": 0.5713,
+      "grad_norm": 1.9991360902786255,
+      "learning_rate": 0.00015047995482778094,
+      "loss": 0.5022,
       "step": 18570
     },
     {
       "epoch": 3.4970826275174103,
-      "grad_norm": 13.005362510681152,
-      "learning_rate": 9.017504234895539e-06,
-      "loss": 0.7295,
+      "grad_norm": 3.4400622844696045,
+      "learning_rate": 0.00015029173724825898,
+      "loss": 0.6502,
       "step": 18580
     },
     {
       "epoch": 3.4989648033126293,
-      "grad_norm": 5.221713066101074,
-      "learning_rate": 9.006211180124223e-06,
-      "loss": 0.5095,
+      "grad_norm": 1.1559652090072632,
+      "learning_rate": 0.00015010351966873706,
+      "loss": 0.5335,
       "step": 18590
     },
     {
       "epoch": 3.5008469791078487,
-      "grad_norm": 6.634870529174805,
-      "learning_rate": 8.994918125352909e-06,
-      "loss": 0.6905,
+      "grad_norm": 1.5509639978408813,
+      "learning_rate": 0.00014991530208921514,
+      "loss": 0.6104,
       "step": 18600
     },
     {
       "epoch": 3.5027291549030677,
-      "grad_norm": 22.765445709228516,
-      "learning_rate": 8.983625070581594e-06,
-      "loss": 0.4576,
+      "grad_norm": 4.370759010314941,
+      "learning_rate": 0.00014972708450969321,
+      "loss": 0.4614,
       "step": 18610
     },
     {
       "epoch": 3.504611330698287,
-      "grad_norm": 12.745418548583984,
-      "learning_rate": 8.972332015810276e-06,
-      "loss": 0.7584,
+      "grad_norm": 5.9260993003845215,
+      "learning_rate": 0.00014953886693017126,
+      "loss": 0.8786,
       "step": 18620
     },
     {
       "epoch": 3.5064935064935066,
-      "grad_norm": 7.202927112579346,
-      "learning_rate": 8.96103896103896e-06,
-      "loss": 0.6597,
+      "grad_norm": 1.7598670721054077,
+      "learning_rate": 0.00014935064935064934,
+      "loss": 0.6033,
       "step": 18630
     },
     {
       "epoch": 3.508375682288726,
-      "grad_norm": 19.406003952026367,
-      "learning_rate": 8.949745906267645e-06,
-      "loss": 0.6717,
+      "grad_norm": 3.866478681564331,
+      "learning_rate": 0.00014916243177112742,
+      "loss": 0.4937,
       "step": 18640
     },
     {
       "epoch": 3.510257858083945,
-      "grad_norm": 9.290009498596191,
-      "learning_rate": 8.938452851496331e-06,
-      "loss": 0.4197,
+      "grad_norm": 1.5161197185516357,
+      "learning_rate": 0.0001489742141916055,
+      "loss": 0.6126,
       "step": 18650
     },
     {
       "epoch": 3.5121400338791644,
-      "grad_norm": 29.786117553710938,
-      "learning_rate": 8.927159796725015e-06,
-      "loss": 1.0078,
+      "grad_norm": 4.176273345947266,
+      "learning_rate": 0.0001487859966120836,
+      "loss": 0.9906,
       "step": 18660
     },
     {
       "epoch": 3.5140222096743834,
-      "grad_norm": 27.318464279174805,
-      "learning_rate": 8.915866741953698e-06,
-      "loss": 0.6775,
+      "grad_norm": 8.102840423583984,
+      "learning_rate": 0.00014859777903256165,
+      "loss": 0.7627,
       "step": 18670
     },
     {
       "epoch": 3.515904385469603,
-      "grad_norm": 3.237138509750366,
-      "learning_rate": 8.904573687182382e-06,
-      "loss": 0.429,
+      "grad_norm": 1.2033134698867798,
+      "learning_rate": 0.00014840956145303972,
+      "loss": 0.3544,
       "step": 18680
     },
     {
       "epoch": 3.5177865612648223,
-      "grad_norm": 5.157870292663574,
-      "learning_rate": 8.893280632411067e-06,
-      "loss": 0.5416,
+      "grad_norm": 2.3178234100341797,
+      "learning_rate": 0.0001482213438735178,
+      "loss": 0.6326,
       "step": 18690
     },
     {
       "epoch": 3.5196687370600412,
-      "grad_norm": 16.835599899291992,
-      "learning_rate": 8.881987577639753e-06,
-      "loss": 0.6814,
+      "grad_norm": 4.31680965423584,
+      "learning_rate": 0.00014803312629399587,
+      "loss": 0.6204,
       "step": 18700
     },
     {
       "epoch": 3.5215509128552607,
-      "grad_norm": 15.013919830322266,
-      "learning_rate": 8.870694522868436e-06,
-      "loss": 0.6731,
+      "grad_norm": 3.589350700378418,
+      "learning_rate": 0.00014784490871447392,
+      "loss": 0.4283,
       "step": 18710
     },
     {
       "epoch": 3.52343308865048,
-      "grad_norm": 20.667890548706055,
-      "learning_rate": 8.85940146809712e-06,
-      "loss": 0.5596,
+      "grad_norm": 3.9112446308135986,
+      "learning_rate": 0.000147656691134952,
+      "loss": 0.647,
       "step": 18720
     },
     {
       "epoch": 3.525315264445699,
-      "grad_norm": 14.67215347290039,
-      "learning_rate": 8.848108413325804e-06,
-      "loss": 0.6222,
+      "grad_norm": 1.732999563217163,
+      "learning_rate": 0.00014746847355543008,
+      "loss": 0.7145,
       "step": 18730
     },
     {
       "epoch": 3.5271974402409185,
-      "grad_norm": 11.272614479064941,
-      "learning_rate": 8.83681535855449e-06,
-      "loss": 0.8376,
+      "grad_norm": 1.423341155052185,
+      "learning_rate": 0.00014728025597590815,
+      "loss": 0.8925,
       "step": 18740
     },
     {
       "epoch": 3.529079616036138,
-      "grad_norm": 10.547713279724121,
-      "learning_rate": 8.825522303783173e-06,
-      "loss": 0.5504,
+      "grad_norm": 2.423460006713867,
+      "learning_rate": 0.0001470920383963862,
+      "loss": 0.6344,
       "step": 18750
     },
     {
       "epoch": 3.530961791831357,
-      "grad_norm": 12.09305191040039,
-      "learning_rate": 8.814229249011858e-06,
-      "loss": 0.5685,
+      "grad_norm": 4.275598049163818,
+      "learning_rate": 0.00014690382081686428,
+      "loss": 0.6458,
       "step": 18760
     },
     {
       "epoch": 3.5328439676265764,
-      "grad_norm": 13.567402839660645,
-      "learning_rate": 8.802936194240542e-06,
-      "loss": 0.4545,
+      "grad_norm": 3.5608088970184326,
+      "learning_rate": 0.00014671560323734238,
+      "loss": 0.6465,
       "step": 18770
     },
     {
       "epoch": 3.5347261434217954,
-      "grad_norm": 8.988545417785645,
-      "learning_rate": 8.791643139469226e-06,
-      "loss": 0.7346,
+      "grad_norm": 8.03825855255127,
+      "learning_rate": 0.00014652738565782046,
+      "loss": 0.5933,
       "step": 18780
     },
     {
       "epoch": 3.536608319217015,
-      "grad_norm": 13.106191635131836,
-      "learning_rate": 8.780350084697912e-06,
-      "loss": 0.6023,
+      "grad_norm": 0.4490392506122589,
+      "learning_rate": 0.00014633916807829854,
+      "loss": 0.6048,
       "step": 18790
     },
     {
       "epoch": 3.5384904950122342,
-      "grad_norm": 28.7813777923584,
-      "learning_rate": 8.769057029926595e-06,
-      "loss": 0.6156,
+      "grad_norm": 5.7437214851379395,
+      "learning_rate": 0.00014615095049877658,
+      "loss": 0.7403,
       "step": 18800
     },
     {
       "epoch": 3.5403726708074537,
-      "grad_norm": 22.4103946685791,
-      "learning_rate": 8.75776397515528e-06,
-      "loss": 0.4925,
+      "grad_norm": 2.6373136043548584,
+      "learning_rate": 0.00014596273291925466,
+      "loss": 0.4953,
       "step": 18810
     },
     {
       "epoch": 3.5422548466026726,
-      "grad_norm": 20.627689361572266,
-      "learning_rate": 8.746470920383964e-06,
-      "loss": 0.6625,
+      "grad_norm": 3.8211348056793213,
+      "learning_rate": 0.00014577451533973274,
+      "loss": 0.6762,
       "step": 18820
     },
     {
       "epoch": 3.544137022397892,
-      "grad_norm": 8.956947326660156,
-      "learning_rate": 8.73517786561265e-06,
-      "loss": 0.5561,
+      "grad_norm": 2.7318568229675293,
+      "learning_rate": 0.00014558629776021081,
+      "loss": 0.6779,
       "step": 18830
     },
     {
       "epoch": 3.546019198193111,
-      "grad_norm": 7.5617594718933105,
-      "learning_rate": 8.723884810841333e-06,
-      "loss": 0.604,
+      "grad_norm": 1.7246413230895996,
+      "learning_rate": 0.00014539808018068886,
+      "loss": 0.5515,
       "step": 18840
     },
     {
       "epoch": 3.5479013739883305,
-      "grad_norm": 17.85495948791504,
-      "learning_rate": 8.712591756070017e-06,
-      "loss": 0.6587,
+      "grad_norm": 3.1513025760650635,
+      "learning_rate": 0.00014520986260116694,
+      "loss": 0.7151,
       "step": 18850
     },
     {
       "epoch": 3.54978354978355,
-      "grad_norm": 8.020273208618164,
-      "learning_rate": 8.701298701298701e-06,
-      "loss": 0.8588,
+      "grad_norm": 5.167235851287842,
+      "learning_rate": 0.00014502164502164502,
+      "loss": 0.8498,
       "step": 18860
     },
     {
       "epoch": 3.551665725578769,
-      "grad_norm": 25.53630256652832,
-      "learning_rate": 8.690005646527386e-06,
-      "loss": 0.6081,
+      "grad_norm": 3.688398838043213,
+      "learning_rate": 0.0001448334274421231,
+      "loss": 0.6606,
       "step": 18870
     },
     {
       "epoch": 3.5535479013739883,
-      "grad_norm": 5.950412273406982,
-      "learning_rate": 8.67871259175607e-06,
-      "loss": 0.3223,
+      "grad_norm": 2.404766082763672,
+      "learning_rate": 0.00014464520986260117,
+      "loss": 0.3633,
       "step": 18880
     },
     {
       "epoch": 3.5554300771692073,
-      "grad_norm": 7.792263031005859,
-      "learning_rate": 8.667419536984755e-06,
-      "loss": 0.6444,
+      "grad_norm": 1.7548309564590454,
+      "learning_rate": 0.00014445699228307925,
+      "loss": 0.7681,
       "step": 18890
     },
     {
       "epoch": 3.5573122529644268,
-      "grad_norm": 8.66256332397461,
-      "learning_rate": 8.656126482213439e-06,
-      "loss": 0.3844,
+      "grad_norm": 2.3239974975585938,
+      "learning_rate": 0.00014426877470355732,
+      "loss": 0.5125,
       "step": 18900
     },
     {
       "epoch": 3.559194428759646,
-      "grad_norm": 22.50640296936035,
-      "learning_rate": 8.644833427442123e-06,
-      "loss": 0.6602,
+      "grad_norm": 3.089728355407715,
+      "learning_rate": 0.0001440805571240354,
+      "loss": 0.7069,
       "step": 18910
     },
     {
       "epoch": 3.5610766045548656,
-      "grad_norm": 2.1984527111053467,
-      "learning_rate": 8.633540372670808e-06,
-      "loss": 0.288,
+      "grad_norm": 2.1467270851135254,
+      "learning_rate": 0.00014389233954451347,
+      "loss": 0.4031,
       "step": 18920
     },
     {
       "epoch": 3.5629587803500846,
-      "grad_norm": 8.957244873046875,
-      "learning_rate": 8.622247317899492e-06,
-      "loss": 0.7115,
+      "grad_norm": 1.7423038482666016,
+      "learning_rate": 0.00014370412196499152,
+      "loss": 0.8048,
       "step": 18930
     },
     {
       "epoch": 3.564840956145304,
-      "grad_norm": 12.717720985412598,
-      "learning_rate": 8.610954263128176e-06,
-      "loss": 0.5397,
+      "grad_norm": 2.2575857639312744,
+      "learning_rate": 0.0001435159043854696,
+      "loss": 0.5316,
       "step": 18940
     },
     {
       "epoch": 3.566723131940523,
-      "grad_norm": 17.717315673828125,
-      "learning_rate": 8.59966120835686e-06,
-      "loss": 0.8032,
+      "grad_norm": 5.246999740600586,
+      "learning_rate": 0.00014332768680594768,
+      "loss": 0.834,
       "step": 18950
     },
     {
       "epoch": 3.5686053077357425,
-      "grad_norm": 35.10055160522461,
-      "learning_rate": 8.588368153585545e-06,
-      "loss": 0.6323,
+      "grad_norm": 2.156754493713379,
+      "learning_rate": 0.00014313946922642575,
+      "loss": 0.6575,
       "step": 18960
     },
     {
       "epoch": 3.570487483530962,
-      "grad_norm": 16.153400421142578,
-      "learning_rate": 8.57707509881423e-06,
-      "loss": 0.9937,
+      "grad_norm": 2.3258817195892334,
+      "learning_rate": 0.0001429512516469038,
+      "loss": 1.1311,
       "step": 18970
     },
     {
       "epoch": 3.5723696593261813,
-      "grad_norm": 7.813040733337402,
-      "learning_rate": 8.565782044042914e-06,
-      "loss": 0.5768,
+      "grad_norm": 4.504016876220703,
+      "learning_rate": 0.00014276303406738188,
+      "loss": 0.5333,
       "step": 18980
     },
     {
       "epoch": 3.5742518351214003,
-      "grad_norm": 38.98237228393555,
-      "learning_rate": 8.554488989271598e-06,
-      "loss": 0.4722,
+      "grad_norm": 2.8779964447021484,
+      "learning_rate": 0.00014257481648785998,
+      "loss": 0.3841,
       "step": 18990
     },
     {
       "epoch": 3.5761340109166198,
-      "grad_norm": 8.980436325073242,
-      "learning_rate": 8.543195934500283e-06,
-      "loss": 0.6281,
+      "grad_norm": 2.204181432723999,
+      "learning_rate": 0.00014238659890833806,
+      "loss": 0.4968,
       "step": 19000
     },
     {
       "epoch": 3.5780161867118387,
-      "grad_norm": 4.206551551818848,
-      "learning_rate": 8.531902879728965e-06,
-      "loss": 0.7266,
+      "grad_norm": 1.2488787174224854,
+      "learning_rate": 0.0001421983813288161,
+      "loss": 0.7569,
       "step": 19010
     },
     {
       "epoch": 3.579898362507058,
-      "grad_norm": 7.171810150146484,
-      "learning_rate": 8.520609824957651e-06,
-      "loss": 0.5348,
+      "grad_norm": 3.739560842514038,
+      "learning_rate": 0.00014201016374929418,
+      "loss": 0.6607,
       "step": 19020
     },
     {
       "epoch": 3.5817805383022776,
-      "grad_norm": 8.998128890991211,
-      "learning_rate": 8.509316770186336e-06,
-      "loss": 0.3459,
+      "grad_norm": 5.1085381507873535,
+      "learning_rate": 0.00014182194616977226,
+      "loss": 0.4285,
       "step": 19030
     },
     {
       "epoch": 3.5836627140974966,
-      "grad_norm": 11.471923828125,
-      "learning_rate": 8.49802371541502e-06,
-      "loss": 0.5549,
+      "grad_norm": 2.485764980316162,
+      "learning_rate": 0.00014163372859025034,
+      "loss": 0.4453,
       "step": 19040
     },
     {
       "epoch": 3.585544889892716,
-      "grad_norm": 1.2988466024398804,
-      "learning_rate": 8.486730660643705e-06,
-      "loss": 0.4202,
+      "grad_norm": 0.32622551918029785,
+      "learning_rate": 0.00014144551101072841,
+      "loss": 0.4728,
       "step": 19050
     },
     {
       "epoch": 3.587427065687935,
-      "grad_norm": 12.022832870483398,
-      "learning_rate": 8.475437605872387e-06,
-      "loss": 0.882,
+      "grad_norm": 1.6413334608078003,
+      "learning_rate": 0.00014125729343120646,
+      "loss": 0.8364,
       "step": 19060
     },
     {
       "epoch": 3.5893092414831544,
-      "grad_norm": 18.806732177734375,
-      "learning_rate": 8.464144551101073e-06,
-      "loss": 0.4618,
+      "grad_norm": 2.5387015342712402,
+      "learning_rate": 0.00014106907585168454,
+      "loss": 0.4835,
       "step": 19070
     },
     {
       "epoch": 3.591191417278374,
-      "grad_norm": 7.265424728393555,
-      "learning_rate": 8.452851496329758e-06,
-      "loss": 0.5997,
+      "grad_norm": 2.042985677719116,
+      "learning_rate": 0.00014088085827216262,
+      "loss": 0.6753,
       "step": 19080
     },
     {
       "epoch": 3.5930735930735933,
-      "grad_norm": 11.584159851074219,
-      "learning_rate": 8.441558441558442e-06,
-      "loss": 0.6913,
+      "grad_norm": 6.939350128173828,
+      "learning_rate": 0.0001406926406926407,
+      "loss": 0.7062,
       "step": 19090
     },
     {
       "epoch": 3.5949557688688123,
-      "grad_norm": 39.18511199951172,
-      "learning_rate": 8.430265386787125e-06,
-      "loss": 0.6348,
+      "grad_norm": 5.651983737945557,
+      "learning_rate": 0.00014050442311311877,
+      "loss": 0.5953,
       "step": 19100
     },
     {
       "epoch": 3.5968379446640317,
-      "grad_norm": 29.593963623046875,
-      "learning_rate": 8.418972332015811e-06,
-      "loss": 0.7804,
+      "grad_norm": 5.715653896331787,
+      "learning_rate": 0.00014031620553359685,
+      "loss": 0.7144,
       "step": 19110
     },
     {
       "epoch": 3.5987201204592507,
-      "grad_norm": 12.343049049377441,
-      "learning_rate": 8.407679277244495e-06,
-      "loss": 0.5887,
+      "grad_norm": 3.6187081336975098,
+      "learning_rate": 0.00014012798795407492,
+      "loss": 0.6545,
       "step": 19120
     },
     {
       "epoch": 3.60060229625447,
-      "grad_norm": 27.028095245361328,
-      "learning_rate": 8.39638622247318e-06,
-      "loss": 0.5748,
+      "grad_norm": 2.5229785442352295,
+      "learning_rate": 0.000139939770374553,
+      "loss": 0.584,
       "step": 19130
     },
     {
       "epoch": 3.6024844720496896,
-      "grad_norm": 10.208629608154297,
-      "learning_rate": 8.385093167701862e-06,
-      "loss": 0.5336,
+      "grad_norm": 3.843477487564087,
+      "learning_rate": 0.00013975155279503105,
+      "loss": 0.5401,
       "step": 19140
     },
     {
       "epoch": 3.6043666478449086,
-      "grad_norm": 6.941380023956299,
-      "learning_rate": 8.373800112930547e-06,
-      "loss": 0.5231,
+      "grad_norm": 1.425986886024475,
+      "learning_rate": 0.00013956333521550912,
+      "loss": 0.416,
       "step": 19150
     },
     {
       "epoch": 3.606248823640128,
-      "grad_norm": 24.04265785217285,
-      "learning_rate": 8.362507058159233e-06,
-      "loss": 0.9771,
+      "grad_norm": 5.949957847595215,
+      "learning_rate": 0.0001393751176359872,
+      "loss": 0.885,
       "step": 19160
     },
     {
       "epoch": 3.608130999435347,
-      "grad_norm": 7.969100475311279,
-      "learning_rate": 8.351214003387917e-06,
-      "loss": 0.6908,
+      "grad_norm": 2.58278489112854,
+      "learning_rate": 0.00013918690005646528,
+      "loss": 0.5372,
       "step": 19170
     },
     {
       "epoch": 3.6100131752305664,
-      "grad_norm": 5.254550457000732,
-      "learning_rate": 8.339920948616602e-06,
-      "loss": 0.4686,
+      "grad_norm": 2.7384581565856934,
+      "learning_rate": 0.00013899868247694335,
+      "loss": 0.5182,
       "step": 19180
     },
     {
       "epoch": 3.611895351025786,
-      "grad_norm": 11.416996955871582,
-      "learning_rate": 8.328627893845284e-06,
-      "loss": 0.9016,
+      "grad_norm": 2.234143018722534,
+      "learning_rate": 0.0001388104648974214,
+      "loss": 0.8945,
       "step": 19190
     },
     {
       "epoch": 3.6137775268210053,
-      "grad_norm": 22.956119537353516,
-      "learning_rate": 8.31733483907397e-06,
-      "loss": 0.718,
+      "grad_norm": 2.7675840854644775,
+      "learning_rate": 0.00013862224731789948,
+      "loss": 0.737,
       "step": 19200
     },
     {
       "epoch": 3.6156597026162243,
-      "grad_norm": 20.53351402282715,
-      "learning_rate": 8.306041784302655e-06,
-      "loss": 0.6639,
+      "grad_norm": 5.574830532073975,
+      "learning_rate": 0.00013843402973837758,
+      "loss": 0.6644,
       "step": 19210
     },
     {
       "epoch": 3.6175418784114437,
-      "grad_norm": 15.131000518798828,
-      "learning_rate": 8.294748729531339e-06,
-      "loss": 0.3244,
+      "grad_norm": 4.628572463989258,
+      "learning_rate": 0.00013824581215885566,
+      "loss": 0.5055,
       "step": 19220
     },
     {
       "epoch": 3.6194240542066627,
-      "grad_norm": 24.35951805114746,
-      "learning_rate": 8.283455674760022e-06,
-      "loss": 0.467,
+      "grad_norm": 5.0951032638549805,
+      "learning_rate": 0.0001380575945793337,
+      "loss": 0.3973,
       "step": 19230
     },
     {
       "epoch": 3.621306230001882,
-      "grad_norm": 10.79614543914795,
-      "learning_rate": 8.272162619988706e-06,
-      "loss": 0.6985,
+      "grad_norm": 3.8626248836517334,
+      "learning_rate": 0.00013786937699981178,
+      "loss": 0.7009,
       "step": 19240
     },
     {
       "epoch": 3.6231884057971016,
-      "grad_norm": 8.274187088012695,
-      "learning_rate": 8.260869565217392e-06,
-      "loss": 0.7224,
+      "grad_norm": 1.549597144126892,
+      "learning_rate": 0.00013768115942028986,
+      "loss": 0.7579,
       "step": 19250
     },
     {
       "epoch": 3.625070581592321,
-      "grad_norm": 0.10666459053754807,
-      "learning_rate": 8.249576510446077e-06,
-      "loss": 0.6622,
+      "grad_norm": 0.22615107893943787,
+      "learning_rate": 0.00013749294184076794,
+      "loss": 0.7572,
       "step": 19260
     },
     {
       "epoch": 3.62695275738754,
-      "grad_norm": 24.340177536010742,
-      "learning_rate": 8.238283455674761e-06,
-      "loss": 0.7348,
+      "grad_norm": 1.8409181833267212,
+      "learning_rate": 0.00013730472426124601,
+      "loss": 0.6716,
       "step": 19270
     },
     {
       "epoch": 3.6288349331827594,
-      "grad_norm": 20.038578033447266,
-      "learning_rate": 8.226990400903444e-06,
-      "loss": 0.8442,
+      "grad_norm": 4.8296027183532715,
+      "learning_rate": 0.00013711650668172406,
+      "loss": 0.8643,
       "step": 19280
     },
     {
       "epoch": 3.6307171089779784,
-      "grad_norm": 11.465856552124023,
-      "learning_rate": 8.215697346132128e-06,
-      "loss": 0.7398,
+      "grad_norm": 1.5565869808197021,
+      "learning_rate": 0.00013692828910220214,
+      "loss": 0.6602,
       "step": 19290
     },
     {
       "epoch": 3.632599284773198,
-      "grad_norm": 13.356294631958008,
-      "learning_rate": 8.204404291360814e-06,
-      "loss": 0.4302,
+      "grad_norm": 3.0663068294525146,
+      "learning_rate": 0.00013674007152268022,
+      "loss": 0.5523,
       "step": 19300
     },
     {
       "epoch": 3.6344814605684173,
-      "grad_norm": 17.840681076049805,
-      "learning_rate": 8.193111236589499e-06,
-      "loss": 0.562,
+      "grad_norm": 4.205318450927734,
+      "learning_rate": 0.0001365518539431583,
+      "loss": 0.6649,
       "step": 19310
     },
     {
       "epoch": 3.6363636363636362,
-      "grad_norm": 0.5151638984680176,
-      "learning_rate": 8.181818181818181e-06,
-      "loss": 0.5005,
+      "grad_norm": 3.561537742614746,
+      "learning_rate": 0.00013636363636363637,
+      "loss": 0.6846,
       "step": 19320
     },
     {
       "epoch": 3.6382458121588557,
-      "grad_norm": 8.19641399383545,
-      "learning_rate": 8.170525127046866e-06,
-      "loss": 0.6157,
+      "grad_norm": 1.4255841970443726,
+      "learning_rate": 0.00013617541878411445,
+      "loss": 0.6412,
       "step": 19330
     },
     {
       "epoch": 3.6401279879540747,
-      "grad_norm": 19.37883949279785,
-      "learning_rate": 8.159232072275552e-06,
-      "loss": 0.5518,
+      "grad_norm": 4.167680263519287,
+      "learning_rate": 0.00013598720120459252,
+      "loss": 0.5654,
       "step": 19340
     },
     {
       "epoch": 3.642010163749294,
-      "grad_norm": 2.3880703449249268,
-      "learning_rate": 8.147939017504236e-06,
-      "loss": 0.5452,
+      "grad_norm": 2.158613443374634,
+      "learning_rate": 0.0001357989836250706,
+      "loss": 0.5479,
       "step": 19350
     },
     {
       "epoch": 3.6438923395445135,
-      "grad_norm": 12.739452362060547,
-      "learning_rate": 8.136645962732919e-06,
-      "loss": 0.945,
+      "grad_norm": 2.723226308822632,
+      "learning_rate": 0.00013561076604554865,
+      "loss": 1.0151,
       "step": 19360
     },
     {
       "epoch": 3.645774515339733,
-      "grad_norm": 14.899114608764648,
-      "learning_rate": 8.125352907961603e-06,
-      "loss": 0.9551,
+      "grad_norm": 2.7262959480285645,
+      "learning_rate": 0.00013542254846602672,
+      "loss": 0.8649,
       "step": 19370
     },
     {
       "epoch": 3.647656691134952,
-      "grad_norm": 24.746118545532227,
-      "learning_rate": 8.114059853190288e-06,
-      "loss": 0.7507,
+      "grad_norm": 3.184572458267212,
+      "learning_rate": 0.0001352343308865048,
+      "loss": 0.8191,
       "step": 19380
     },
     {
       "epoch": 3.6495388669301714,
-      "grad_norm": 10.281793594360352,
-      "learning_rate": 8.102766798418974e-06,
-      "loss": 0.5547,
+      "grad_norm": 2.500969409942627,
+      "learning_rate": 0.00013504611330698288,
+      "loss": 0.6022,
       "step": 19390
     },
     {
       "epoch": 3.6514210427253904,
-      "grad_norm": 0.5525919198989868,
-      "learning_rate": 8.091473743647658e-06,
-      "loss": 0.5547,
+      "grad_norm": 0.1688629537820816,
+      "learning_rate": 0.00013485789572746095,
+      "loss": 0.7283,
       "step": 19400
     },
     {
       "epoch": 3.65330321852061,
-      "grad_norm": 51.09165573120117,
-      "learning_rate": 8.08018068887634e-06,
-      "loss": 0.6717,
+      "grad_norm": 4.442159175872803,
+      "learning_rate": 0.000134669678147939,
+      "loss": 0.7114,
       "step": 19410
     },
     {
       "epoch": 3.6551853943158292,
-      "grad_norm": 18.030263900756836,
-      "learning_rate": 8.068887634105025e-06,
-      "loss": 0.6471,
+      "grad_norm": 3.3339366912841797,
+      "learning_rate": 0.00013448146056841708,
+      "loss": 0.8124,
       "step": 19420
     },
     {
       "epoch": 3.657067570111048,
-      "grad_norm": 10.15441608428955,
-      "learning_rate": 8.05759457933371e-06,
-      "loss": 0.5754,
+      "grad_norm": 3.360987663269043,
+      "learning_rate": 0.00013429324298889518,
+      "loss": 0.5944,
       "step": 19430
     },
     {
       "epoch": 3.6589497459062676,
-      "grad_norm": 8.153667449951172,
-      "learning_rate": 8.046301524562396e-06,
-      "loss": 0.4898,
+      "grad_norm": 1.367207407951355,
+      "learning_rate": 0.00013410502540937326,
+      "loss": 0.5212,
       "step": 19440
     },
     {
       "epoch": 3.660831921701487,
-      "grad_norm": 22.26079750061035,
-      "learning_rate": 8.035008469791078e-06,
-      "loss": 0.6954,
+      "grad_norm": 2.3143234252929688,
+      "learning_rate": 0.0001339168078298513,
+      "loss": 0.7531,
       "step": 19450
     },
     {
       "epoch": 3.662714097496706,
-      "grad_norm": 8.062853813171387,
-      "learning_rate": 8.023715415019763e-06,
-      "loss": 0.5817,
+      "grad_norm": 2.4220802783966064,
+      "learning_rate": 0.00013372859025032938,
+      "loss": 0.573,
       "step": 19460
     },
     {
       "epoch": 3.6645962732919255,
-      "grad_norm": 20.452653884887695,
-      "learning_rate": 8.012422360248447e-06,
-      "loss": 0.6119,
+      "grad_norm": 3.9013524055480957,
+      "learning_rate": 0.00013354037267080746,
+      "loss": 0.6934,
       "step": 19470
     },
     {
       "epoch": 3.666478449087145,
-      "grad_norm": 16.677806854248047,
-      "learning_rate": 8.001129305477133e-06,
-      "loss": 0.5955,
+      "grad_norm": 2.481264114379883,
+      "learning_rate": 0.00013335215509128554,
+      "loss": 0.6393,
       "step": 19480
     },
     {
       "epoch": 3.668360624882364,
-      "grad_norm": 13.450298309326172,
-      "learning_rate": 7.989836250705816e-06,
-      "loss": 0.4935,
+      "grad_norm": 1.6518964767456055,
+      "learning_rate": 0.0001331639375117636,
+      "loss": 0.4722,
       "step": 19490
     },
     {
       "epoch": 3.6702428006775834,
-      "grad_norm": 12.314119338989258,
-      "learning_rate": 7.9785431959345e-06,
-      "loss": 0.7169,
+      "grad_norm": 3.1066956520080566,
+      "learning_rate": 0.00013297571993224166,
+      "loss": 0.722,
       "step": 19500
     },
     {
       "epoch": 3.6721249764728023,
-      "grad_norm": 13.286541938781738,
-      "learning_rate": 7.967250141163185e-06,
-      "loss": 0.7006,
+      "grad_norm": 3.954586982727051,
+      "learning_rate": 0.00013278750235271974,
+      "loss": 0.5826,
       "step": 19510
     },
     {
       "epoch": 3.6740071522680218,
-      "grad_norm": 8.442977905273438,
-      "learning_rate": 7.955957086391869e-06,
-      "loss": 0.4225,
+      "grad_norm": 1.8204858303070068,
+      "learning_rate": 0.00013259928477319782,
+      "loss": 0.6208,
       "step": 19520
     },
     {
       "epoch": 3.675889328063241,
-      "grad_norm": 18.344675064086914,
-      "learning_rate": 7.944664031620555e-06,
-      "loss": 0.5021,
+      "grad_norm": 2.552551746368408,
+      "learning_rate": 0.0001324110671936759,
+      "loss": 0.4284,
       "step": 19530
     },
     {
       "epoch": 3.6777715038584606,
-      "grad_norm": 22.64541244506836,
-      "learning_rate": 7.933370976849238e-06,
-      "loss": 0.596,
+      "grad_norm": 4.546378135681152,
+      "learning_rate": 0.00013222284961415397,
+      "loss": 0.6156,
       "step": 19540
     },
     {
       "epoch": 3.6796536796536796,
-      "grad_norm": 20.384031295776367,
-      "learning_rate": 7.922077922077922e-06,
-      "loss": 0.6877,
+      "grad_norm": 3.2740485668182373,
+      "learning_rate": 0.00013203463203463205,
+      "loss": 0.6867,
       "step": 19550
     },
     {
       "epoch": 3.681535855448899,
-      "grad_norm": 8.203783988952637,
-      "learning_rate": 7.910784867306606e-06,
-      "loss": 0.6984,
+      "grad_norm": 5.2918877601623535,
+      "learning_rate": 0.00013184641445511012,
+      "loss": 0.6754,
       "step": 19560
     },
     {
       "epoch": 3.683418031244118,
-      "grad_norm": 8.949334144592285,
-      "learning_rate": 7.89949181253529e-06,
-      "loss": 0.642,
+      "grad_norm": 2.1436519622802734,
+      "learning_rate": 0.0001316581968755882,
+      "loss": 0.674,
       "step": 19570
     },
     {
       "epoch": 3.6853002070393375,
-      "grad_norm": 9.0608491897583,
-      "learning_rate": 7.888198757763975e-06,
-      "loss": 0.6827,
+      "grad_norm": 1.6020269393920898,
+      "learning_rate": 0.00013146997929606625,
+      "loss": 0.6801,
       "step": 19580
     },
     {
       "epoch": 3.687182382834557,
-      "grad_norm": 11.608131408691406,
-      "learning_rate": 7.87690570299266e-06,
-      "loss": 0.5154,
+      "grad_norm": 3.0209949016571045,
+      "learning_rate": 0.00013128176171654432,
+      "loss": 0.5285,
       "step": 19590
     },
     {
       "epoch": 3.689064558629776,
-      "grad_norm": 42.18769836425781,
-      "learning_rate": 7.865612648221344e-06,
-      "loss": 0.6003,
+      "grad_norm": 6.093017101287842,
+      "learning_rate": 0.0001310935441370224,
+      "loss": 0.8178,
       "step": 19600
     },
     {
       "epoch": 3.6909467344249953,
-      "grad_norm": 10.675950050354004,
-      "learning_rate": 7.854319593450028e-06,
-      "loss": 0.4056,
+      "grad_norm": 2.7886905670166016,
+      "learning_rate": 0.00013090532655750048,
+      "loss": 0.5863,
       "step": 19610
     },
     {
       "epoch": 3.6928289102202143,
-      "grad_norm": 21.224416732788086,
-      "learning_rate": 7.843026538678713e-06,
-      "loss": 0.5917,
+      "grad_norm": 3.8390111923217773,
+      "learning_rate": 0.00013071710897797853,
+      "loss": 0.5651,
       "step": 19620
     },
     {
       "epoch": 3.6947110860154337,
-      "grad_norm": 13.734648704528809,
-      "learning_rate": 7.831733483907397e-06,
-      "loss": 0.6431,
+      "grad_norm": 4.392716407775879,
+      "learning_rate": 0.0001305288913984566,
+      "loss": 0.7295,
       "step": 19630
     },
     {
       "epoch": 3.696593261810653,
-      "grad_norm": 5.891987323760986,
-      "learning_rate": 7.820440429136081e-06,
-      "loss": 0.4731,
+      "grad_norm": 2.144402503967285,
+      "learning_rate": 0.00013034067381893468,
+      "loss": 0.6048,
       "step": 19640
     },
     {
       "epoch": 3.6984754376058726,
-      "grad_norm": 6.475367546081543,
-      "learning_rate": 7.809147374364766e-06,
-      "loss": 0.5094,
+      "grad_norm": 0.8373141288757324,
+      "learning_rate": 0.00013015245623941278,
+      "loss": 0.6461,
       "step": 19650
     },
     {
       "epoch": 3.7003576134010916,
-      "grad_norm": 5.192591667175293,
-      "learning_rate": 7.79785431959345e-06,
-      "loss": 0.508,
+      "grad_norm": 1.1086032390594482,
+      "learning_rate": 0.00012996423865989086,
+      "loss": 0.5463,
       "step": 19660
     },
     {
       "epoch": 3.702239789196311,
-      "grad_norm": 5.926048755645752,
-      "learning_rate": 7.786561264822135e-06,
-      "loss": 0.5057,
+      "grad_norm": 4.724222183227539,
+      "learning_rate": 0.0001297760210803689,
+      "loss": 0.7127,
       "step": 19670
     },
     {
       "epoch": 3.70412196499153,
-      "grad_norm": 41.563819885253906,
-      "learning_rate": 7.775268210050819e-06,
-      "loss": 0.7474,
+      "grad_norm": 5.788373947143555,
+      "learning_rate": 0.00012958780350084698,
+      "loss": 0.7291,
       "step": 19680
     },
     {
       "epoch": 3.7060041407867494,
-      "grad_norm": 22.899145126342773,
-      "learning_rate": 7.763975155279503e-06,
-      "loss": 0.4772,
+      "grad_norm": 2.612990617752075,
+      "learning_rate": 0.00012939958592132506,
+      "loss": 0.5288,
       "step": 19690
     },
     {
       "epoch": 3.707886316581969,
-      "grad_norm": 1.0630342960357666,
-      "learning_rate": 7.752682100508188e-06,
-      "loss": 0.3682,
+      "grad_norm": 0.08691232651472092,
+      "learning_rate": 0.00012921136834180314,
+      "loss": 0.3476,
       "step": 19700
     },
     {
       "epoch": 3.709768492377188,
-      "grad_norm": 10.798871994018555,
-      "learning_rate": 7.741389045736872e-06,
-      "loss": 0.7488,
+      "grad_norm": 4.636226654052734,
+      "learning_rate": 0.0001290231507622812,
+      "loss": 0.786,
       "step": 19710
     },
     {
       "epoch": 3.7116506681724073,
-      "grad_norm": 9.957785606384277,
-      "learning_rate": 7.730095990965557e-06,
-      "loss": 0.4364,
+      "grad_norm": 3.9436421394348145,
+      "learning_rate": 0.00012883493318275926,
+      "loss": 0.4371,
       "step": 19720
     },
     {
       "epoch": 3.7135328439676267,
-      "grad_norm": 3.4311161041259766,
-      "learning_rate": 7.718802936194241e-06,
-      "loss": 0.4192,
+      "grad_norm": 1.0998643636703491,
+      "learning_rate": 0.00012864671560323734,
+      "loss": 0.5665,
       "step": 19730
     },
     {
       "epoch": 3.7154150197628457,
-      "grad_norm": 10.924822807312012,
-      "learning_rate": 7.707509881422925e-06,
-      "loss": 0.5174,
+      "grad_norm": 5.126893520355225,
+      "learning_rate": 0.00012845849802371542,
+      "loss": 0.6374,
       "step": 19740
     },
     {
       "epoch": 3.717297195558065,
-      "grad_norm": 11.127129554748535,
-      "learning_rate": 7.696216826651608e-06,
-      "loss": 0.4632,
+      "grad_norm": 1.9523310661315918,
+      "learning_rate": 0.00012827028044419347,
+      "loss": 0.6891,
       "step": 19750
     },
     {
       "epoch": 3.7191793713532846,
-      "grad_norm": 2.7011196613311768,
-      "learning_rate": 7.684923771880294e-06,
-      "loss": 0.499,
+      "grad_norm": 1.3989789485931396,
+      "learning_rate": 0.00012808206286467157,
+      "loss": 0.5985,
       "step": 19760
     },
     {
       "epoch": 3.7210615471485036,
-      "grad_norm": 3.9069223403930664,
-      "learning_rate": 7.673630717108978e-06,
-      "loss": 0.6029,
+      "grad_norm": 0.2790946960449219,
+      "learning_rate": 0.00012789384528514965,
+      "loss": 0.7122,
       "step": 19770
     },
     {
       "epoch": 3.722943722943723,
-      "grad_norm": 4.461427211761475,
-      "learning_rate": 7.662337662337663e-06,
-      "loss": 0.7491,
+      "grad_norm": 2.048234224319458,
+      "learning_rate": 0.00012770562770562772,
+      "loss": 0.7603,
       "step": 19780
     },
     {
       "epoch": 3.724825898738942,
-      "grad_norm": 19.533382415771484,
-      "learning_rate": 7.651044607566347e-06,
-      "loss": 0.6521,
+      "grad_norm": 3.0206234455108643,
+      "learning_rate": 0.0001275174101261058,
+      "loss": 0.7907,
       "step": 19790
     },
     {
       "epoch": 3.7267080745341614,
-      "grad_norm": 2.6040291786193848,
-      "learning_rate": 7.63975155279503e-06,
-      "loss": 0.4989,
+      "grad_norm": 2.8944666385650635,
+      "learning_rate": 0.00012732919254658385,
+      "loss": 0.6928,
       "step": 19800
     },
     {
       "epoch": 3.728590250329381,
-      "grad_norm": 11.266732215881348,
-      "learning_rate": 7.628458498023715e-06,
-      "loss": 0.5427,
+      "grad_norm": 2.925626754760742,
+      "learning_rate": 0.00012714097496706192,
+      "loss": 0.6453,
       "step": 19810
     },
     {
       "epoch": 3.7304724261246003,
-      "grad_norm": 0.7041659951210022,
-      "learning_rate": 7.6171654432524e-06,
-      "loss": 0.482,
+      "grad_norm": 2.9843673706054688,
+      "learning_rate": 0.00012695275738754,
+      "loss": 0.5517,
       "step": 19820
     },
     {
       "epoch": 3.7323546019198193,
-      "grad_norm": 7.669864654541016,
-      "learning_rate": 7.605872388481085e-06,
-      "loss": 0.4862,
+      "grad_norm": 2.6966941356658936,
+      "learning_rate": 0.00012676453980801808,
+      "loss": 0.6548,
       "step": 19830
     },
     {
       "epoch": 3.7342367777150387,
-      "grad_norm": 10.8816499710083,
-      "learning_rate": 7.594579333709768e-06,
-      "loss": 0.7602,
+      "grad_norm": 2.4660897254943848,
+      "learning_rate": 0.00012657632222849613,
+      "loss": 0.6221,
       "step": 19840
     },
     {
       "epoch": 3.7361189535102577,
-      "grad_norm": 12.468655586242676,
-      "learning_rate": 7.583286278938453e-06,
-      "loss": 0.4088,
+      "grad_norm": 2.4245846271514893,
+      "learning_rate": 0.0001263881046489742,
+      "loss": 0.4908,
       "step": 19850
     },
     {
       "epoch": 3.738001129305477,
-      "grad_norm": 0.618268609046936,
-      "learning_rate": 7.571993224167137e-06,
-      "loss": 0.2964,
+      "grad_norm": 2.279542922973633,
+      "learning_rate": 0.00012619988706945228,
+      "loss": 0.3842,
       "step": 19860
     },
     {
       "epoch": 3.7398833051006966,
-      "grad_norm": 19.64562225341797,
-      "learning_rate": 7.560700169395822e-06,
-      "loss": 0.7775,
+      "grad_norm": 4.801576614379883,
+      "learning_rate": 0.00012601166948993038,
+      "loss": 0.7761,
       "step": 19870
     },
     {
       "epoch": 3.7417654808959155,
-      "grad_norm": 18.206167221069336,
-      "learning_rate": 7.549407114624507e-06,
-      "loss": 0.7153,
+      "grad_norm": 7.5804901123046875,
+      "learning_rate": 0.00012582345191040846,
+      "loss": 0.8545,
       "step": 19880
     },
     {
       "epoch": 3.743647656691135,
-      "grad_norm": 7.192927837371826,
-      "learning_rate": 7.53811405985319e-06,
-      "loss": 0.4181,
+      "grad_norm": 2.122183322906494,
+      "learning_rate": 0.0001256352343308865,
+      "loss": 0.566,
       "step": 19890
     },
     {
       "epoch": 3.745529832486354,
-      "grad_norm": 29.97498893737793,
-      "learning_rate": 7.5268210050818746e-06,
-      "loss": 0.3835,
+      "grad_norm": 0.906879186630249,
+      "learning_rate": 0.00012544701675136458,
+      "loss": 0.3437,
       "step": 19900
     },
     {
       "epoch": 3.7474120082815734,
-      "grad_norm": 5.730542182922363,
-      "learning_rate": 7.51552795031056e-06,
-      "loss": 0.5346,
+      "grad_norm": 1.77112877368927,
+      "learning_rate": 0.00012525879917184266,
+      "loss": 0.5665,
       "step": 19910
     },
     {
       "epoch": 3.749294184076793,
-      "grad_norm": 15.303305625915527,
-      "learning_rate": 7.504234895539244e-06,
-      "loss": 0.5668,
+      "grad_norm": 2.0256805419921875,
+      "learning_rate": 0.00012507058159232074,
+      "loss": 0.5925,
       "step": 19920
     },
     {
       "epoch": 3.7511763598720123,
-      "grad_norm": 7.135068893432617,
-      "learning_rate": 7.4929418407679286e-06,
-      "loss": 0.5419,
+      "grad_norm": 3.9407191276550293,
+      "learning_rate": 0.00012488236401279881,
+      "loss": 0.523,
       "step": 19930
     },
     {
       "epoch": 3.7530585356672312,
-      "grad_norm": 18.504470825195312,
-      "learning_rate": 7.481648785996612e-06,
-      "loss": 0.6651,
+      "grad_norm": 7.084681034088135,
+      "learning_rate": 0.00012469414643327686,
+      "loss": 0.7849,
       "step": 19940
     },
     {
       "epoch": 3.7549407114624507,
-      "grad_norm": 4.049873352050781,
-      "learning_rate": 7.4703557312252965e-06,
-      "loss": 0.6697,
+      "grad_norm": 3.3468875885009766,
+      "learning_rate": 0.00012450592885375494,
+      "loss": 0.7904,
       "step": 19950
     },
     {
       "epoch": 3.7568228872576697,
-      "grad_norm": 0.3904266953468323,
-      "learning_rate": 7.459062676453981e-06,
-      "loss": 0.5437,
+      "grad_norm": 0.175517737865448,
+      "learning_rate": 0.00012431771127423302,
+      "loss": 0.4774,
       "step": 19960
     },
     {
       "epoch": 3.758705063052889,
-      "grad_norm": 2.0129761695861816,
-      "learning_rate": 7.447769621682665e-06,
-      "loss": 0.4477,
+      "grad_norm": 4.669493198394775,
+      "learning_rate": 0.0001241294936947111,
+      "loss": 0.6702,
       "step": 19970
     },
     {
       "epoch": 3.7605872388481085,
-      "grad_norm": 25.443471908569336,
-      "learning_rate": 7.43647656691135e-06,
-      "loss": 0.6205,
+      "grad_norm": 4.180354118347168,
+      "learning_rate": 0.00012394127611518917,
+      "loss": 0.7476,
       "step": 19980
     },
     {
       "epoch": 3.762469414643328,
-      "grad_norm": 27.52094841003418,
-      "learning_rate": 7.425183512140034e-06,
-      "loss": 0.4978,
+      "grad_norm": 4.086312294006348,
+      "learning_rate": 0.00012375305853566725,
+      "loss": 0.5892,
       "step": 19990
     },
     {
       "epoch": 3.764351590438547,
-      "grad_norm": 0.09147056192159653,
-      "learning_rate": 7.413890457368718e-06,
-      "loss": 0.398,
+      "grad_norm": 0.048435747623443604,
+      "learning_rate": 0.0001235648409561453,
+      "loss": 0.415,
       "step": 20000
     },
     {
       "epoch": 3.7662337662337664,
-      "grad_norm": 0.8058493733406067,
-      "learning_rate": 7.402597402597403e-06,
-      "loss": 0.4703,
+      "grad_norm": 0.6088755130767822,
+      "learning_rate": 0.00012337662337662337,
+      "loss": 0.5425,
       "step": 20010
     },
     {
       "epoch": 3.7681159420289854,
-      "grad_norm": 8.946433067321777,
-      "learning_rate": 7.391304347826087e-06,
-      "loss": 0.6018,
+      "grad_norm": 3.2463440895080566,
+      "learning_rate": 0.00012318840579710145,
+      "loss": 0.6901,
       "step": 20020
     },
     {
       "epoch": 3.769998117824205,
-      "grad_norm": 12.180794715881348,
-      "learning_rate": 7.3800112930547715e-06,
-      "loss": 0.5685,
+      "grad_norm": 2.989475965499878,
+      "learning_rate": 0.00012300018821757952,
+      "loss": 0.5331,
       "step": 20030
     },
     {
       "epoch": 3.7718802936194242,
-      "grad_norm": 11.476252555847168,
-      "learning_rate": 7.368718238283456e-06,
-      "loss": 0.5504,
+      "grad_norm": 6.133965492248535,
+      "learning_rate": 0.0001228119706380576,
+      "loss": 0.5936,
       "step": 20040
     },
     {
       "epoch": 3.773762469414643,
-      "grad_norm": 14.593816757202148,
-      "learning_rate": 7.35742518351214e-06,
-      "loss": 0.5439,
+      "grad_norm": 2.2449915409088135,
+      "learning_rate": 0.00012262375305853568,
+      "loss": 0.573,
       "step": 20050
     },
     {
       "epoch": 3.7756446452098626,
-      "grad_norm": 42.94921112060547,
-      "learning_rate": 7.346132128740825e-06,
-      "loss": 0.7765,
+      "grad_norm": 6.024438858032227,
+      "learning_rate": 0.00012243553547901375,
+      "loss": 0.7494,
       "step": 20060
     },
     {
       "epoch": 3.7775268210050816,
-      "grad_norm": 8.45206069946289,
-      "learning_rate": 7.334839073969509e-06,
-      "loss": 0.4416,
+      "grad_norm": 0.22473078966140747,
+      "learning_rate": 0.0001222473178994918,
+      "loss": 0.5063,
       "step": 20070
     },
     {
       "epoch": 3.779408996800301,
-      "grad_norm": 11.394601821899414,
-      "learning_rate": 7.3235460191981934e-06,
-      "loss": 0.5376,
+      "grad_norm": 4.407290458679199,
+      "learning_rate": 0.00012205910031996989,
+      "loss": 0.6881,
       "step": 20080
     },
     {
       "epoch": 3.7812911725955205,
-      "grad_norm": 0.5047250986099243,
-      "learning_rate": 7.312252964426877e-06,
-      "loss": 0.5465,
+      "grad_norm": 3.2443549633026123,
+      "learning_rate": 0.00012187088274044796,
+      "loss": 0.659,
       "step": 20090
     },
     {
       "epoch": 3.78317334839074,
-      "grad_norm": 8.610776901245117,
-      "learning_rate": 7.300959909655562e-06,
-      "loss": 0.4937,
+      "grad_norm": 2.6818675994873047,
+      "learning_rate": 0.00012168266516092603,
+      "loss": 0.6821,
       "step": 20100
     },
     {
       "epoch": 3.785055524185959,
-      "grad_norm": 16.135894775390625,
-      "learning_rate": 7.289666854884246e-06,
-      "loss": 0.3992,
+      "grad_norm": 3.2473978996276855,
+      "learning_rate": 0.0001214944475814041,
+      "loss": 0.2873,
       "step": 20110
     },
     {
       "epoch": 3.7869376999811784,
-      "grad_norm": 10.323850631713867,
-      "learning_rate": 7.278373800112931e-06,
-      "loss": 0.6229,
+      "grad_norm": 2.7352027893066406,
+      "learning_rate": 0.00012130623000188218,
+      "loss": 0.8633,
       "step": 20120
     },
     {
       "epoch": 3.7888198757763973,
-      "grad_norm": 24.070560455322266,
-      "learning_rate": 7.267080745341615e-06,
-      "loss": 0.6724,
+      "grad_norm": 3.0541322231292725,
+      "learning_rate": 0.00012111801242236026,
+      "loss": 0.7022,
       "step": 20130
     },
     {
       "epoch": 3.7907020515716168,
-      "grad_norm": 15.11103343963623,
-      "learning_rate": 7.2557876905703e-06,
-      "loss": 0.4975,
+      "grad_norm": 3.460502862930298,
+      "learning_rate": 0.00012092979484283832,
+      "loss": 0.4832,
       "step": 20140
     },
     {
       "epoch": 3.792584227366836,
-      "grad_norm": 14.725980758666992,
-      "learning_rate": 7.244494635798984e-06,
-      "loss": 0.65,
+      "grad_norm": 3.218946933746338,
+      "learning_rate": 0.0001207415772633164,
+      "loss": 0.8061,
       "step": 20150
     },
     {
       "epoch": 3.794466403162055,
-      "grad_norm": 13.276191711425781,
-      "learning_rate": 7.233201581027668e-06,
-      "loss": 0.2699,
+      "grad_norm": 2.4092910289764404,
+      "learning_rate": 0.00012055335968379446,
+      "loss": 0.3311,
       "step": 20160
     },
     {
       "epoch": 3.7963485789572746,
-      "grad_norm": 16.452234268188477,
-      "learning_rate": 7.221908526256353e-06,
-      "loss": 0.7108,
+      "grad_norm": 1.346550703048706,
+      "learning_rate": 0.00012036514210427254,
+      "loss": 0.609,
       "step": 20170
     },
     {
       "epoch": 3.7982307547524936,
-      "grad_norm": 18.935544967651367,
-      "learning_rate": 7.210615471485036e-06,
-      "loss": 0.4207,
+      "grad_norm": 3.535579204559326,
+      "learning_rate": 0.00012017692452475062,
+      "loss": 0.4008,
       "step": 20180
     },
     {
       "epoch": 3.800112930547713,
-      "grad_norm": 14.351372718811035,
-      "learning_rate": 7.199322416713722e-06,
-      "loss": 0.554,
+      "grad_norm": 3.253265857696533,
+      "learning_rate": 0.00011998870694522869,
+      "loss": 0.5573,
       "step": 20190
     },
     {
       "epoch": 3.8019951063429325,
-      "grad_norm": 2.5707204341888428,
-      "learning_rate": 7.188029361942405e-06,
-      "loss": 0.392,
+      "grad_norm": 0.9760240316390991,
+      "learning_rate": 0.00011980048936570676,
+      "loss": 0.5194,
       "step": 20200
     },
     {
       "epoch": 3.803877282138152,
-      "grad_norm": 1.579572081565857,
-      "learning_rate": 7.17673630717109e-06,
-      "loss": 0.3267,
+      "grad_norm": 4.288007736206055,
+      "learning_rate": 0.00011961227178618483,
+      "loss": 0.514,
       "step": 20210
     },
     {
       "epoch": 3.805759457933371,
-      "grad_norm": 4.0918660163879395,
-      "learning_rate": 7.165443252399774e-06,
-      "loss": 0.6002,
+      "grad_norm": 7.366967678070068,
+      "learning_rate": 0.0001194240542066629,
+      "loss": 0.5861,
       "step": 20220
     },
     {
       "epoch": 3.8076416337285903,
-      "grad_norm": 16.16851043701172,
-      "learning_rate": 7.154150197628458e-06,
-      "loss": 0.3712,
+      "grad_norm": 1.3266446590423584,
+      "learning_rate": 0.00011923583662714098,
+      "loss": 0.3292,
       "step": 20230
     },
     {
       "epoch": 3.8095238095238093,
-      "grad_norm": 11.626598358154297,
-      "learning_rate": 7.142857142857143e-06,
-      "loss": 0.8826,
+      "grad_norm": 2.6739280223846436,
+      "learning_rate": 0.00011904761904761905,
+      "loss": 0.934,
       "step": 20240
     },
     {
       "epoch": 3.8114059853190287,
-      "grad_norm": 17.99329948425293,
-      "learning_rate": 7.131564088085827e-06,
-      "loss": 0.8901,
+      "grad_norm": 4.105169773101807,
+      "learning_rate": 0.00011885940146809712,
+      "loss": 0.9709,
       "step": 20250
     },
     {
       "epoch": 3.813288161114248,
-      "grad_norm": 0.3546718955039978,
-      "learning_rate": 7.120271033314512e-06,
-      "loss": 0.7524,
+      "grad_norm": 2.8270390033721924,
+      "learning_rate": 0.0001186711838885752,
+      "loss": 0.7253,
       "step": 20260
     },
     {
       "epoch": 3.8151703369094676,
-      "grad_norm": 6.967835426330566,
-      "learning_rate": 7.108977978543196e-06,
-      "loss": 0.3611,
+      "grad_norm": 3.085862398147583,
+      "learning_rate": 0.00011848296630905326,
+      "loss": 0.3944,
       "step": 20270
     },
     {
       "epoch": 3.8170525127046866,
-      "grad_norm": 0.7506970763206482,
-      "learning_rate": 7.097684923771881e-06,
-      "loss": 0.6239,
+      "grad_norm": 2.2109007835388184,
+      "learning_rate": 0.00011829474872953134,
+      "loss": 0.6252,
       "step": 20280
     },
     {
       "epoch": 3.818934688499906,
-      "grad_norm": 7.021599769592285,
-      "learning_rate": 7.086391869000565e-06,
-      "loss": 0.8331,
+      "grad_norm": 1.0854592323303223,
+      "learning_rate": 0.00011810653115000942,
+      "loss": 0.8545,
       "step": 20290
     },
     {
       "epoch": 3.820816864295125,
-      "grad_norm": 0.8781418800354004,
-      "learning_rate": 7.075098814229249e-06,
-      "loss": 0.5364,
+      "grad_norm": 2.4989964962005615,
+      "learning_rate": 0.00011791831357048749,
+      "loss": 0.6113,
       "step": 20300
     },
     {
       "epoch": 3.8226990400903444,
-      "grad_norm": 10.943838119506836,
-      "learning_rate": 7.063805759457933e-06,
-      "loss": 0.3288,
+      "grad_norm": 2.8781096935272217,
+      "learning_rate": 0.00011773009599096556,
+      "loss": 0.3864,
       "step": 20310
     },
     {
       "epoch": 3.824581215885564,
-      "grad_norm": 6.258754730224609,
-      "learning_rate": 7.052512704686618e-06,
-      "loss": 0.2297,
+      "grad_norm": 5.14607572555542,
+      "learning_rate": 0.00011754187841144363,
+      "loss": 0.4674,
       "step": 20320
     },
     {
       "epoch": 3.826463391680783,
-      "grad_norm": 6.189356803894043,
-      "learning_rate": 7.041219649915302e-06,
-      "loss": 0.4229,
+      "grad_norm": 2.13116455078125,
+      "learning_rate": 0.0001173536608319217,
+      "loss": 0.5366,
       "step": 20330
     },
     {
       "epoch": 3.8283455674760023,
-      "grad_norm": 0.3370305299758911,
-      "learning_rate": 7.0299265951439865e-06,
-      "loss": 0.8312,
+      "grad_norm": 0.10455270856618881,
+      "learning_rate": 0.00011716544325239978,
+      "loss": 0.9137,
       "step": 20340
     },
     {
       "epoch": 3.8302277432712213,
-      "grad_norm": 0.39705947041511536,
-      "learning_rate": 7.018633540372671e-06,
-      "loss": 0.6544,
+      "grad_norm": 0.04126379266381264,
+      "learning_rate": 0.00011697722567287785,
+      "loss": 0.6901,
       "step": 20350
     },
     {
       "epoch": 3.8321099190664407,
-      "grad_norm": 5.138480186462402,
-      "learning_rate": 7.007340485601355e-06,
-      "loss": 0.5875,
+      "grad_norm": 2.3497095108032227,
+      "learning_rate": 0.00011678900809335592,
+      "loss": 0.5338,
       "step": 20360
     },
     {
       "epoch": 3.83399209486166,
-      "grad_norm": 16.162595748901367,
-      "learning_rate": 6.9960474308300405e-06,
-      "loss": 0.694,
+      "grad_norm": 4.9876179695129395,
+      "learning_rate": 0.000116600790513834,
+      "loss": 0.8286,
       "step": 20370
     },
     {
       "epoch": 3.8358742706568796,
-      "grad_norm": 10.226083755493164,
-      "learning_rate": 6.984754376058724e-06,
-      "loss": 0.6075,
+      "grad_norm": 7.40828800201416,
+      "learning_rate": 0.00011641257293431206,
+      "loss": 0.6535,
       "step": 20380
     },
     {
       "epoch": 3.8377564464520986,
-      "grad_norm": 11.182906150817871,
-      "learning_rate": 6.9734613212874084e-06,
-      "loss": 0.5489,
+      "grad_norm": 2.155172109603882,
+      "learning_rate": 0.00011622435535479014,
+      "loss": 0.5168,
       "step": 20390
     },
     {
       "epoch": 3.839638622247318,
-      "grad_norm": 16.98402214050293,
-      "learning_rate": 6.962168266516093e-06,
-      "loss": 0.5389,
+      "grad_norm": 4.699321269989014,
+      "learning_rate": 0.00011603613777526822,
+      "loss": 0.4554,
       "step": 20400
     },
     {
       "epoch": 3.841520798042537,
-      "grad_norm": 7.632681846618652,
-      "learning_rate": 6.950875211744777e-06,
-      "loss": 0.4979,
+      "grad_norm": 2.950913429260254,
+      "learning_rate": 0.00011584792019574629,
+      "loss": 0.5303,
       "step": 20410
     },
     {
       "epoch": 3.8434029738377564,
-      "grad_norm": 11.018415451049805,
-      "learning_rate": 6.939582156973462e-06,
-      "loss": 0.792,
+      "grad_norm": 1.8240998983383179,
+      "learning_rate": 0.00011565970261622436,
+      "loss": 0.7742,
       "step": 20420
     },
     {
       "epoch": 3.845285149632976,
-      "grad_norm": 18.204879760742188,
-      "learning_rate": 6.928289102202146e-06,
-      "loss": 0.4916,
+      "grad_norm": 5.689794063568115,
+      "learning_rate": 0.00011547148503670243,
+      "loss": 0.5243,
       "step": 20430
     },
     {
       "epoch": 3.847167325428195,
-      "grad_norm": 17.605222702026367,
-      "learning_rate": 6.91699604743083e-06,
-      "loss": 0.2598,
+      "grad_norm": 3.2978150844573975,
+      "learning_rate": 0.0001152832674571805,
+      "loss": 0.354,
       "step": 20440
     },
     {
       "epoch": 3.8490495012234143,
-      "grad_norm": 31.94579315185547,
-      "learning_rate": 6.905702992659515e-06,
-      "loss": 0.843,
+      "grad_norm": 4.142574787139893,
+      "learning_rate": 0.00011509504987765858,
+      "loss": 0.6876,
       "step": 20450
     },
     {
       "epoch": 3.8509316770186337,
-      "grad_norm": 16.11570167541504,
-      "learning_rate": 6.894409937888198e-06,
-      "loss": 0.5118,
+      "grad_norm": 3.3246145248413086,
+      "learning_rate": 0.00011490683229813665,
+      "loss": 0.7418,
       "step": 20460
     },
     {
       "epoch": 3.8528138528138527,
-      "grad_norm": 9.326797485351562,
-      "learning_rate": 6.8831168831168835e-06,
-      "loss": 0.5509,
+      "grad_norm": 1.9157615900039673,
+      "learning_rate": 0.00011471861471861472,
+      "loss": 0.6626,
       "step": 20470
     },
     {
       "epoch": 3.854696028609072,
-      "grad_norm": 12.51727294921875,
-      "learning_rate": 6.871823828345567e-06,
-      "loss": 0.5935,
+      "grad_norm": 2.236476182937622,
+      "learning_rate": 0.00011453039713909279,
+      "loss": 0.6926,
       "step": 20480
     },
     {
       "epoch": 3.8565782044042916,
-      "grad_norm": 18.62541389465332,
-      "learning_rate": 6.860530773574252e-06,
-      "loss": 0.7294,
+      "grad_norm": 2.5907342433929443,
+      "learning_rate": 0.00011434217955957086,
+      "loss": 0.8657,
       "step": 20490
     },
     {
       "epoch": 3.8584603801995105,
-      "grad_norm": 18.53264045715332,
-      "learning_rate": 6.849237718802937e-06,
-      "loss": 0.4357,
+      "grad_norm": 3.643303871154785,
+      "learning_rate": 0.00011415396198004894,
+      "loss": 0.6398,
       "step": 20500
     },
     {
       "epoch": 3.86034255599473,
-      "grad_norm": 22.299819946289062,
-      "learning_rate": 6.837944664031621e-06,
-      "loss": 0.5943,
+      "grad_norm": 4.363615036010742,
+      "learning_rate": 0.00011396574440052702,
+      "loss": 0.7104,
       "step": 20510
     },
     {
       "epoch": 3.862224731789949,
-      "grad_norm": 23.825881958007812,
-      "learning_rate": 6.826651609260305e-06,
-      "loss": 0.5521,
+      "grad_norm": 3.6445014476776123,
+      "learning_rate": 0.00011377752682100509,
+      "loss": 0.7257,
       "step": 20520
     },
     {
       "epoch": 3.8641069075851684,
-      "grad_norm": 20.267616271972656,
-      "learning_rate": 6.815358554488989e-06,
-      "loss": 0.8127,
+      "grad_norm": 1.6341899633407593,
+      "learning_rate": 0.00011358930924148316,
+      "loss": 0.7366,
       "step": 20530
     },
     {
       "epoch": 3.865989083380388,
-      "grad_norm": 29.187458038330078,
-      "learning_rate": 6.804065499717674e-06,
-      "loss": 0.5979,
+      "grad_norm": 3.796801805496216,
+      "learning_rate": 0.00011340109166196123,
+      "loss": 0.6457,
       "step": 20540
     },
     {
       "epoch": 3.8678712591756073,
-      "grad_norm": 15.423782348632812,
-      "learning_rate": 6.792772444946358e-06,
-      "loss": 0.6075,
+      "grad_norm": 2.6889047622680664,
+      "learning_rate": 0.0001132128740824393,
+      "loss": 0.5895,
       "step": 20550
     },
     {
       "epoch": 3.8697534349708262,
-      "grad_norm": 8.876800537109375,
-      "learning_rate": 6.781479390175043e-06,
-      "loss": 0.8203,
+      "grad_norm": 1.7333699464797974,
+      "learning_rate": 0.00011302465650291738,
+      "loss": 1.041,
       "step": 20560
     },
     {
       "epoch": 3.8716356107660457,
-      "grad_norm": 0.373295396566391,
-      "learning_rate": 6.7701863354037265e-06,
-      "loss": 0.7442,
+      "grad_norm": 0.7206604480743408,
+      "learning_rate": 0.00011283643892339545,
+      "loss": 0.7296,
       "step": 20570
     },
     {
       "epoch": 3.8735177865612647,
-      "grad_norm": 3.6619198322296143,
-      "learning_rate": 6.758893280632412e-06,
-      "loss": 0.2888,
+      "grad_norm": 0.6784987449645996,
+      "learning_rate": 0.00011264822134387352,
+      "loss": 0.3055,
       "step": 20580
     },
     {
       "epoch": 3.875399962356484,
-      "grad_norm": 7.333616256713867,
-      "learning_rate": 6.747600225861095e-06,
-      "loss": 0.5505,
+      "grad_norm": 3.1988539695739746,
+      "learning_rate": 0.00011246000376435159,
+      "loss": 0.7054,
       "step": 20590
     },
     {
       "epoch": 3.8772821381517035,
-      "grad_norm": 11.97646427154541,
-      "learning_rate": 6.73630717108978e-06,
-      "loss": 0.5715,
+      "grad_norm": 2.1100261211395264,
+      "learning_rate": 0.00011227178618482966,
+      "loss": 0.6783,
       "step": 20600
     },
     {
       "epoch": 3.8791643139469225,
-      "grad_norm": 17.42437744140625,
-      "learning_rate": 6.725014116318464e-06,
-      "loss": 0.7322,
+      "grad_norm": 3.418389320373535,
+      "learning_rate": 0.00011208356860530773,
+      "loss": 0.8839,
       "step": 20610
     },
     {
       "epoch": 3.881046489742142,
-      "grad_norm": 4.155119895935059,
-      "learning_rate": 6.713721061547148e-06,
-      "loss": 0.5882,
+      "grad_norm": 1.9344919919967651,
+      "learning_rate": 0.00011189535102578582,
+      "loss": 0.603,
       "step": 20620
     },
     {
       "epoch": 3.882928665537361,
-      "grad_norm": 20.934206008911133,
-      "learning_rate": 6.702428006775834e-06,
-      "loss": 0.7397,
+      "grad_norm": 4.261674880981445,
+      "learning_rate": 0.00011170713344626389,
+      "loss": 0.794,
       "step": 20630
     },
     {
       "epoch": 3.8848108413325804,
-      "grad_norm": 12.857183456420898,
-      "learning_rate": 6.691134952004517e-06,
-      "loss": 0.4704,
+      "grad_norm": 2.8918564319610596,
+      "learning_rate": 0.00011151891586674196,
+      "loss": 0.4593,
       "step": 20640
     },
     {
       "epoch": 3.8866930171278,
-      "grad_norm": 8.88806438446045,
-      "learning_rate": 6.679841897233202e-06,
-      "loss": 0.3095,
+      "grad_norm": 1.8943965435028076,
+      "learning_rate": 0.00011133069828722003,
+      "loss": 0.4184,
       "step": 20650
     },
     {
       "epoch": 3.8885751929230192,
-      "grad_norm": 11.002625465393066,
-      "learning_rate": 6.668548842461886e-06,
-      "loss": 0.5045,
+      "grad_norm": 3.679342269897461,
+      "learning_rate": 0.0001111424807076981,
+      "loss": 0.5149,
       "step": 20660
     },
     {
       "epoch": 3.890457368718238,
-      "grad_norm": 7.77680778503418,
-      "learning_rate": 6.65725578769057e-06,
-      "loss": 0.8244,
+      "grad_norm": 3.0474326610565186,
+      "learning_rate": 0.00011095426312817617,
+      "loss": 0.8624,
       "step": 20670
     },
     {
       "epoch": 3.8923395445134576,
-      "grad_norm": 0.3390113115310669,
-      "learning_rate": 6.645962732919255e-06,
-      "loss": 0.5562,
+      "grad_norm": 0.4835365116596222,
+      "learning_rate": 0.00011076604554865425,
+      "loss": 0.5949,
       "step": 20680
     },
     {
       "epoch": 3.8942217203086766,
-      "grad_norm": 19.375041961669922,
-      "learning_rate": 6.634669678147939e-06,
-      "loss": 0.4,
+      "grad_norm": 4.019068241119385,
+      "learning_rate": 0.00011057782796913232,
+      "loss": 0.4482,
       "step": 20690
     },
     {
       "epoch": 3.896103896103896,
-      "grad_norm": 47.01906967163086,
-      "learning_rate": 6.6233766233766234e-06,
-      "loss": 0.4626,
+      "grad_norm": 3.6996610164642334,
+      "learning_rate": 0.00011038961038961039,
+      "loss": 0.5172,
       "step": 20700
     },
     {
       "epoch": 3.8979860718991155,
-      "grad_norm": 12.001082420349121,
-      "learning_rate": 6.612083568605308e-06,
-      "loss": 0.7142,
+      "grad_norm": 2.3082823753356934,
+      "learning_rate": 0.00011020139281008846,
+      "loss": 0.6819,
       "step": 20710
     },
     {
       "epoch": 3.8998682476943345,
-      "grad_norm": 12.084139823913574,
-      "learning_rate": 6.600790513833992e-06,
-      "loss": 0.4786,
+      "grad_norm": 3.6274797916412354,
+      "learning_rate": 0.00011001317523056653,
+      "loss": 0.6246,
       "step": 20720
     },
     {
       "epoch": 3.901750423489554,
-      "grad_norm": 10.616601943969727,
-      "learning_rate": 6.589497459062677e-06,
-      "loss": 0.4461,
+      "grad_norm": 2.167787790298462,
+      "learning_rate": 0.00010982495765104462,
+      "loss": 0.5745,
       "step": 20730
     },
     {
       "epoch": 3.9036325992847734,
-      "grad_norm": 15.530953407287598,
-      "learning_rate": 6.578204404291361e-06,
-      "loss": 0.6108,
+      "grad_norm": 3.232084274291992,
+      "learning_rate": 0.00010963674007152269,
+      "loss": 0.6898,
       "step": 20740
     },
     {
       "epoch": 3.9055147750799923,
-      "grad_norm": 0.3299574851989746,
-      "learning_rate": 6.566911349520045e-06,
-      "loss": 0.5048,
+      "grad_norm": 1.4799513816833496,
+      "learning_rate": 0.00010944852249200076,
+      "loss": 0.5645,
       "step": 20750
     },
     {
       "epoch": 3.9073969508752118,
-      "grad_norm": 0.755125880241394,
-      "learning_rate": 6.55561829474873e-06,
-      "loss": 0.3229,
+      "grad_norm": 0.08950027078390121,
+      "learning_rate": 0.00010926030491247883,
+      "loss": 0.3951,
       "step": 20760
     },
     {
       "epoch": 3.909279126670431,
-      "grad_norm": 0.48004603385925293,
-      "learning_rate": 6.544325239977414e-06,
-      "loss": 0.4146,
+      "grad_norm": 0.2402801364660263,
+      "learning_rate": 0.0001090720873329569,
+      "loss": 0.3863,
       "step": 20770
     },
     {
       "epoch": 3.91116130246565,
-      "grad_norm": 33.05055236816406,
-      "learning_rate": 6.5330321852060985e-06,
-      "loss": 0.4969,
+      "grad_norm": 5.756802558898926,
+      "learning_rate": 0.00010888386975343497,
+      "loss": 0.5367,
       "step": 20780
     },
     {
       "epoch": 3.9130434782608696,
-      "grad_norm": 2.9246139526367188,
-      "learning_rate": 6.521739130434783e-06,
-      "loss": 1.0341,
+      "grad_norm": 2.1738858222961426,
+      "learning_rate": 0.00010869565217391305,
+      "loss": 0.9819,
       "step": 20790
     },
     {
       "epoch": 3.9149256540560886,
-      "grad_norm": 8.560576438903809,
-      "learning_rate": 6.510446075663467e-06,
-      "loss": 0.4784,
+      "grad_norm": 1.9287985563278198,
+      "learning_rate": 0.00010850743459439112,
+      "loss": 0.5903,
       "step": 20800
     },
     {
       "epoch": 3.916807829851308,
-      "grad_norm": 14.475516319274902,
-      "learning_rate": 6.499153020892152e-06,
-      "loss": 0.422,
+      "grad_norm": 2.640705108642578,
+      "learning_rate": 0.00010831921701486919,
+      "loss": 0.4065,
       "step": 20810
     },
     {
       "epoch": 3.9186900056465275,
-      "grad_norm": 23.599973678588867,
-      "learning_rate": 6.487859966120836e-06,
-      "loss": 0.7579,
+      "grad_norm": 5.2311296463012695,
+      "learning_rate": 0.00010813099943534726,
+      "loss": 0.8136,
       "step": 20820
     },
     {
       "epoch": 3.920572181441747,
-      "grad_norm": 21.075658798217773,
-      "learning_rate": 6.4765669113495196e-06,
-      "loss": 0.5062,
+      "grad_norm": 1.2336487770080566,
+      "learning_rate": 0.00010794278185582533,
+      "loss": 0.5395,
       "step": 20830
     },
     {
       "epoch": 3.922454357236966,
-      "grad_norm": 11.33735466003418,
-      "learning_rate": 6.465273856578205e-06,
-      "loss": 0.7134,
+      "grad_norm": 2.3239247798919678,
+      "learning_rate": 0.00010775456427630342,
+      "loss": 0.7615,
       "step": 20840
     },
     {
       "epoch": 3.9243365330321853,
-      "grad_norm": 13.300880432128906,
-      "learning_rate": 6.453980801806888e-06,
-      "loss": 0.4887,
+      "grad_norm": 2.2713286876678467,
+      "learning_rate": 0.00010756634669678148,
+      "loss": 0.6509,
       "step": 20850
     },
     {
       "epoch": 3.9262187088274043,
-      "grad_norm": 16.807415008544922,
-      "learning_rate": 6.4426877470355736e-06,
-      "loss": 0.3613,
+      "grad_norm": 2.7799088954925537,
+      "learning_rate": 0.00010737812911725956,
+      "loss": 0.3079,
       "step": 20860
     },
     {
       "epoch": 3.9281008846226237,
-      "grad_norm": 10.84011459350586,
-      "learning_rate": 6.431394692264258e-06,
-      "loss": 0.5069,
+      "grad_norm": 4.758141994476318,
+      "learning_rate": 0.00010718991153773763,
+      "loss": 0.5902,
       "step": 20870
     },
     {
       "epoch": 3.929983060417843,
-      "grad_norm": 21.17901611328125,
-      "learning_rate": 6.420101637492942e-06,
-      "loss": 0.6276,
+      "grad_norm": 5.282011985778809,
+      "learning_rate": 0.0001070016939582157,
+      "loss": 0.5354,
       "step": 20880
     },
     {
       "epoch": 3.931865236213062,
-      "grad_norm": 12.811206817626953,
-      "learning_rate": 6.408808582721627e-06,
-      "loss": 0.5739,
+      "grad_norm": 2.415411949157715,
+      "learning_rate": 0.00010681347637869377,
+      "loss": 0.6325,
       "step": 20890
     },
     {
       "epoch": 3.9337474120082816,
-      "grad_norm": 2.8694605827331543,
-      "learning_rate": 6.39751552795031e-06,
-      "loss": 0.3009,
+      "grad_norm": 0.12160457670688629,
+      "learning_rate": 0.00010662525879917185,
+      "loss": 0.3388,
       "step": 20900
     },
     {
       "epoch": 3.9356295878035006,
-      "grad_norm": 10.879847526550293,
-      "learning_rate": 6.3862224731789955e-06,
-      "loss": 0.5047,
+      "grad_norm": 4.200422286987305,
+      "learning_rate": 0.00010643704121964992,
+      "loss": 0.6201,
       "step": 20910
     },
     {
       "epoch": 3.93751176359872,
-      "grad_norm": 12.21065616607666,
-      "learning_rate": 6.374929418407679e-06,
-      "loss": 0.6257,
+      "grad_norm": 4.672911167144775,
+      "learning_rate": 0.00010624882364012799,
+      "loss": 0.5476,
       "step": 20920
     },
     {
       "epoch": 3.9393939393939394,
-      "grad_norm": 12.078612327575684,
-      "learning_rate": 6.363636363636364e-06,
-      "loss": 0.7487,
+      "grad_norm": 3.7594289779663086,
+      "learning_rate": 0.00010606060606060606,
+      "loss": 0.6746,
       "step": 20930
     },
     {
       "epoch": 3.941276115189159,
-      "grad_norm": 7.435342311859131,
-      "learning_rate": 6.352343308865048e-06,
-      "loss": 0.4637,
+      "grad_norm": 1.4969550371170044,
+      "learning_rate": 0.00010587238848108413,
+      "loss": 0.4174,
       "step": 20940
     },
     {
       "epoch": 3.943158290984378,
-      "grad_norm": 16.511627197265625,
-      "learning_rate": 6.341050254093733e-06,
-      "loss": 0.6262,
+      "grad_norm": 0.9225571155548096,
+      "learning_rate": 0.00010568417090156222,
+      "loss": 0.5711,
       "step": 20950
     },
     {
       "epoch": 3.9450404667795973,
-      "grad_norm": 0.3107624053955078,
-      "learning_rate": 6.3297571993224165e-06,
-      "loss": 0.5074,
+      "grad_norm": 4.92745304107666,
+      "learning_rate": 0.00010549595332204028,
+      "loss": 0.6358,
       "step": 20960
     },
     {
       "epoch": 3.9469226425748163,
-      "grad_norm": 0.1293262541294098,
-      "learning_rate": 6.318464144551101e-06,
-      "loss": 0.7009,
+      "grad_norm": 0.09020210057497025,
+      "learning_rate": 0.00010530773574251836,
+      "loss": 0.7014,
       "step": 20970
     },
     {
       "epoch": 3.9488048183700357,
-      "grad_norm": 16.236751556396484,
-      "learning_rate": 6.307171089779785e-06,
-      "loss": 0.9354,
+      "grad_norm": 4.352484226226807,
+      "learning_rate": 0.00010511951816299642,
+      "loss": 1.125,
       "step": 20980
     },
     {
       "epoch": 3.950686994165255,
-      "grad_norm": 4.352895259857178,
-      "learning_rate": 6.29587803500847e-06,
-      "loss": 0.7928,
+      "grad_norm": 1.2882801294326782,
+      "learning_rate": 0.0001049313005834745,
+      "loss": 0.7509,
       "step": 20990
     },
     {
       "epoch": 3.9525691699604746,
-      "grad_norm": 7.811060428619385,
-      "learning_rate": 6.284584980237155e-06,
-      "loss": 0.384,
+      "grad_norm": 2.5389184951782227,
+      "learning_rate": 0.00010474308300395257,
+      "loss": 0.5879,
       "step": 21000
     },
     {
       "epoch": 3.9544513457556936,
-      "grad_norm": 30.867467880249023,
-      "learning_rate": 6.2732919254658384e-06,
-      "loss": 0.4535,
+      "grad_norm": 4.555139541625977,
+      "learning_rate": 0.00010455486542443065,
+      "loss": 0.6353,
       "step": 21010
     },
     {
       "epoch": 3.956333521550913,
-      "grad_norm": 9.880651473999023,
-      "learning_rate": 6.261998870694524e-06,
-      "loss": 0.4897,
+      "grad_norm": 2.414466381072998,
+      "learning_rate": 0.00010436664784490872,
+      "loss": 0.5994,
       "step": 21020
     },
     {
       "epoch": 3.958215697346132,
-      "grad_norm": 26.8701114654541,
-      "learning_rate": 6.250705815923207e-06,
-      "loss": 0.5893,
+      "grad_norm": 2.828115701675415,
+      "learning_rate": 0.00010417843026538679,
+      "loss": 0.6584,
       "step": 21030
     },
     {
       "epoch": 3.9600978731413514,
-      "grad_norm": 21.494901657104492,
-      "learning_rate": 6.239412761151892e-06,
-      "loss": 0.5116,
+      "grad_norm": 4.323361873626709,
+      "learning_rate": 0.00010399021268586486,
+      "loss": 0.568,
       "step": 21040
     },
     {
       "epoch": 3.961980048936571,
-      "grad_norm": 13.416314125061035,
-      "learning_rate": 6.228119706380576e-06,
-      "loss": 0.6689,
+      "grad_norm": 1.821200966835022,
+      "learning_rate": 0.00010380199510634293,
+      "loss": 0.6949,
       "step": 21050
     },
     {
       "epoch": 3.96386222473179,
-      "grad_norm": 7.51889705657959,
-      "learning_rate": 6.21682665160926e-06,
-      "loss": 0.4839,
+      "grad_norm": 7.879891395568848,
+      "learning_rate": 0.00010361377752682102,
+      "loss": 0.5554,
       "step": 21060
     },
     {
       "epoch": 3.9657444005270093,
-      "grad_norm": 12.946698188781738,
-      "learning_rate": 6.205533596837945e-06,
-      "loss": 0.6613,
+      "grad_norm": 2.9675745964050293,
+      "learning_rate": 0.00010342555994729908,
+      "loss": 0.5815,
       "step": 21070
     },
     {
       "epoch": 3.9676265763222283,
-      "grad_norm": 18.94002342224121,
-      "learning_rate": 6.194240542066629e-06,
-      "loss": 0.607,
+      "grad_norm": 3.8450276851654053,
+      "learning_rate": 0.00010323734236777716,
+      "loss": 0.734,
       "step": 21080
     },
     {
       "epoch": 3.9695087521174477,
-      "grad_norm": 13.655399322509766,
-      "learning_rate": 6.1829474872953135e-06,
-      "loss": 0.6801,
+      "grad_norm": 3.7981014251708984,
+      "learning_rate": 0.00010304912478825522,
+      "loss": 0.7174,
       "step": 21090
     },
     {
       "epoch": 3.971390927912667,
-      "grad_norm": 12.325647354125977,
-      "learning_rate": 6.171654432523998e-06,
-      "loss": 0.6112,
+      "grad_norm": 3.7332234382629395,
+      "learning_rate": 0.0001028609072087333,
+      "loss": 0.5115,
       "step": 21100
     },
     {
       "epoch": 3.9732731037078866,
-      "grad_norm": 6.132079601287842,
-      "learning_rate": 6.160361377752682e-06,
-      "loss": 0.6071,
+      "grad_norm": 1.9932327270507812,
+      "learning_rate": 0.00010267268962921137,
+      "loss": 0.7141,
       "step": 21110
     },
     {
       "epoch": 3.9751552795031055,
-      "grad_norm": 0.7968364357948303,
-      "learning_rate": 6.149068322981367e-06,
-      "loss": 0.5114,
+      "grad_norm": 1.1441056728363037,
+      "learning_rate": 0.00010248447204968945,
+      "loss": 0.5968,
       "step": 21120
     },
     {
       "epoch": 3.977037455298325,
-      "grad_norm": 5.779326438903809,
-      "learning_rate": 6.137775268210051e-06,
-      "loss": 0.6626,
+      "grad_norm": 1.9460235834121704,
+      "learning_rate": 0.00010229625447016752,
+      "loss": 0.6997,
       "step": 21130
     },
     {
       "epoch": 3.978919631093544,
-      "grad_norm": 15.42678451538086,
-      "learning_rate": 6.126482213438735e-06,
-      "loss": 0.6293,
+      "grad_norm": 2.354952096939087,
+      "learning_rate": 0.00010210803689064559,
+      "loss": 0.5117,
       "step": 21140
     },
     {
       "epoch": 3.9808018068887634,
-      "grad_norm": 23.818042755126953,
-      "learning_rate": 6.11518915866742e-06,
-      "loss": 0.7765,
+      "grad_norm": 3.873466968536377,
+      "learning_rate": 0.00010191981931112366,
+      "loss": 0.7877,
       "step": 21150
     },
     {
       "epoch": 3.982683982683983,
-      "grad_norm": 15.091184616088867,
-      "learning_rate": 6.103896103896104e-06,
-      "loss": 0.6724,
+      "grad_norm": 5.34775972366333,
+      "learning_rate": 0.00010173160173160173,
+      "loss": 0.6326,
       "step": 21160
     },
     {
       "epoch": 3.984566158479202,
-      "grad_norm": 13.259024620056152,
-      "learning_rate": 6.0926030491247886e-06,
-      "loss": 0.6242,
+      "grad_norm": 2.9042680263519287,
+      "learning_rate": 0.00010154338415207982,
+      "loss": 0.762,
       "step": 21170
     },
     {
       "epoch": 3.9864483342744212,
-      "grad_norm": 10.13891887664795,
-      "learning_rate": 6.081309994353472e-06,
-      "loss": 0.807,
+      "grad_norm": 4.9012675285339355,
+      "learning_rate": 0.00010135516657255788,
+      "loss": 0.9534,
       "step": 21180
     },
     {
       "epoch": 3.9883305100696402,
-      "grad_norm": 10.566964149475098,
-      "learning_rate": 6.070016939582157e-06,
-      "loss": 0.4579,
+      "grad_norm": 3.0978500843048096,
+      "learning_rate": 0.00010116694899303596,
+      "loss": 0.5921,
       "step": 21190
     },
     {
       "epoch": 3.9902126858648597,
-      "grad_norm": 9.14899730682373,
-      "learning_rate": 6.058723884810841e-06,
-      "loss": 0.5235,
+      "grad_norm": 3.7232377529144287,
+      "learning_rate": 0.00010097873141351402,
+      "loss": 0.4836,
       "step": 21200
     },
     {
       "epoch": 3.992094861660079,
-      "grad_norm": 2.9169421195983887,
-      "learning_rate": 6.047430830039526e-06,
-      "loss": 0.3648,
+      "grad_norm": 2.5855820178985596,
+      "learning_rate": 0.0001007905138339921,
+      "loss": 0.478,
       "step": 21210
     },
     {
       "epoch": 3.9939770374552985,
-      "grad_norm": 14.589804649353027,
-      "learning_rate": 6.03613777526821e-06,
-      "loss": 0.4533,
+      "grad_norm": 2.2673332691192627,
+      "learning_rate": 0.00010060229625447016,
+      "loss": 0.52,
       "step": 21220
     },
     {
       "epoch": 3.9958592132505175,
-      "grad_norm": 24.296232223510742,
-      "learning_rate": 6.024844720496895e-06,
-      "loss": 0.5101,
+      "grad_norm": 3.1919641494750977,
+      "learning_rate": 0.00010041407867494825,
+      "loss": 0.4867,
       "step": 21230
     },
     {
       "epoch": 3.997741389045737,
-      "grad_norm": 33.149715423583984,
-      "learning_rate": 6.013551665725579e-06,
-      "loss": 0.6524,
+      "grad_norm": 5.216162204742432,
+      "learning_rate": 0.00010022586109542632,
+      "loss": 0.7683,
       "step": 21240
     },
     {
       "epoch": 3.999623564840956,
-      "grad_norm": 15.509395599365234,
-      "learning_rate": 6.002258610954264e-06,
-      "loss": 0.3432,
+      "grad_norm": 3.5101685523986816,
+      "learning_rate": 0.00010003764351590439,
+      "loss": 0.3582,
       "step": 21250
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.9209333333333334,
-      "eval_loss": 0.2990027368068695,
-      "eval_runtime": 253.833,
-      "eval_samples_per_second": 29.547,
-      "eval_steps_per_second": 3.695,
+      "eval_accuracy": 0.9232,
+      "eval_loss": 0.25806137919425964,
+      "eval_runtime": 293.9532,
+      "eval_samples_per_second": 25.514,
+      "eval_steps_per_second": 3.191,
       "step": 21252
     },
     {
       "epoch": 4.001505740636175,
-      "grad_norm": 10.212028503417969,
-      "learning_rate": 5.990965556182948e-06,
-      "loss": 0.751,
+      "grad_norm": 2.5859482288360596,
+      "learning_rate": 9.984942593638246e-05,
+      "loss": 0.6965,
       "step": 21260
     },
     {
       "epoch": 4.003387916431395,
-      "grad_norm": 12.337179183959961,
-      "learning_rate": 5.9796725014116315e-06,
-      "loss": 0.4736,
+      "grad_norm": 2.421527862548828,
+      "learning_rate": 9.966120835686053e-05,
+      "loss": 0.5074,
       "step": 21270
     },
     {
       "epoch": 4.005270092226614,
-      "grad_norm": 1.4147861003875732,
-      "learning_rate": 5.968379446640317e-06,
-      "loss": 0.3634,
+      "grad_norm": 0.755755603313446,
+      "learning_rate": 9.947299077733862e-05,
+      "loss": 0.3855,
       "step": 21280
     },
     {
       "epoch": 4.007152268021834,
-      "grad_norm": 21.909902572631836,
-      "learning_rate": 5.957086391869e-06,
-      "loss": 0.5764,
+      "grad_norm": 1.626813292503357,
+      "learning_rate": 9.928477319781668e-05,
+      "loss": 0.6369,
       "step": 21290
     },
     {
       "epoch": 4.009034443817052,
-      "grad_norm": 8.32170295715332,
-      "learning_rate": 5.9457933370976855e-06,
-      "loss": 0.5818,
+      "grad_norm": 1.9735829830169678,
+      "learning_rate": 9.909655561829476e-05,
+      "loss": 0.6928,
       "step": 21300
     },
     {
       "epoch": 4.010916619612272,
-      "grad_norm": 4.794271945953369,
-      "learning_rate": 5.934500282326369e-06,
-      "loss": 0.7417,
+      "grad_norm": 2.7982254028320312,
+      "learning_rate": 9.890833803877282e-05,
+      "loss": 0.7617,
       "step": 21310
     },
     {
       "epoch": 4.012798795407491,
-      "grad_norm": 25.876705169677734,
-      "learning_rate": 5.923207227555054e-06,
-      "loss": 0.4546,
+      "grad_norm": 3.808712959289551,
+      "learning_rate": 9.87201204592509e-05,
+      "loss": 0.5099,
       "step": 21320
     },
     {
       "epoch": 4.0146809712027105,
-      "grad_norm": 26.065282821655273,
-      "learning_rate": 5.911914172783738e-06,
-      "loss": 0.7006,
+      "grad_norm": 4.210937976837158,
+      "learning_rate": 9.853190287972896e-05,
+      "loss": 0.7595,
       "step": 21330
     },
     {
       "epoch": 4.01656314699793,
-      "grad_norm": 34.19929122924805,
-      "learning_rate": 5.900621118012422e-06,
-      "loss": 0.6222,
+      "grad_norm": 8.570923805236816,
+      "learning_rate": 9.834368530020705e-05,
+      "loss": 0.6814,
       "step": 21340
     },
     {
       "epoch": 4.0184453227931485,
-      "grad_norm": 1.1643229722976685,
-      "learning_rate": 5.889328063241107e-06,
-      "loss": 0.4165,
+      "grad_norm": 1.556383728981018,
+      "learning_rate": 9.815546772068511e-05,
+      "loss": 0.4653,
       "step": 21350
     },
     {
       "epoch": 4.020327498588368,
-      "grad_norm": 10.392925262451172,
-      "learning_rate": 5.878035008469791e-06,
-      "loss": 0.5281,
+      "grad_norm": 2.515901803970337,
+      "learning_rate": 9.796725014116319e-05,
+      "loss": 0.5543,
       "step": 21360
     },
     {
       "epoch": 4.022209674383587,
-      "grad_norm": 25.98088264465332,
-      "learning_rate": 5.866741953698476e-06,
-      "loss": 0.4653,
+      "grad_norm": 1.1342847347259521,
+      "learning_rate": 9.777903256164126e-05,
+      "loss": 0.577,
       "step": 21370
     },
     {
       "epoch": 4.024091850178807,
-      "grad_norm": 19.15206527709961,
-      "learning_rate": 5.85544889892716e-06,
-      "loss": 0.6925,
+      "grad_norm": 2.5764143466949463,
+      "learning_rate": 9.759081498211933e-05,
+      "loss": 0.7136,
       "step": 21380
     },
     {
       "epoch": 4.025974025974026,
-      "grad_norm": 8.375141143798828,
-      "learning_rate": 5.844155844155845e-06,
-      "loss": 0.975,
+      "grad_norm": 3.2266030311584473,
+      "learning_rate": 9.740259740259742e-05,
+      "loss": 0.9059,
       "step": 21390
     },
     {
       "epoch": 4.027856201769246,
-      "grad_norm": 11.092159271240234,
-      "learning_rate": 5.8328627893845285e-06,
-      "loss": 0.4277,
+      "grad_norm": 2.965477466583252,
+      "learning_rate": 9.721437982307548e-05,
+      "loss": 0.537,
       "step": 21400
     },
     {
       "epoch": 4.029738377564464,
-      "grad_norm": 17.455846786499023,
-      "learning_rate": 5.821569734613213e-06,
-      "loss": 0.7711,
+      "grad_norm": 2.400604248046875,
+      "learning_rate": 9.702616224355356e-05,
+      "loss": 0.7048,
       "step": 21410
     },
     {
       "epoch": 4.031620553359684,
-      "grad_norm": 5.300505638122559,
-      "learning_rate": 5.810276679841897e-06,
-      "loss": 0.6583,
+      "grad_norm": 2.111332893371582,
+      "learning_rate": 9.683794466403162e-05,
+      "loss": 0.7618,
       "step": 21420
     },
     {
       "epoch": 4.033502729154903,
-      "grad_norm": 0.19853296875953674,
-      "learning_rate": 5.798983625070582e-06,
-      "loss": 0.5738,
+      "grad_norm": 0.10220424830913544,
+      "learning_rate": 9.66497270845097e-05,
+      "loss": 0.6055,
       "step": 21430
     },
     {
       "epoch": 4.0353849049501225,
-      "grad_norm": 19.96097183227539,
-      "learning_rate": 5.787690570299266e-06,
-      "loss": 0.3168,
+      "grad_norm": 6.471917152404785,
+      "learning_rate": 9.646150950498776e-05,
+      "loss": 0.4042,
       "step": 21440
     },
     {
       "epoch": 4.037267080745342,
-      "grad_norm": 10.578010559082031,
-      "learning_rate": 5.77639751552795e-06,
-      "loss": 0.3206,
+      "grad_norm": 2.8826186656951904,
+      "learning_rate": 9.627329192546585e-05,
+      "loss": 0.442,
       "step": 21450
     },
     {
       "epoch": 4.0391492565405605,
-      "grad_norm": 7.319991111755371,
-      "learning_rate": 5.765104460756635e-06,
-      "loss": 0.6817,
+      "grad_norm": 1.8660509586334229,
+      "learning_rate": 9.608507434594391e-05,
+      "loss": 0.5408,
       "step": 21460
     },
     {
       "epoch": 4.04103143233578,
-      "grad_norm": 1.0113133192062378,
-      "learning_rate": 5.753811405985319e-06,
-      "loss": 0.4977,
+      "grad_norm": 0.3903960585594177,
+      "learning_rate": 9.589685676642199e-05,
+      "loss": 0.5129,
       "step": 21470
     },
     {
       "epoch": 4.042913608130999,
-      "grad_norm": 0.2078145146369934,
-      "learning_rate": 5.7425183512140036e-06,
-      "loss": 0.3547,
+      "grad_norm": 4.441036701202393,
+      "learning_rate": 9.570863918690006e-05,
+      "loss": 0.3228,
       "step": 21480
     },
     {
       "epoch": 4.044795783926219,
-      "grad_norm": 19.170316696166992,
-      "learning_rate": 5.731225296442688e-06,
-      "loss": 0.3471,
+      "grad_norm": 3.8729546070098877,
+      "learning_rate": 9.552042160737813e-05,
+      "loss": 0.45,
       "step": 21490
     },
     {
       "epoch": 4.046677959721438,
-      "grad_norm": 20.73012924194336,
-      "learning_rate": 5.719932241671372e-06,
-      "loss": 0.8461,
+      "grad_norm": 3.6702287197113037,
+      "learning_rate": 9.53322040278562e-05,
+      "loss": 1.0155,
       "step": 21500
     },
     {
       "epoch": 4.048560135516658,
-      "grad_norm": 33.84339141845703,
-      "learning_rate": 5.708639186900057e-06,
-      "loss": 0.406,
+      "grad_norm": 3.694718360900879,
+      "learning_rate": 9.514398644833428e-05,
+      "loss": 0.4059,
       "step": 21510
     },
     {
       "epoch": 4.050442311311876,
-      "grad_norm": 0.295187771320343,
-      "learning_rate": 5.697346132128741e-06,
-      "loss": 0.2405,
+      "grad_norm": 1.9021106958389282,
+      "learning_rate": 9.495576886881236e-05,
+      "loss": 0.3598,
       "step": 21520
     },
     {
       "epoch": 4.052324487107096,
-      "grad_norm": 8.998740196228027,
-      "learning_rate": 5.6860530773574255e-06,
-      "loss": 0.4896,
+      "grad_norm": 3.583451986312866,
+      "learning_rate": 9.476755128929042e-05,
+      "loss": 0.3496,
       "step": 21530
     },
     {
       "epoch": 4.054206662902315,
-      "grad_norm": 2.618791103363037,
-      "learning_rate": 5.67476002258611e-06,
-      "loss": 0.3719,
+      "grad_norm": 2.1199567317962646,
+      "learning_rate": 9.45793337097685e-05,
+      "loss": 0.4274,
       "step": 21540
     },
     {
       "epoch": 4.0560888386975344,
-      "grad_norm": 12.432368278503418,
-      "learning_rate": 5.663466967814793e-06,
-      "loss": 0.9172,
+      "grad_norm": 1.4874187707901,
+      "learning_rate": 9.439111613024656e-05,
+      "loss": 0.9868,
       "step": 21550
     },
     {
       "epoch": 4.057971014492754,
-      "grad_norm": 22.641036987304688,
-      "learning_rate": 5.652173913043479e-06,
-      "loss": 0.498,
+      "grad_norm": 2.7171809673309326,
+      "learning_rate": 9.420289855072465e-05,
+      "loss": 0.4751,
       "step": 21560
     },
     {
       "epoch": 4.059853190287973,
-      "grad_norm": 1.2676461935043335,
-      "learning_rate": 5.640880858272162e-06,
-      "loss": 0.4709,
+      "grad_norm": 0.14315591752529144,
+      "learning_rate": 9.401468097120271e-05,
+      "loss": 0.3579,
       "step": 21570
     },
     {
       "epoch": 4.061735366083192,
-      "grad_norm": 8.317973136901855,
-      "learning_rate": 5.629587803500847e-06,
-      "loss": 0.463,
+      "grad_norm": 1.0544301271438599,
+      "learning_rate": 9.382646339168079e-05,
+      "loss": 0.4521,
       "step": 21580
     },
     {
       "epoch": 4.063617541878411,
-      "grad_norm": 12.644018173217773,
-      "learning_rate": 5.618294748729531e-06,
-      "loss": 0.4324,
+      "grad_norm": 3.1227588653564453,
+      "learning_rate": 9.363824581215885e-05,
+      "loss": 0.6073,
       "step": 21590
     },
     {
       "epoch": 4.065499717673631,
-      "grad_norm": 6.407628536224365,
-      "learning_rate": 5.607001693958216e-06,
-      "loss": 0.256,
+      "grad_norm": 2.726243019104004,
+      "learning_rate": 9.345002823263693e-05,
+      "loss": 0.3532,
       "step": 21600
     },
     {
       "epoch": 4.06738189346885,
-      "grad_norm": 0.3914599120616913,
-      "learning_rate": 5.5957086391869005e-06,
-      "loss": 0.6861,
+      "grad_norm": 2.325624942779541,
+      "learning_rate": 9.3261810653115e-05,
+      "loss": 0.7147,
       "step": 21610
     },
     {
       "epoch": 4.06926406926407,
-      "grad_norm": 9.922120094299316,
-      "learning_rate": 5.584415584415584e-06,
-      "loss": 0.4863,
+      "grad_norm": 2.536409854888916,
+      "learning_rate": 9.307359307359308e-05,
+      "loss": 0.5313,
       "step": 21620
     },
     {
       "epoch": 4.071146245059288,
-      "grad_norm": 0.7256487011909485,
-      "learning_rate": 5.573122529644269e-06,
-      "loss": 0.585,
+      "grad_norm": 0.09779352694749832,
+      "learning_rate": 9.288537549407116e-05,
+      "loss": 0.7419,
       "step": 21630
     },
     {
       "epoch": 4.073028420854508,
-      "grad_norm": 10.62320613861084,
-      "learning_rate": 5.561829474872953e-06,
-      "loss": 0.43,
+      "grad_norm": 1.094053864479065,
+      "learning_rate": 9.269715791454922e-05,
+      "loss": 0.4456,
       "step": 21640
     },
     {
       "epoch": 4.074910596649727,
-      "grad_norm": 2.3755974769592285,
-      "learning_rate": 5.550536420101638e-06,
-      "loss": 0.4797,
+      "grad_norm": 2.9910216331481934,
+      "learning_rate": 9.25089403350273e-05,
+      "loss": 0.4874,
       "step": 21650
     },
     {
       "epoch": 4.076792772444946,
-      "grad_norm": 0.20227846503257751,
-      "learning_rate": 5.539243365330322e-06,
-      "loss": 0.4626,
+      "grad_norm": 0.17736339569091797,
+      "learning_rate": 9.232072275550536e-05,
+      "loss": 0.3904,
       "step": 21660
     },
     {
       "epoch": 4.078674948240166,
-      "grad_norm": 12.040499687194824,
-      "learning_rate": 5.527950310559007e-06,
-      "loss": 0.3619,
+      "grad_norm": 3.1419992446899414,
+      "learning_rate": 9.213250517598345e-05,
+      "loss": 0.3972,
       "step": 21670
     },
     {
       "epoch": 4.080557124035385,
-      "grad_norm": 1.3578740358352661,
-      "learning_rate": 5.51665725578769e-06,
-      "loss": 0.4352,
+      "grad_norm": 2.997563600540161,
+      "learning_rate": 9.194428759646151e-05,
+      "loss": 0.468,
       "step": 21680
     },
     {
       "epoch": 4.082439299830604,
-      "grad_norm": 7.650432109832764,
-      "learning_rate": 5.505364201016376e-06,
-      "loss": 0.673,
+      "grad_norm": 2.441227912902832,
+      "learning_rate": 9.175607001693959e-05,
+      "loss": 0.7764,
       "step": 21690
     },
     {
       "epoch": 4.084321475625823,
-      "grad_norm": 7.1012983322143555,
-      "learning_rate": 5.494071146245059e-06,
-      "loss": 0.5103,
+      "grad_norm": 3.6427226066589355,
+      "learning_rate": 9.156785243741765e-05,
+      "loss": 0.5896,
       "step": 21700
     },
     {
       "epoch": 4.086203651421043,
-      "grad_norm": 28.039716720581055,
-      "learning_rate": 5.4827780914737435e-06,
-      "loss": 0.4092,
+      "grad_norm": 4.410248756408691,
+      "learning_rate": 9.137963485789573e-05,
+      "loss": 0.4323,
       "step": 21710
     },
     {
       "epoch": 4.088085827216262,
-      "grad_norm": 7.539644718170166,
-      "learning_rate": 5.471485036702429e-06,
-      "loss": 0.6337,
+      "grad_norm": 7.132479667663574,
+      "learning_rate": 9.11914172783738e-05,
+      "loss": 0.7631,
       "step": 21720
     },
     {
       "epoch": 4.089968003011482,
-      "grad_norm": 28.933773040771484,
-      "learning_rate": 5.460191981931112e-06,
-      "loss": 0.4827,
+      "grad_norm": 2.9803075790405273,
+      "learning_rate": 9.100319969885188e-05,
+      "loss": 0.567,
       "step": 21730
     },
     {
       "epoch": 4.0918501788067,
-      "grad_norm": 25.65105628967285,
-      "learning_rate": 5.4488989271597975e-06,
-      "loss": 0.5597,
+      "grad_norm": 2.718230724334717,
+      "learning_rate": 9.081498211932996e-05,
+      "loss": 0.606,
       "step": 21740
     },
     {
       "epoch": 4.0937323546019195,
-      "grad_norm": 27.194011688232422,
-      "learning_rate": 5.437605872388481e-06,
-      "loss": 0.5449,
+      "grad_norm": 8.262181282043457,
+      "learning_rate": 9.062676453980802e-05,
+      "loss": 0.6464,
       "step": 21750
     },
     {
       "epoch": 4.095614530397139,
-      "grad_norm": 11.716024398803711,
-      "learning_rate": 5.426312817617166e-06,
-      "loss": 0.6295,
+      "grad_norm": 3.146902084350586,
+      "learning_rate": 9.04385469602861e-05,
+      "loss": 0.6862,
       "step": 21760
     },
     {
       "epoch": 4.097496706192358,
-      "grad_norm": 17.159563064575195,
-      "learning_rate": 5.41501976284585e-06,
-      "loss": 0.4375,
+      "grad_norm": 2.2113897800445557,
+      "learning_rate": 9.025032938076416e-05,
+      "loss": 0.6772,
       "step": 21770
     },
     {
       "epoch": 4.099378881987578,
-      "grad_norm": 6.972395420074463,
-      "learning_rate": 5.403726708074534e-06,
-      "loss": 0.6678,
+      "grad_norm": 2.7351415157318115,
+      "learning_rate": 9.006211180124225e-05,
+      "loss": 0.6737,
       "step": 21780
     },
     {
       "epoch": 4.101261057782797,
-      "grad_norm": 2.577281951904297,
-      "learning_rate": 5.3924336533032186e-06,
-      "loss": 0.3085,
+      "grad_norm": 2.420198678970337,
+      "learning_rate": 8.987389422172031e-05,
+      "loss": 0.3986,
       "step": 21790
     },
     {
       "epoch": 4.103143233578016,
-      "grad_norm": 8.173493385314941,
-      "learning_rate": 5.381140598531903e-06,
-      "loss": 0.4501,
+      "grad_norm": 5.054831027984619,
+      "learning_rate": 8.968567664219839e-05,
+      "loss": 0.6978,
       "step": 21800
     },
     {
       "epoch": 4.105025409373235,
-      "grad_norm": 0.4988132417201996,
-      "learning_rate": 5.369847543760587e-06,
-      "loss": 0.4765,
+      "grad_norm": 0.5988843441009521,
+      "learning_rate": 8.949745906267645e-05,
+      "loss": 0.5788,
       "step": 21810
     },
     {
       "epoch": 4.106907585168455,
-      "grad_norm": 0.59678715467453,
-      "learning_rate": 5.358554488989272e-06,
-      "loss": 0.3523,
+      "grad_norm": 5.237829685211182,
+      "learning_rate": 8.930924148315453e-05,
+      "loss": 0.5006,
       "step": 21820
     },
     {
       "epoch": 4.108789760963674,
-      "grad_norm": 22.480344772338867,
-      "learning_rate": 5.347261434217956e-06,
-      "loss": 0.5372,
+      "grad_norm": 6.762528419494629,
+      "learning_rate": 8.912102390363259e-05,
+      "loss": 0.5774,
       "step": 21830
     },
     {
       "epoch": 4.1106719367588935,
-      "grad_norm": 6.5898284912109375,
-      "learning_rate": 5.3359683794466405e-06,
-      "loss": 0.6108,
+      "grad_norm": 2.084718704223633,
+      "learning_rate": 8.893280632411068e-05,
+      "loss": 0.6407,
       "step": 21840
     },
     {
       "epoch": 4.112554112554113,
-      "grad_norm": 11.716509819030762,
-      "learning_rate": 5.324675324675325e-06,
-      "loss": 0.3293,
+      "grad_norm": 1.660806655883789,
+      "learning_rate": 8.874458874458876e-05,
+      "loss": 0.4582,
       "step": 21850
     },
     {
       "epoch": 4.1144362883493315,
-      "grad_norm": 9.849950790405273,
-      "learning_rate": 5.313382269904009e-06,
-      "loss": 0.5861,
+      "grad_norm": 3.171679973602295,
+      "learning_rate": 8.855637116506682e-05,
+      "loss": 0.5192,
       "step": 21860
     },
     {
       "epoch": 4.116318464144551,
-      "grad_norm": 16.31063461303711,
-      "learning_rate": 5.302089215132694e-06,
-      "loss": 0.3767,
+      "grad_norm": 2.277590751647949,
+      "learning_rate": 8.83681535855449e-05,
+      "loss": 0.4837,
       "step": 21870
     },
     {
       "epoch": 4.11820063993977,
-      "grad_norm": 18.616395950317383,
-      "learning_rate": 5.290796160361378e-06,
-      "loss": 0.5778,
+      "grad_norm": 2.3669557571411133,
+      "learning_rate": 8.817993600602296e-05,
+      "loss": 0.5819,
       "step": 21880
     },
     {
       "epoch": 4.12008281573499,
-      "grad_norm": 12.037795066833496,
-      "learning_rate": 5.279503105590062e-06,
-      "loss": 0.5588,
+      "grad_norm": 2.413088798522949,
+      "learning_rate": 8.799171842650105e-05,
+      "loss": 0.5921,
       "step": 21890
     },
     {
       "epoch": 4.121964991530209,
-      "grad_norm": 5.515803337097168,
-      "learning_rate": 5.268210050818747e-06,
-      "loss": 0.2835,
+      "grad_norm": 5.965948104858398,
+      "learning_rate": 8.780350084697911e-05,
+      "loss": 0.3192,
       "step": 21900
     },
     {
       "epoch": 4.123847167325428,
-      "grad_norm": 8.707112312316895,
-      "learning_rate": 5.256916996047431e-06,
-      "loss": 0.3438,
+      "grad_norm": 1.5568970441818237,
+      "learning_rate": 8.761528326745719e-05,
+      "loss": 0.4662,
       "step": 21910
     },
     {
       "epoch": 4.125729343120647,
-      "grad_norm": 0.553016185760498,
-      "learning_rate": 5.245623941276115e-06,
-      "loss": 0.3274,
+      "grad_norm": 0.12782663106918335,
+      "learning_rate": 8.742706568793525e-05,
+      "loss": 0.4925,
       "step": 21920
     },
     {
       "epoch": 4.127611518915867,
-      "grad_norm": 14.9821195602417,
-      "learning_rate": 5.2343308865048e-06,
-      "loss": 0.5166,
+      "grad_norm": 1.7786810398101807,
+      "learning_rate": 8.723884810841333e-05,
+      "loss": 0.5588,
       "step": 21930
     },
     {
       "epoch": 4.129493694711086,
-      "grad_norm": 34.491355895996094,
-      "learning_rate": 5.2230378317334834e-06,
-      "loss": 0.4849,
+      "grad_norm": 6.479440212249756,
+      "learning_rate": 8.705063052889139e-05,
+      "loss": 0.5754,
       "step": 21940
     },
     {
       "epoch": 4.1313758705063055,
-      "grad_norm": 14.449700355529785,
-      "learning_rate": 5.211744776962169e-06,
-      "loss": 0.768,
+      "grad_norm": 2.8325908184051514,
+      "learning_rate": 8.686241294936948e-05,
+      "loss": 0.9421,
       "step": 21950
     },
     {
       "epoch": 4.133258046301525,
-      "grad_norm": 17.19442367553711,
-      "learning_rate": 5.200451722190852e-06,
-      "loss": 0.3727,
+      "grad_norm": 4.928275108337402,
+      "learning_rate": 8.667419536984754e-05,
+      "loss": 0.469,
       "step": 21960
     },
     {
       "epoch": 4.1351402220967435,
-      "grad_norm": 2.9677646160125732,
-      "learning_rate": 5.1891586674195374e-06,
-      "loss": 0.6443,
+      "grad_norm": 0.39288848638534546,
+      "learning_rate": 8.648597779032562e-05,
+      "loss": 0.5645,
       "step": 21970
     },
     {
       "epoch": 4.137022397891963,
-      "grad_norm": 8.971100807189941,
-      "learning_rate": 5.177865612648222e-06,
-      "loss": 0.744,
+      "grad_norm": 1.3426021337509155,
+      "learning_rate": 8.62977602108037e-05,
+      "loss": 0.822,
       "step": 21980
     },
     {
       "epoch": 4.138904573687182,
-      "grad_norm": 14.347002029418945,
-      "learning_rate": 5.166572557876905e-06,
-      "loss": 0.4241,
+      "grad_norm": 2.0728249549865723,
+      "learning_rate": 8.610954263128176e-05,
+      "loss": 0.5174,
       "step": 21990
     },
     {
       "epoch": 4.140786749482402,
-      "grad_norm": 3.654014825820923,
-      "learning_rate": 5.155279503105591e-06,
-      "loss": 0.6624,
+      "grad_norm": 2.072270631790161,
+      "learning_rate": 8.592132505175985e-05,
+      "loss": 0.5649,
       "step": 22000
     },
     {
       "epoch": 4.142668925277621,
-      "grad_norm": 18.717609405517578,
-      "learning_rate": 5.143986448334274e-06,
-      "loss": 0.6139,
+      "grad_norm": 3.1153640747070312,
+      "learning_rate": 8.573310747223791e-05,
+      "loss": 0.5106,
       "step": 22010
     },
     {
       "epoch": 4.144551101072841,
-      "grad_norm": 25.252634048461914,
-      "learning_rate": 5.132693393562959e-06,
-      "loss": 0.5933,
+      "grad_norm": 4.662202835083008,
+      "learning_rate": 8.554488989271599e-05,
+      "loss": 0.6303,
       "step": 22020
     },
     {
       "epoch": 4.146433276868059,
-      "grad_norm": 8.735464096069336,
-      "learning_rate": 5.121400338791643e-06,
-      "loss": 0.782,
+      "grad_norm": 3.654289484024048,
+      "learning_rate": 8.535667231319405e-05,
+      "loss": 0.8146,
       "step": 22030
     },
     {
       "epoch": 4.148315452663279,
-      "grad_norm": 2.7719712257385254,
-      "learning_rate": 5.110107284020328e-06,
-      "loss": 0.4075,
+      "grad_norm": 1.2205557823181152,
+      "learning_rate": 8.516845473367213e-05,
+      "loss": 0.4592,
       "step": 22040
     },
     {
       "epoch": 4.150197628458498,
-      "grad_norm": 14.126623153686523,
-      "learning_rate": 5.098814229249012e-06,
-      "loss": 0.4613,
+      "grad_norm": 2.4779789447784424,
+      "learning_rate": 8.498023715415019e-05,
+      "loss": 0.467,
       "step": 22050
     },
     {
       "epoch": 4.1520798042537175,
-      "grad_norm": 1.2759391069412231,
-      "learning_rate": 5.087521174477696e-06,
-      "loss": 0.7308,
+      "grad_norm": 0.1437557190656662,
+      "learning_rate": 8.479201957462828e-05,
+      "loss": 0.7689,
       "step": 22060
     },
     {
       "epoch": 4.153961980048937,
-      "grad_norm": 2.6802024841308594,
-      "learning_rate": 5.07622811970638e-06,
-      "loss": 0.3868,
+      "grad_norm": 1.2576069831848145,
+      "learning_rate": 8.460380199510634e-05,
+      "loss": 0.5398,
       "step": 22070
     },
     {
       "epoch": 4.1558441558441555,
-      "grad_norm": 20.169902801513672,
-      "learning_rate": 5.064935064935065e-06,
-      "loss": 0.6009,
+      "grad_norm": 6.528464317321777,
+      "learning_rate": 8.441558441558442e-05,
+      "loss": 0.6382,
       "step": 22080
     },
     {
       "epoch": 4.157726331639375,
-      "grad_norm": 8.015284538269043,
-      "learning_rate": 5.05364201016375e-06,
-      "loss": 0.5751,
+      "grad_norm": 2.254312038421631,
+      "learning_rate": 8.42273668360625e-05,
+      "loss": 0.659,
       "step": 22090
     },
     {
       "epoch": 4.159608507434594,
-      "grad_norm": 9.399229049682617,
-      "learning_rate": 5.0423489553924336e-06,
-      "loss": 0.472,
+      "grad_norm": 2.7255895137786865,
+      "learning_rate": 8.403914925654056e-05,
+      "loss": 0.527,
       "step": 22100
     },
     {
       "epoch": 4.161490683229814,
-      "grad_norm": 16.64990997314453,
-      "learning_rate": 5.031055900621119e-06,
-      "loss": 0.2494,
+      "grad_norm": 2.356421947479248,
+      "learning_rate": 8.385093167701865e-05,
+      "loss": 0.2846,
       "step": 22110
     },
     {
       "epoch": 4.163372859025033,
-      "grad_norm": 34.64622116088867,
-      "learning_rate": 5.019762845849802e-06,
-      "loss": 0.5575,
+      "grad_norm": 5.929986000061035,
+      "learning_rate": 8.366271409749671e-05,
+      "loss": 0.625,
       "step": 22120
     },
     {
       "epoch": 4.165255034820253,
-      "grad_norm": 43.623146057128906,
-      "learning_rate": 5.0084697910784875e-06,
-      "loss": 0.4865,
+      "grad_norm": 3.677624225616455,
+      "learning_rate": 8.347449651797479e-05,
+      "loss": 0.4381,
       "step": 22130
     },
     {
       "epoch": 4.167137210615471,
-      "grad_norm": 11.624351501464844,
-      "learning_rate": 4.997176736307171e-06,
-      "loss": 0.6536,
+      "grad_norm": 4.380472660064697,
+      "learning_rate": 8.328627893845285e-05,
+      "loss": 0.6449,
       "step": 22140
     },
     {
       "epoch": 4.169019386410691,
-      "grad_norm": 6.294399261474609,
-      "learning_rate": 4.9858836815358555e-06,
-      "loss": 0.6794,
+      "grad_norm": 3.776270866394043,
+      "learning_rate": 8.309806135893093e-05,
+      "loss": 0.7702,
       "step": 22150
     },
     {
       "epoch": 4.17090156220591,
-      "grad_norm": 16.353845596313477,
-      "learning_rate": 4.97459062676454e-06,
-      "loss": 0.2492,
+      "grad_norm": 3.1007485389709473,
+      "learning_rate": 8.290984377940899e-05,
+      "loss": 0.286,
       "step": 22160
     },
     {
       "epoch": 4.1727837380011294,
-      "grad_norm": 6.7267608642578125,
-      "learning_rate": 4.963297571993224e-06,
-      "loss": 0.5912,
+      "grad_norm": 4.839145660400391,
+      "learning_rate": 8.272162619988708e-05,
+      "loss": 0.7047,
       "step": 22170
     },
     {
       "epoch": 4.174665913796349,
-      "grad_norm": 0.19940780103206635,
-      "learning_rate": 4.952004517221909e-06,
-      "loss": 0.3615,
+      "grad_norm": 0.20469647645950317,
+      "learning_rate": 8.253340862036514e-05,
+      "loss": 0.5551,
       "step": 22180
     },
     {
       "epoch": 4.176548089591567,
-      "grad_norm": 13.068621635437012,
-      "learning_rate": 4.940711462450593e-06,
-      "loss": 0.7576,
+      "grad_norm": 2.0428857803344727,
+      "learning_rate": 8.234519104084322e-05,
+      "loss": 0.7462,
       "step": 22190
     },
     {
       "epoch": 4.178430265386787,
-      "grad_norm": 4.3102240562438965,
-      "learning_rate": 4.929418407679277e-06,
-      "loss": 0.5176,
+      "grad_norm": 1.7860795259475708,
+      "learning_rate": 8.215697346132128e-05,
+      "loss": 0.6495,
       "step": 22200
     },
     {
       "epoch": 4.180312441182006,
-      "grad_norm": 17.297819137573242,
-      "learning_rate": 4.918125352907962e-06,
-      "loss": 0.7968,
+      "grad_norm": 3.704061508178711,
+      "learning_rate": 8.196875588179936e-05,
+      "loss": 0.8011,
       "step": 22210
     },
     {
       "epoch": 4.182194616977226,
-      "grad_norm": 29.88180923461914,
-      "learning_rate": 4.906832298136646e-06,
-      "loss": 0.5499,
+      "grad_norm": 5.398728370666504,
+      "learning_rate": 8.178053830227745e-05,
+      "loss": 0.6543,
       "step": 22220
     },
     {
       "epoch": 4.184076792772445,
-      "grad_norm": 35.76202392578125,
-      "learning_rate": 4.8955392433653305e-06,
-      "loss": 0.2527,
+      "grad_norm": 4.630385875701904,
+      "learning_rate": 8.159232072275551e-05,
+      "loss": 0.3238,
       "step": 22230
     },
     {
       "epoch": 4.185958968567665,
-      "grad_norm": 23.927745819091797,
-      "learning_rate": 4.884246188594015e-06,
-      "loss": 0.4063,
+      "grad_norm": 3.8577587604522705,
+      "learning_rate": 8.140410314323359e-05,
+      "loss": 0.4467,
       "step": 22240
     },
     {
       "epoch": 4.187841144362883,
-      "grad_norm": 7.700712203979492,
-      "learning_rate": 4.872953133822699e-06,
-      "loss": 0.4893,
+      "grad_norm": 2.0660860538482666,
+      "learning_rate": 8.121588556371165e-05,
+      "loss": 0.4921,
       "step": 22250
     },
     {
       "epoch": 4.189723320158103,
-      "grad_norm": 14.746334075927734,
-      "learning_rate": 4.861660079051384e-06,
-      "loss": 0.7473,
+      "grad_norm": 2.0969619750976562,
+      "learning_rate": 8.102766798418973e-05,
+      "loss": 0.8799,
       "step": 22260
     },
     {
       "epoch": 4.191605495953322,
-      "grad_norm": 0.5177063941955566,
-      "learning_rate": 4.850367024280068e-06,
-      "loss": 0.5172,
+      "grad_norm": 1.2200628519058228,
+      "learning_rate": 8.083945040466779e-05,
+      "loss": 0.5306,
       "step": 22270
     },
     {
       "epoch": 4.193487671748541,
-      "grad_norm": 5.81657075881958,
-      "learning_rate": 4.8390739695087524e-06,
-      "loss": 0.5498,
+      "grad_norm": 1.9948633909225464,
+      "learning_rate": 8.065123282514588e-05,
+      "loss": 0.6008,
       "step": 22280
     },
     {
       "epoch": 4.195369847543761,
-      "grad_norm": 0.22678916156291962,
-      "learning_rate": 4.827780914737436e-06,
-      "loss": 0.4705,
+      "grad_norm": 0.05267168954014778,
+      "learning_rate": 8.046301524562394e-05,
+      "loss": 0.6233,
       "step": 22290
     },
     {
       "epoch": 4.19725202333898,
-      "grad_norm": 18.123903274536133,
-      "learning_rate": 4.816487859966121e-06,
-      "loss": 0.4834,
+      "grad_norm": 4.359118461608887,
+      "learning_rate": 8.027479766610202e-05,
+      "loss": 0.5167,
       "step": 22300
     },
     {
       "epoch": 4.199134199134199,
-      "grad_norm": 22.923141479492188,
-      "learning_rate": 4.805194805194805e-06,
-      "loss": 0.6182,
+      "grad_norm": 2.881960391998291,
+      "learning_rate": 8.008658008658008e-05,
+      "loss": 0.6885,
       "step": 22310
     },
     {
       "epoch": 4.201016374929418,
-      "grad_norm": 30.303974151611328,
-      "learning_rate": 4.79390175042349e-06,
-      "loss": 0.844,
+      "grad_norm": 5.036227703094482,
+      "learning_rate": 7.989836250705816e-05,
+      "loss": 0.9313,
       "step": 22320
     },
     {
       "epoch": 4.202898550724638,
-      "grad_norm": 0.405670702457428,
-      "learning_rate": 4.7826086956521735e-06,
-      "loss": 0.5439,
+      "grad_norm": 0.4130703806877136,
+      "learning_rate": 7.971014492753622e-05,
+      "loss": 0.6305,
       "step": 22330
     },
     {
       "epoch": 4.204780726519857,
-      "grad_norm": 19.21993064880371,
-      "learning_rate": 4.771315640880859e-06,
-      "loss": 0.4976,
+      "grad_norm": 5.823428630828857,
+      "learning_rate": 7.952192734801431e-05,
+      "loss": 0.5402,
       "step": 22340
     },
     {
       "epoch": 4.206662902315077,
-      "grad_norm": 18.2979679107666,
-      "learning_rate": 4.760022586109543e-06,
-      "loss": 0.7726,
+      "grad_norm": 4.27875280380249,
+      "learning_rate": 7.933370976849239e-05,
+      "loss": 0.8322,
       "step": 22350
     },
     {
       "epoch": 4.208545078110295,
-      "grad_norm": 4.73086404800415,
-      "learning_rate": 4.748729531338227e-06,
-      "loss": 0.6009,
+      "grad_norm": 0.5691286325454712,
+      "learning_rate": 7.914549218897045e-05,
+      "loss": 0.5322,
       "step": 22360
     },
     {
       "epoch": 4.2104272539055145,
-      "grad_norm": 0.11267448216676712,
-      "learning_rate": 4.737436476566912e-06,
-      "loss": 0.4873,
+      "grad_norm": 0.19783131778240204,
+      "learning_rate": 7.895727460944853e-05,
+      "loss": 0.5621,
       "step": 22370
     },
     {
       "epoch": 4.212309429700734,
-      "grad_norm": 13.262358665466309,
-      "learning_rate": 4.726143421795595e-06,
-      "loss": 0.588,
+      "grad_norm": 4.439708709716797,
+      "learning_rate": 7.876905702992659e-05,
+      "loss": 0.7453,
       "step": 22380
     },
     {
       "epoch": 4.214191605495953,
-      "grad_norm": 5.213744163513184,
-      "learning_rate": 4.714850367024281e-06,
-      "loss": 0.4623,
+      "grad_norm": 1.6272809505462646,
+      "learning_rate": 7.858083945040468e-05,
+      "loss": 0.5086,
       "step": 22390
     },
     {
       "epoch": 4.216073781291173,
-      "grad_norm": 16.470924377441406,
-      "learning_rate": 4.703557312252964e-06,
-      "loss": 0.6373,
+      "grad_norm": 4.118422031402588,
+      "learning_rate": 7.839262187088274e-05,
+      "loss": 0.6641,
       "step": 22400
     },
     {
       "epoch": 4.217955957086392,
-      "grad_norm": 10.145905494689941,
-      "learning_rate": 4.692264257481649e-06,
-      "loss": 0.6058,
+      "grad_norm": 2.0501649379730225,
+      "learning_rate": 7.820440429136082e-05,
+      "loss": 0.5865,
       "step": 22410
     },
     {
       "epoch": 4.219838132881611,
-      "grad_norm": 1.2421616315841675,
-      "learning_rate": 4.680971202710333e-06,
-      "loss": 0.4736,
+      "grad_norm": 0.5741236209869385,
+      "learning_rate": 7.801618671183888e-05,
+      "loss": 0.5987,
       "step": 22420
     },
     {
       "epoch": 4.22172030867683,
-      "grad_norm": 19.461740493774414,
-      "learning_rate": 4.669678147939017e-06,
-      "loss": 0.4471,
+      "grad_norm": 1.6564340591430664,
+      "learning_rate": 7.782796913231696e-05,
+      "loss": 0.4888,
       "step": 22430
     },
     {
       "epoch": 4.22360248447205,
-      "grad_norm": 11.891735076904297,
-      "learning_rate": 4.658385093167702e-06,
-      "loss": 0.5111,
+      "grad_norm": 3.18048357963562,
+      "learning_rate": 7.763975155279502e-05,
+      "loss": 0.7065,
       "step": 22440
     },
     {
       "epoch": 4.225484660267269,
-      "grad_norm": 42.17231750488281,
-      "learning_rate": 4.647092038396386e-06,
-      "loss": 1.0541,
+      "grad_norm": 6.997757911682129,
+      "learning_rate": 7.745153397327311e-05,
+      "loss": 1.157,
       "step": 22450
     },
     {
       "epoch": 4.2273668360624885,
-      "grad_norm": 8.65245246887207,
-      "learning_rate": 4.635798983625071e-06,
-      "loss": 0.6871,
+      "grad_norm": 2.3890907764434814,
+      "learning_rate": 7.726331639375119e-05,
+      "loss": 0.8599,
       "step": 22460
     },
     {
       "epoch": 4.229249011857707,
-      "grad_norm": 0.8841702938079834,
-      "learning_rate": 4.624505928853755e-06,
-      "loss": 0.478,
+      "grad_norm": 1.6332005262374878,
+      "learning_rate": 7.707509881422925e-05,
+      "loss": 0.487,
       "step": 22470
     },
     {
       "epoch": 4.2311311876529265,
-      "grad_norm": 7.112485408782959,
-      "learning_rate": 4.61321287408244e-06,
-      "loss": 0.7968,
+      "grad_norm": 2.324662208557129,
+      "learning_rate": 7.688688123470733e-05,
+      "loss": 0.8454,
       "step": 22480
     },
     {
       "epoch": 4.233013363448146,
-      "grad_norm": 11.136316299438477,
-      "learning_rate": 4.601919819311124e-06,
-      "loss": 0.3291,
+      "grad_norm": 3.145379066467285,
+      "learning_rate": 7.669866365518539e-05,
+      "loss": 0.4188,
       "step": 22490
     },
     {
       "epoch": 4.234895539243365,
-      "grad_norm": 11.090184211730957,
-      "learning_rate": 4.590626764539808e-06,
-      "loss": 0.5061,
+      "grad_norm": 1.542700171470642,
+      "learning_rate": 7.651044607566348e-05,
+      "loss": 0.5681,
       "step": 22500
     },
     {
       "epoch": 4.236777715038585,
-      "grad_norm": 3.2487759590148926,
-      "learning_rate": 4.579333709768492e-06,
-      "loss": 0.4647,
+      "grad_norm": 4.345068454742432,
+      "learning_rate": 7.632222849614154e-05,
+      "loss": 0.5611,
       "step": 22510
     },
     {
       "epoch": 4.238659890833804,
-      "grad_norm": 30.191455841064453,
-      "learning_rate": 4.568040654997177e-06,
-      "loss": 0.3747,
+      "grad_norm": 3.63413667678833,
+      "learning_rate": 7.613401091661962e-05,
+      "loss": 0.5058,
       "step": 22520
     },
     {
       "epoch": 4.240542066629023,
-      "grad_norm": 12.943588256835938,
-      "learning_rate": 4.556747600225861e-06,
-      "loss": 0.5714,
+      "grad_norm": 2.347446918487549,
+      "learning_rate": 7.594579333709768e-05,
+      "loss": 0.72,
       "step": 22530
     },
     {
       "epoch": 4.242424242424242,
-      "grad_norm": 6.223907947540283,
-      "learning_rate": 4.5454545454545455e-06,
-      "loss": 0.37,
+      "grad_norm": 4.7115092277526855,
+      "learning_rate": 7.575757575757576e-05,
+      "loss": 0.4005,
       "step": 22540
     },
     {
       "epoch": 4.244306418219462,
-      "grad_norm": 23.05088996887207,
-      "learning_rate": 4.53416149068323e-06,
-      "loss": 0.456,
+      "grad_norm": 2.4576327800750732,
+      "learning_rate": 7.556935817805382e-05,
+      "loss": 0.5354,
       "step": 22550
     },
     {
       "epoch": 4.246188594014681,
-      "grad_norm": 21.883832931518555,
-      "learning_rate": 4.522868435911914e-06,
-      "loss": 0.3142,
+      "grad_norm": 4.690468788146973,
+      "learning_rate": 7.538114059853191e-05,
+      "loss": 0.406,
       "step": 22560
     },
     {
       "epoch": 4.2480707698099005,
-      "grad_norm": 10.83736515045166,
-      "learning_rate": 4.511575381140599e-06,
-      "loss": 0.5158,
+      "grad_norm": 2.983757495880127,
+      "learning_rate": 7.519292301900997e-05,
+      "loss": 0.6362,
       "step": 22570
     },
     {
       "epoch": 4.24995294560512,
-      "grad_norm": 2.516878366470337,
-      "learning_rate": 4.500282326369283e-06,
-      "loss": 0.623,
+      "grad_norm": 2.384744644165039,
+      "learning_rate": 7.500470543948805e-05,
+      "loss": 0.8287,
       "step": 22580
     },
     {
       "epoch": 4.2518351214003385,
-      "grad_norm": 0.16205434501171112,
-      "learning_rate": 4.4889892715979674e-06,
-      "loss": 0.5343,
+      "grad_norm": 3.082204580307007,
+      "learning_rate": 7.481648785996613e-05,
+      "loss": 0.4311,
       "step": 22590
     },
     {
       "epoch": 4.253717297195558,
-      "grad_norm": 8.219768524169922,
-      "learning_rate": 4.477696216826652e-06,
-      "loss": 0.612,
+      "grad_norm": 2.5956528186798096,
+      "learning_rate": 7.462827028044419e-05,
+      "loss": 0.6546,
       "step": 22600
     },
     {
       "epoch": 4.255599472990777,
-      "grad_norm": 13.677404403686523,
-      "learning_rate": 4.466403162055336e-06,
-      "loss": 0.6012,
+      "grad_norm": 5.551482200622559,
+      "learning_rate": 7.444005270092228e-05,
+      "loss": 0.7867,
       "step": 22610
     },
     {
       "epoch": 4.257481648785997,
-      "grad_norm": 10.702381134033203,
-      "learning_rate": 4.4551101072840206e-06,
-      "loss": 0.3911,
+      "grad_norm": 4.27431583404541,
+      "learning_rate": 7.425183512140034e-05,
+      "loss": 0.657,
       "step": 22620
     },
     {
       "epoch": 4.259363824581216,
-      "grad_norm": 0.8724668622016907,
-      "learning_rate": 4.443817052512705e-06,
-      "loss": 0.7228,
+      "grad_norm": 1.6535813808441162,
+      "learning_rate": 7.406361754187842e-05,
+      "loss": 0.6216,
       "step": 22630
     },
     {
       "epoch": 4.261246000376435,
-      "grad_norm": 40.3377799987793,
-      "learning_rate": 4.432523997741389e-06,
-      "loss": 0.7456,
+      "grad_norm": 3.505030870437622,
+      "learning_rate": 7.387539996235648e-05,
+      "loss": 0.709,
       "step": 22640
     },
     {
       "epoch": 4.263128176171654,
-      "grad_norm": 15.633112907409668,
-      "learning_rate": 4.421230942970074e-06,
-      "loss": 0.5719,
+      "grad_norm": 2.203139543533325,
+      "learning_rate": 7.368718238283456e-05,
+      "loss": 0.6866,
       "step": 22650
     },
     {
       "epoch": 4.265010351966874,
-      "grad_norm": 12.120109558105469,
-      "learning_rate": 4.409937888198757e-06,
-      "loss": 0.6745,
+      "grad_norm": 2.7182722091674805,
+      "learning_rate": 7.349896480331262e-05,
+      "loss": 0.8602,
       "step": 22660
     },
     {
       "epoch": 4.266892527762093,
-      "grad_norm": 17.709228515625,
-      "learning_rate": 4.3986448334274425e-06,
-      "loss": 0.5516,
+      "grad_norm": 1.7698827981948853,
+      "learning_rate": 7.331074722379071e-05,
+      "loss": 0.707,
       "step": 22670
     },
     {
       "epoch": 4.2687747035573125,
-      "grad_norm": 8.81289005279541,
-      "learning_rate": 4.387351778656126e-06,
-      "loss": 0.4194,
+      "grad_norm": 2.303340196609497,
+      "learning_rate": 7.312252964426877e-05,
+      "loss": 0.4766,
       "step": 22680
     },
     {
       "epoch": 4.270656879352532,
-      "grad_norm": 14.428284645080566,
-      "learning_rate": 4.376058723884811e-06,
-      "loss": 0.4524,
+      "grad_norm": 2.723996639251709,
+      "learning_rate": 7.293431206474685e-05,
+      "loss": 0.5091,
       "step": 22690
     },
     {
       "epoch": 4.2725390551477505,
-      "grad_norm": 15.037586212158203,
-      "learning_rate": 4.364765669113496e-06,
-      "loss": 0.6796,
+      "grad_norm": 2.801182746887207,
+      "learning_rate": 7.274609448522493e-05,
+      "loss": 0.7886,
       "step": 22700
     },
     {
       "epoch": 4.27442123094297,
-      "grad_norm": 7.700161933898926,
-      "learning_rate": 4.35347261434218e-06,
-      "loss": 0.4788,
+      "grad_norm": 1.358647346496582,
+      "learning_rate": 7.255787690570299e-05,
+      "loss": 0.5111,
       "step": 22710
     },
     {
       "epoch": 4.276303406738189,
-      "grad_norm": 3.1909353733062744,
-      "learning_rate": 4.342179559570864e-06,
-      "loss": 0.4486,
+      "grad_norm": 1.516494870185852,
+      "learning_rate": 7.236965932618108e-05,
+      "loss": 0.6179,
       "step": 22720
     },
     {
       "epoch": 4.278185582533409,
-      "grad_norm": 17.010690689086914,
-      "learning_rate": 4.330886504799548e-06,
-      "loss": 0.6104,
+      "grad_norm": 2.8411619663238525,
+      "learning_rate": 7.218144174665914e-05,
+      "loss": 0.5986,
       "step": 22730
     },
     {
       "epoch": 4.280067758328628,
-      "grad_norm": 8.390015602111816,
-      "learning_rate": 4.319593450028233e-06,
-      "loss": 0.5245,
+      "grad_norm": 1.708974838256836,
+      "learning_rate": 7.199322416713722e-05,
+      "loss": 0.6238,
       "step": 22740
     },
     {
       "epoch": 4.281949934123848,
-      "grad_norm": 14.018758773803711,
-      "learning_rate": 4.308300395256917e-06,
-      "loss": 0.4835,
+      "grad_norm": 3.111238718032837,
+      "learning_rate": 7.180500658761528e-05,
+      "loss": 0.4499,
       "step": 22750
     },
     {
       "epoch": 4.283832109919066,
-      "grad_norm": 0.4540165662765503,
-      "learning_rate": 4.297007340485602e-06,
-      "loss": 0.5203,
+      "grad_norm": 0.690464198589325,
+      "learning_rate": 7.161678900809336e-05,
+      "loss": 0.4686,
       "step": 22760
     },
     {
       "epoch": 4.285714285714286,
-      "grad_norm": 2.0470516681671143,
-      "learning_rate": 4.2857142857142855e-06,
-      "loss": 0.4774,
+      "grad_norm": 0.1224987581372261,
+      "learning_rate": 7.142857142857142e-05,
+      "loss": 0.3938,
       "step": 22770
     },
     {
       "epoch": 4.287596461509505,
-      "grad_norm": 9.436823844909668,
-      "learning_rate": 4.274421230942971e-06,
-      "loss": 0.3557,
+      "grad_norm": 3.601773738861084,
+      "learning_rate": 7.124035384904951e-05,
+      "loss": 0.4035,
       "step": 22780
     },
     {
       "epoch": 4.2894786373047245,
-      "grad_norm": 14.441892623901367,
-      "learning_rate": 4.263128176171654e-06,
-      "loss": 0.4613,
+      "grad_norm": 2.2314205169677734,
+      "learning_rate": 7.105213626952757e-05,
+      "loss": 0.5302,
       "step": 22790
     },
     {
       "epoch": 4.291360813099944,
-      "grad_norm": 21.169797897338867,
-      "learning_rate": 4.251835121400339e-06,
-      "loss": 0.4245,
+      "grad_norm": 1.4461051225662231,
+      "learning_rate": 7.086391869000565e-05,
+      "loss": 0.3793,
       "step": 22800
     },
     {
       "epoch": 4.293242988895162,
-      "grad_norm": 5.513450622558594,
-      "learning_rate": 4.240542066629023e-06,
-      "loss": 0.5082,
+      "grad_norm": 1.465973138809204,
+      "learning_rate": 7.067570111048371e-05,
+      "loss": 0.6275,
       "step": 22810
     },
     {
       "epoch": 4.295125164690382,
-      "grad_norm": 9.460198402404785,
-      "learning_rate": 4.229249011857707e-06,
-      "loss": 0.6396,
+      "grad_norm": 4.16806173324585,
+      "learning_rate": 7.048748353096179e-05,
+      "loss": 0.8245,
       "step": 22820
     },
     {
       "epoch": 4.297007340485601,
-      "grad_norm": 16.516008377075195,
-      "learning_rate": 4.217955957086393e-06,
-      "loss": 0.4007,
+      "grad_norm": 1.0273966789245605,
+      "learning_rate": 7.029926595143988e-05,
+      "loss": 0.4617,
       "step": 22830
     },
     {
       "epoch": 4.298889516280821,
-      "grad_norm": 12.221315383911133,
-      "learning_rate": 4.206662902315076e-06,
-      "loss": 0.541,
+      "grad_norm": 1.136577844619751,
+      "learning_rate": 7.011104837191794e-05,
+      "loss": 0.4239,
       "step": 22840
     },
     {
       "epoch": 4.30077169207604,
-      "grad_norm": 24.764720916748047,
-      "learning_rate": 4.195369847543761e-06,
-      "loss": 0.6108,
+      "grad_norm": 3.0240557193756104,
+      "learning_rate": 6.992283079239602e-05,
+      "loss": 0.5731,
       "step": 22850
     },
     {
       "epoch": 4.30265386787126,
-      "grad_norm": 16.752914428710938,
-      "learning_rate": 4.184076792772445e-06,
-      "loss": 0.6381,
+      "grad_norm": 2.209796190261841,
+      "learning_rate": 6.973461321287408e-05,
+      "loss": 0.7176,
       "step": 22860
     },
     {
       "epoch": 4.304536043666478,
-      "grad_norm": 30.31788444519043,
-      "learning_rate": 4.172783738001129e-06,
-      "loss": 0.5743,
+      "grad_norm": 4.721385478973389,
+      "learning_rate": 6.954639563335216e-05,
+      "loss": 0.7321,
       "step": 22870
     },
     {
       "epoch": 4.306418219461698,
-      "grad_norm": 14.529326438903809,
-      "learning_rate": 4.161490683229814e-06,
-      "loss": 0.4165,
+      "grad_norm": 5.0673747062683105,
+      "learning_rate": 6.935817805383022e-05,
+      "loss": 0.4734,
       "step": 22880
     },
     {
       "epoch": 4.308300395256917,
-      "grad_norm": 5.0346808433532715,
-      "learning_rate": 4.150197628458498e-06,
-      "loss": 0.6804,
+      "grad_norm": 2.6815907955169678,
+      "learning_rate": 6.916996047430831e-05,
+      "loss": 0.7312,
       "step": 22890
     },
     {
       "epoch": 4.310182571052136,
-      "grad_norm": 12.99887466430664,
-      "learning_rate": 4.1389045736871824e-06,
-      "loss": 0.3671,
+      "grad_norm": 3.661799669265747,
+      "learning_rate": 6.898174289478637e-05,
+      "loss": 0.4466,
       "step": 22900
     },
     {
       "epoch": 4.312064746847356,
-      "grad_norm": 26.515596389770508,
-      "learning_rate": 4.127611518915867e-06,
-      "loss": 0.479,
+      "grad_norm": 4.6033935546875,
+      "learning_rate": 6.879352531526445e-05,
+      "loss": 0.5665,
       "step": 22910
     },
     {
       "epoch": 4.313946922642575,
-      "grad_norm": 21.167415618896484,
-      "learning_rate": 4.116318464144551e-06,
-      "loss": 0.5802,
+      "grad_norm": 1.0957890748977661,
+      "learning_rate": 6.860530773574251e-05,
+      "loss": 0.5804,
       "step": 22920
     },
     {
       "epoch": 4.315829098437794,
-      "grad_norm": 16.790386199951172,
-      "learning_rate": 4.1050254093732356e-06,
-      "loss": 0.6888,
+      "grad_norm": 2.1576459407806396,
+      "learning_rate": 6.841709015622059e-05,
+      "loss": 0.6856,
       "step": 22930
     },
     {
       "epoch": 4.317711274233013,
-      "grad_norm": 10.245500564575195,
-      "learning_rate": 4.093732354601919e-06,
-      "loss": 0.7874,
+      "grad_norm": 1.8269578218460083,
+      "learning_rate": 6.822887257669865e-05,
+      "loss": 0.6718,
       "step": 22940
     },
     {
       "epoch": 4.319593450028233,
-      "grad_norm": 21.216289520263672,
-      "learning_rate": 4.082439299830604e-06,
-      "loss": 0.3578,
+      "grad_norm": 1.6212677955627441,
+      "learning_rate": 6.804065499717674e-05,
+      "loss": 0.5032,
       "step": 22950
     },
     {
       "epoch": 4.321475625823452,
-      "grad_norm": 18.367511749267578,
-      "learning_rate": 4.071146245059289e-06,
-      "loss": 0.6691,
+      "grad_norm": 3.279130697250366,
+      "learning_rate": 6.785243741765482e-05,
+      "loss": 0.743,
       "step": 22960
     },
     {
       "epoch": 4.323357801618672,
-      "grad_norm": 24.98716163635254,
-      "learning_rate": 4.059853190287973e-06,
-      "loss": 0.5337,
+      "grad_norm": 0.7260986566543579,
+      "learning_rate": 6.766421983813288e-05,
+      "loss": 0.5386,
       "step": 22970
     },
     {
       "epoch": 4.32523997741389,
-      "grad_norm": 0.125027135014534,
-      "learning_rate": 4.0485601355166575e-06,
-      "loss": 0.6233,
+      "grad_norm": 0.01941692642867565,
+      "learning_rate": 6.747600225861096e-05,
+      "loss": 0.4597,
       "step": 22980
     },
     {
       "epoch": 4.3271221532091095,
-      "grad_norm": 19.9777889251709,
-      "learning_rate": 4.037267080745342e-06,
-      "loss": 0.5926,
+      "grad_norm": 4.9999847412109375,
+      "learning_rate": 6.728778467908902e-05,
+      "loss": 0.6838,
       "step": 22990
     },
     {
       "epoch": 4.329004329004329,
-      "grad_norm": 20.854936599731445,
-      "learning_rate": 4.025974025974026e-06,
-      "loss": 0.5644,
+      "grad_norm": 2.1625707149505615,
+      "learning_rate": 6.709956709956711e-05,
+      "loss": 0.6977,
       "step": 23000
     },
     {
       "epoch": 4.330886504799548,
-      "grad_norm": 7.7878546714782715,
-      "learning_rate": 4.014680971202711e-06,
-      "loss": 0.5073,
+      "grad_norm": 1.335459589958191,
+      "learning_rate": 6.691134952004517e-05,
+      "loss": 0.5518,
       "step": 23010
     },
     {
       "epoch": 4.332768680594768,
-      "grad_norm": 7.36667013168335,
-      "learning_rate": 4.003387916431395e-06,
-      "loss": 0.1648,
+      "grad_norm": 2.090069055557251,
+      "learning_rate": 6.672313194052325e-05,
+      "loss": 0.2411,
       "step": 23020
     },
     {
       "epoch": 4.334650856389986,
-      "grad_norm": 9.753989219665527,
-      "learning_rate": 3.9920948616600785e-06,
-      "loss": 0.4856,
+      "grad_norm": 3.832279682159424,
+      "learning_rate": 6.653491436100131e-05,
+      "loss": 0.5272,
       "step": 23030
     },
     {
       "epoch": 4.336533032185206,
-      "grad_norm": 14.510024070739746,
-      "learning_rate": 3.980801806888764e-06,
-      "loss": 0.4652,
+      "grad_norm": 5.671814918518066,
+      "learning_rate": 6.634669678147939e-05,
+      "loss": 0.469,
       "step": 23040
     },
     {
       "epoch": 4.338415207980425,
-      "grad_norm": 15.549747467041016,
-      "learning_rate": 3.969508752117447e-06,
-      "loss": 0.4127,
+      "grad_norm": 2.7710330486297607,
+      "learning_rate": 6.615847920195745e-05,
+      "loss": 0.4091,
       "step": 23050
     },
     {
       "epoch": 4.340297383775645,
-      "grad_norm": 8.210662841796875,
-      "learning_rate": 3.9582156973461325e-06,
-      "loss": 0.5347,
+      "grad_norm": 2.5829217433929443,
+      "learning_rate": 6.597026162243554e-05,
+      "loss": 0.4833,
       "step": 23060
     },
     {
       "epoch": 4.342179559570864,
-      "grad_norm": 7.637692451477051,
-      "learning_rate": 3.946922642574817e-06,
-      "loss": 0.5743,
+      "grad_norm": 1.8346978425979614,
+      "learning_rate": 6.578204404291362e-05,
+      "loss": 0.5997,
       "step": 23070
     },
     {
       "epoch": 4.3440617353660835,
-      "grad_norm": 0.12690700590610504,
-      "learning_rate": 3.935629587803501e-06,
-      "loss": 0.6929,
+      "grad_norm": 0.10548819601535797,
+      "learning_rate": 6.559382646339168e-05,
+      "loss": 0.6695,
       "step": 23080
     },
     {
       "epoch": 4.345943911161302,
-      "grad_norm": 4.598480701446533,
-      "learning_rate": 3.924336533032186e-06,
-      "loss": 0.3635,
+      "grad_norm": 0.4382118582725525,
+      "learning_rate": 6.540560888386976e-05,
+      "loss": 0.3085,
       "step": 23090
     },
     {
       "epoch": 4.3478260869565215,
-      "grad_norm": 2.793907403945923,
-      "learning_rate": 3.913043478260869e-06,
-      "loss": 0.5651,
+      "grad_norm": 0.8366402387619019,
+      "learning_rate": 6.521739130434782e-05,
+      "loss": 0.55,
       "step": 23100
     },
     {
       "epoch": 4.349708262751741,
-      "grad_norm": 32.544918060302734,
-      "learning_rate": 3.9017504234895545e-06,
-      "loss": 0.5491,
+      "grad_norm": 6.871497631072998,
+      "learning_rate": 6.502917372482591e-05,
+      "loss": 0.5395,
       "step": 23110
     },
     {
       "epoch": 4.35159043854696,
-      "grad_norm": 15.509510040283203,
-      "learning_rate": 3.890457368718238e-06,
-      "loss": 1.0166,
+      "grad_norm": 5.250075340270996,
+      "learning_rate": 6.484095614530397e-05,
+      "loss": 0.9494,
       "step": 23120
     },
     {
       "epoch": 4.35347261434218,
-      "grad_norm": 3.5904457569122314,
-      "learning_rate": 3.879164313946923e-06,
-      "loss": 0.6177,
+      "grad_norm": 0.5241207480430603,
+      "learning_rate": 6.465273856578205e-05,
+      "loss": 0.6133,
       "step": 23130
     },
     {
       "epoch": 4.355354790137399,
-      "grad_norm": 7.414836406707764,
-      "learning_rate": 3.867871259175607e-06,
-      "loss": 0.4744,
+      "grad_norm": 2.0515754222869873,
+      "learning_rate": 6.446452098626011e-05,
+      "loss": 0.4578,
       "step": 23140
     },
     {
       "epoch": 4.357236965932618,
-      "grad_norm": 6.72654390335083,
-      "learning_rate": 3.856578204404292e-06,
-      "loss": 0.4433,
+      "grad_norm": 2.292283535003662,
+      "learning_rate": 6.427630340673819e-05,
+      "loss": 0.5271,
       "step": 23150
     },
     {
       "epoch": 4.359119141727837,
-      "grad_norm": 1.880739450454712,
-      "learning_rate": 3.8452851496329755e-06,
-      "loss": 0.4546,
+      "grad_norm": 1.606432557106018,
+      "learning_rate": 6.408808582721625e-05,
+      "loss": 0.5568,
       "step": 23160
     },
     {
       "epoch": 4.361001317523057,
-      "grad_norm": 15.60661792755127,
-      "learning_rate": 3.83399209486166e-06,
-      "loss": 0.6959,
+      "grad_norm": 5.097689151763916,
+      "learning_rate": 6.389986824769434e-05,
+      "loss": 0.6891,
       "step": 23170
     },
     {
       "epoch": 4.362883493318276,
-      "grad_norm": 0.70307856798172,
-      "learning_rate": 3.822699040090344e-06,
-      "loss": 0.5917,
+      "grad_norm": 0.15392223000526428,
+      "learning_rate": 6.37116506681724e-05,
+      "loss": 0.6539,
       "step": 23180
     },
     {
       "epoch": 4.3647656691134955,
-      "grad_norm": 16.895275115966797,
-      "learning_rate": 3.811405985319029e-06,
-      "loss": 0.6483,
+      "grad_norm": 7.145848274230957,
+      "learning_rate": 6.352343308865048e-05,
+      "loss": 0.612,
       "step": 23190
     },
     {
       "epoch": 4.366647844908714,
-      "grad_norm": 14.281275749206543,
-      "learning_rate": 3.8001129305477135e-06,
-      "loss": 0.4527,
+      "grad_norm": 2.0338876247406006,
+      "learning_rate": 6.333521550912856e-05,
+      "loss": 0.4728,
       "step": 23200
     },
     {
       "epoch": 4.3685300207039335,
-      "grad_norm": 16.391109466552734,
-      "learning_rate": 3.7888198757763974e-06,
-      "loss": 0.8592,
+      "grad_norm": 4.034317493438721,
+      "learning_rate": 6.314699792960662e-05,
+      "loss": 0.855,
       "step": 23210
     },
     {
       "epoch": 4.370412196499153,
-      "grad_norm": 10.493399620056152,
-      "learning_rate": 3.7775268210050822e-06,
-      "loss": 0.8362,
+      "grad_norm": 2.8626930713653564,
+      "learning_rate": 6.295878035008471e-05,
+      "loss": 0.8229,
       "step": 23220
     },
     {
       "epoch": 4.372294372294372,
-      "grad_norm": 4.833158493041992,
-      "learning_rate": 3.766233766233766e-06,
-      "loss": 0.4836,
+      "grad_norm": 1.998652696609497,
+      "learning_rate": 6.277056277056277e-05,
+      "loss": 0.521,
       "step": 23230
     },
     {
       "epoch": 4.374176548089592,
-      "grad_norm": 19.595304489135742,
-      "learning_rate": 3.754940711462451e-06,
-      "loss": 0.3935,
+      "grad_norm": 1.864508867263794,
+      "learning_rate": 6.258234519104085e-05,
+      "loss": 0.4708,
       "step": 23240
     },
     {
       "epoch": 4.376058723884811,
-      "grad_norm": 9.765375137329102,
-      "learning_rate": 3.743647656691135e-06,
-      "loss": 0.642,
+      "grad_norm": 2.864326000213623,
+      "learning_rate": 6.239412761151891e-05,
+      "loss": 0.6477,
       "step": 23250
     },
     {
       "epoch": 4.37794089968003,
-      "grad_norm": 33.9497184753418,
-      "learning_rate": 3.7323546019198198e-06,
-      "loss": 0.5137,
+      "grad_norm": 4.261322498321533,
+      "learning_rate": 6.220591003199699e-05,
+      "loss": 0.5401,
       "step": 23260
     },
     {
       "epoch": 4.379823075475249,
-      "grad_norm": 13.267662048339844,
-      "learning_rate": 3.7210615471485037e-06,
-      "loss": 0.3194,
+      "grad_norm": 4.250820636749268,
+      "learning_rate": 6.201769245247507e-05,
+      "loss": 0.3505,
       "step": 23270
     },
     {
       "epoch": 4.381705251270469,
-      "grad_norm": 25.924053192138672,
-      "learning_rate": 3.709768492377188e-06,
-      "loss": 0.3042,
+      "grad_norm": 3.9857656955718994,
+      "learning_rate": 6.182947487295314e-05,
+      "loss": 0.3791,
       "step": 23280
     },
     {
       "epoch": 4.383587427065688,
-      "grad_norm": 0.388808935880661,
-      "learning_rate": 3.6984754376058725e-06,
-      "loss": 0.5223,
+      "grad_norm": 0.5790356397628784,
+      "learning_rate": 6.16412572934312e-05,
+      "loss": 0.523,
       "step": 23290
     },
     {
       "epoch": 4.3854696028609075,
-      "grad_norm": 28.93108558654785,
-      "learning_rate": 3.687182382834557e-06,
-      "loss": 0.6084,
+      "grad_norm": 4.269297122955322,
+      "learning_rate": 6.145303971390928e-05,
+      "loss": 0.6019,
       "step": 23300
     },
     {
       "epoch": 4.387351778656127,
-      "grad_norm": 14.811662673950195,
-      "learning_rate": 3.6758893280632412e-06,
-      "loss": 0.522,
+      "grad_norm": 4.494009971618652,
+      "learning_rate": 6.126482213438736e-05,
+      "loss": 0.7931,
       "step": 23310
     },
     {
       "epoch": 4.3892339544513455,
-      "grad_norm": 11.806975364685059,
-      "learning_rate": 3.6645962732919256e-06,
-      "loss": 0.4666,
+      "grad_norm": 2.184983015060425,
+      "learning_rate": 6.107660455486542e-05,
+      "loss": 0.6101,
       "step": 23320
     },
     {
       "epoch": 4.391116130246565,
-      "grad_norm": 76.58401489257812,
-      "learning_rate": 3.65330321852061e-06,
-      "loss": 0.4884,
+      "grad_norm": 2.364179849624634,
+      "learning_rate": 6.08883869753435e-05,
+      "loss": 0.4942,
       "step": 23330
     },
     {
       "epoch": 4.392998306041784,
-      "grad_norm": 0.8018186688423157,
-      "learning_rate": 3.6420101637492944e-06,
-      "loss": 0.5054,
+      "grad_norm": 0.16114617884159088,
+      "learning_rate": 6.0700169395821566e-05,
+      "loss": 0.526,
       "step": 23340
     },
     {
       "epoch": 4.394880481837004,
-      "grad_norm": 6.801021099090576,
-      "learning_rate": 3.6307171089779784e-06,
-      "loss": 0.339,
+      "grad_norm": 2.0968990325927734,
+      "learning_rate": 6.051195181629964e-05,
+      "loss": 0.2826,
       "step": 23350
     },
     {
       "epoch": 4.396762657632223,
-      "grad_norm": 12.988116264343262,
-      "learning_rate": 3.6194240542066627e-06,
-      "loss": 0.2921,
+      "grad_norm": 2.517059087753296,
+      "learning_rate": 6.032373423677771e-05,
+      "loss": 0.5407,
       "step": 23360
     },
     {
       "epoch": 4.398644833427442,
-      "grad_norm": 19.021095275878906,
-      "learning_rate": 3.608130999435347e-06,
-      "loss": 0.4502,
+      "grad_norm": 2.5663349628448486,
+      "learning_rate": 6.013551665725579e-05,
+      "loss": 0.4843,
       "step": 23370
     },
     {
       "epoch": 4.400527009222661,
-      "grad_norm": 8.15141487121582,
-      "learning_rate": 3.596837944664032e-06,
-      "loss": 0.6093,
+      "grad_norm": 1.0666170120239258,
+      "learning_rate": 5.9947299077733865e-05,
+      "loss": 0.6544,
       "step": 23380
     },
     {
       "epoch": 4.402409185017881,
-      "grad_norm": 0.6214210987091064,
-      "learning_rate": 3.5855448898927163e-06,
-      "loss": 0.6574,
+      "grad_norm": 0.42081281542778015,
+      "learning_rate": 5.9759081498211935e-05,
+      "loss": 0.7458,
       "step": 23390
     },
     {
       "epoch": 4.4042913608131,
-      "grad_norm": 14.339752197265625,
-      "learning_rate": 3.5742518351214007e-06,
-      "loss": 0.4536,
+      "grad_norm": 2.0457112789154053,
+      "learning_rate": 5.957086391869001e-05,
+      "loss": 0.4973,
       "step": 23400
     },
     {
       "epoch": 4.4061735366083195,
-      "grad_norm": 16.84188461303711,
-      "learning_rate": 3.562958780350085e-06,
-      "loss": 0.2564,
+      "grad_norm": 5.91927433013916,
+      "learning_rate": 5.938264633916808e-05,
+      "loss": 0.3556,
       "step": 23410
     },
     {
       "epoch": 4.408055712403539,
-      "grad_norm": 18.416526794433594,
-      "learning_rate": 3.551665725578769e-06,
-      "loss": 0.4888,
+      "grad_norm": 2.4589664936065674,
+      "learning_rate": 5.919442875964615e-05,
+      "loss": 0.5741,
       "step": 23420
     },
     {
       "epoch": 4.409937888198757,
-      "grad_norm": 0.2957004904747009,
-      "learning_rate": 3.5403726708074534e-06,
-      "loss": 0.4994,
+      "grad_norm": 0.18644341826438904,
+      "learning_rate": 5.900621118012423e-05,
+      "loss": 0.5359,
       "step": 23430
     },
     {
       "epoch": 4.411820063993977,
-      "grad_norm": 14.493985176086426,
-      "learning_rate": 3.529079616036138e-06,
-      "loss": 0.5243,
+      "grad_norm": 3.3280177116394043,
+      "learning_rate": 5.88179936006023e-05,
+      "loss": 0.5731,
       "step": 23440
     },
     {
       "epoch": 4.413702239789196,
-      "grad_norm": 22.472604751586914,
-      "learning_rate": 3.517786561264822e-06,
-      "loss": 0.8668,
+      "grad_norm": 6.775310039520264,
+      "learning_rate": 5.8629776021080366e-05,
+      "loss": 0.8888,
       "step": 23450
     },
     {
       "epoch": 4.415584415584416,
-      "grad_norm": 1.9218357801437378,
-      "learning_rate": 3.5064935064935066e-06,
-      "loss": 0.55,
+      "grad_norm": 1.3114447593688965,
+      "learning_rate": 5.844155844155844e-05,
+      "loss": 0.6345,
       "step": 23460
     },
     {
       "epoch": 4.417466591379635,
-      "grad_norm": 15.064146995544434,
-      "learning_rate": 3.495200451722191e-06,
-      "loss": 0.6122,
+      "grad_norm": 2.811979055404663,
+      "learning_rate": 5.825334086203651e-05,
+      "loss": 0.5995,
       "step": 23470
     },
     {
       "epoch": 4.419348767174855,
-      "grad_norm": 29.05467987060547,
-      "learning_rate": 3.4839073969508753e-06,
-      "loss": 0.2962,
+      "grad_norm": 0.5569702982902527,
+      "learning_rate": 5.806512328251459e-05,
+      "loss": 0.394,
       "step": 23480
     },
     {
       "epoch": 4.421230942970073,
-      "grad_norm": 10.816146850585938,
-      "learning_rate": 3.4726143421795593e-06,
-      "loss": 0.4214,
+      "grad_norm": 2.1078720092773438,
+      "learning_rate": 5.787690570299266e-05,
+      "loss": 0.5194,
       "step": 23490
     },
     {
       "epoch": 4.423113118765293,
-      "grad_norm": 8.355408668518066,
-      "learning_rate": 3.461321287408244e-06,
-      "loss": 0.5314,
+      "grad_norm": 1.9005637168884277,
+      "learning_rate": 5.7688688123470735e-05,
+      "loss": 0.5483,
       "step": 23500
     },
     {
       "epoch": 4.424995294560512,
-      "grad_norm": 13.588001251220703,
-      "learning_rate": 3.4500282326369285e-06,
-      "loss": 0.8294,
+      "grad_norm": 3.41694712638855,
+      "learning_rate": 5.750047054394881e-05,
+      "loss": 0.7021,
       "step": 23510
     },
     {
       "epoch": 4.426877470355731,
-      "grad_norm": 20.003461837768555,
-      "learning_rate": 3.438735177865613e-06,
-      "loss": 0.7186,
+      "grad_norm": 4.113175868988037,
+      "learning_rate": 5.731225296442688e-05,
+      "loss": 0.7608,
       "step": 23520
     },
     {
       "epoch": 4.428759646150951,
-      "grad_norm": 5.769153594970703,
-      "learning_rate": 3.4274421230942972e-06,
-      "loss": 0.3721,
+      "grad_norm": 1.1039425134658813,
+      "learning_rate": 5.712403538490495e-05,
+      "loss": 0.4509,
       "step": 23530
     },
     {
       "epoch": 4.430641821946169,
-      "grad_norm": 5.754838466644287,
-      "learning_rate": 3.4161490683229816e-06,
-      "loss": 0.5478,
+      "grad_norm": 1.165903091430664,
+      "learning_rate": 5.693581780538303e-05,
+      "loss": 0.5474,
       "step": 23540
     },
     {
       "epoch": 4.432523997741389,
-      "grad_norm": 0.23657679557800293,
-      "learning_rate": 3.404856013551666e-06,
-      "loss": 0.3118,
+      "grad_norm": 5.634236812591553,
+      "learning_rate": 5.67476002258611e-05,
+      "loss": 0.3828,
       "step": 23550
     },
     {
       "epoch": 4.434406173536608,
-      "grad_norm": 15.234368324279785,
-      "learning_rate": 3.3935629587803504e-06,
-      "loss": 0.4746,
+      "grad_norm": 2.44069242477417,
+      "learning_rate": 5.6559382646339166e-05,
+      "loss": 0.5159,
       "step": 23560
     },
     {
       "epoch": 4.436288349331828,
-      "grad_norm": 2.7576918601989746,
-      "learning_rate": 3.3822699040090343e-06,
-      "loss": 0.6348,
+      "grad_norm": 1.1310418844223022,
+      "learning_rate": 5.637116506681724e-05,
+      "loss": 0.7407,
       "step": 23570
     },
     {
       "epoch": 4.438170525127047,
-      "grad_norm": 20.79529571533203,
-      "learning_rate": 3.3709768492377187e-06,
-      "loss": 0.4821,
+      "grad_norm": 3.2341320514678955,
+      "learning_rate": 5.618294748729531e-05,
+      "loss": 0.5031,
       "step": 23580
     },
     {
       "epoch": 4.440052700922266,
-      "grad_norm": 15.99216365814209,
-      "learning_rate": 3.359683794466403e-06,
-      "loss": 0.4527,
+      "grad_norm": 2.2330312728881836,
+      "learning_rate": 5.599472990777338e-05,
+      "loss": 0.6094,
       "step": 23590
     },
     {
       "epoch": 4.441934876717485,
-      "grad_norm": 11.632766723632812,
-      "learning_rate": 3.3483907396950875e-06,
-      "loss": 0.4984,
+      "grad_norm": 4.402441024780273,
+      "learning_rate": 5.580651232825146e-05,
+      "loss": 0.6228,
       "step": 23600
     },
     {
       "epoch": 4.4438170525127045,
-      "grad_norm": 13.765467643737793,
-      "learning_rate": 3.337097684923772e-06,
-      "loss": 0.6588,
+      "grad_norm": 3.7783493995666504,
+      "learning_rate": 5.561829474872953e-05,
+      "loss": 0.6626,
       "step": 23610
     },
     {
       "epoch": 4.445699228307924,
-      "grad_norm": 2.9188790321350098,
-      "learning_rate": 3.3258046301524562e-06,
-      "loss": 0.5091,
+      "grad_norm": 3.7687785625457764,
+      "learning_rate": 5.5430077169207605e-05,
+      "loss": 0.542,
       "step": 23620
     },
     {
       "epoch": 4.447581404103143,
-      "grad_norm": 33.282928466796875,
-      "learning_rate": 3.314511575381141e-06,
-      "loss": 0.665,
+      "grad_norm": 5.35054349899292,
+      "learning_rate": 5.524185958968568e-05,
+      "loss": 0.749,
       "step": 23630
     },
     {
       "epoch": 4.449463579898363,
-      "grad_norm": 0.6511355042457581,
-      "learning_rate": 3.303218520609825e-06,
-      "loss": 0.5527,
+      "grad_norm": 2.302736520767212,
+      "learning_rate": 5.505364201016375e-05,
+      "loss": 0.6489,
       "step": 23640
     },
     {
       "epoch": 4.451345755693581,
-      "grad_norm": 12.938034057617188,
-      "learning_rate": 3.2919254658385094e-06,
-      "loss": 0.5798,
+      "grad_norm": 2.1349406242370605,
+      "learning_rate": 5.486542443064183e-05,
+      "loss": 0.8253,
       "step": 23650
     },
     {
       "epoch": 4.453227931488801,
-      "grad_norm": 19.567174911499023,
-      "learning_rate": 3.2806324110671938e-06,
-      "loss": 0.624,
+      "grad_norm": 4.4704132080078125,
+      "learning_rate": 5.46772068511199e-05,
+      "loss": 0.5917,
       "step": 23660
     },
     {
       "epoch": 4.45511010728402,
-      "grad_norm": 0.04238367825746536,
-      "learning_rate": 3.269339356295878e-06,
-      "loss": 0.2131,
+      "grad_norm": 0.0549953430891037,
+      "learning_rate": 5.4488989271597966e-05,
+      "loss": 0.3432,
       "step": 23670
     },
     {
       "epoch": 4.45699228307924,
-      "grad_norm": 14.076987266540527,
-      "learning_rate": 3.2580463015245625e-06,
-      "loss": 0.6618,
+      "grad_norm": 2.6983518600463867,
+      "learning_rate": 5.430077169207604e-05,
+      "loss": 0.6768,
       "step": 23680
     },
     {
       "epoch": 4.458874458874459,
-      "grad_norm": 7.118105888366699,
-      "learning_rate": 3.246753246753247e-06,
-      "loss": 0.4138,
+      "grad_norm": 2.198878049850464,
+      "learning_rate": 5.411255411255411e-05,
+      "loss": 0.3771,
       "step": 23690
     },
     {
       "epoch": 4.4607566346696785,
-      "grad_norm": 13.331232070922852,
-      "learning_rate": 3.2354601919819313e-06,
-      "loss": 0.3546,
+      "grad_norm": 3.5589592456817627,
+      "learning_rate": 5.392433653303218e-05,
+      "loss": 0.3683,
       "step": 23700
     },
     {
       "epoch": 4.462638810464897,
-      "grad_norm": 19.263700485229492,
-      "learning_rate": 3.2241671372106153e-06,
-      "loss": 0.4596,
+      "grad_norm": 4.189352512359619,
+      "learning_rate": 5.373611895351026e-05,
+      "loss": 0.4458,
       "step": 23710
     },
     {
       "epoch": 4.4645209862601165,
-      "grad_norm": 23.13578987121582,
-      "learning_rate": 3.2128740824392996e-06,
-      "loss": 0.6003,
+      "grad_norm": 6.152927875518799,
+      "learning_rate": 5.354790137398833e-05,
+      "loss": 0.6845,
       "step": 23720
     },
     {
       "epoch": 4.466403162055336,
-      "grad_norm": 2.132723331451416,
-      "learning_rate": 3.201581027667984e-06,
-      "loss": 0.6584,
+      "grad_norm": 0.19540274143218994,
+      "learning_rate": 5.3359683794466405e-05,
+      "loss": 0.6013,
       "step": 23730
     },
     {
       "epoch": 4.468285337850555,
-      "grad_norm": 10.117582321166992,
-      "learning_rate": 3.1902879728966684e-06,
-      "loss": 0.3707,
+      "grad_norm": 1.5613901615142822,
+      "learning_rate": 5.3171466214944474e-05,
+      "loss": 0.4262,
       "step": 23740
     },
     {
       "epoch": 4.470167513645775,
-      "grad_norm": 1.6187089681625366,
-      "learning_rate": 3.1789949181253532e-06,
-      "loss": 0.3961,
+      "grad_norm": 0.4342918395996094,
+      "learning_rate": 5.298324863542255e-05,
+      "loss": 0.4913,
       "step": 23750
     },
     {
       "epoch": 4.472049689440993,
-      "grad_norm": 8.55710506439209,
-      "learning_rate": 3.1677018633540376e-06,
-      "loss": 0.3721,
+      "grad_norm": 1.4780311584472656,
+      "learning_rate": 5.279503105590063e-05,
+      "loss": 0.2758,
       "step": 23760
     },
     {
       "epoch": 4.473931865236213,
-      "grad_norm": 11.327536582946777,
-      "learning_rate": 3.156408808582722e-06,
-      "loss": 0.6747,
+      "grad_norm": 1.1511306762695312,
+      "learning_rate": 5.26068134763787e-05,
+      "loss": 0.5309,
       "step": 23770
     },
     {
       "epoch": 4.475814041031432,
-      "grad_norm": 7.744390487670898,
-      "learning_rate": 3.1451157538114064e-06,
-      "loss": 0.6423,
+      "grad_norm": 3.953789710998535,
+      "learning_rate": 5.2418595896856766e-05,
+      "loss": 0.6147,
       "step": 23780
     },
     {
       "epoch": 4.477696216826652,
-      "grad_norm": 1.3057587146759033,
-      "learning_rate": 3.1338226990400903e-06,
-      "loss": 0.2579,
+      "grad_norm": 0.7184831500053406,
+      "learning_rate": 5.223037831733484e-05,
+      "loss": 0.34,
       "step": 23790
     },
     {
       "epoch": 4.479578392621871,
-      "grad_norm": 10.926691055297852,
-      "learning_rate": 3.1225296442687747e-06,
-      "loss": 0.5395,
+      "grad_norm": 0.6724153757095337,
+      "learning_rate": 5.204216073781291e-05,
+      "loss": 0.5609,
       "step": 23800
     },
     {
       "epoch": 4.4814605684170905,
-      "grad_norm": 13.025545120239258,
-      "learning_rate": 3.111236589497459e-06,
-      "loss": 0.5213,
+      "grad_norm": 2.403193950653076,
+      "learning_rate": 5.185394315829098e-05,
+      "loss": 0.6468,
       "step": 23810
     },
     {
       "epoch": 4.483342744212309,
-      "grad_norm": 8.658554077148438,
-      "learning_rate": 3.0999435347261435e-06,
-      "loss": 0.522,
+      "grad_norm": 2.3661391735076904,
+      "learning_rate": 5.166572557876906e-05,
+      "loss": 0.5268,
       "step": 23820
     },
     {
       "epoch": 4.4852249200075285,
-      "grad_norm": 0.7279936671257019,
-      "learning_rate": 3.088650479954828e-06,
-      "loss": 0.3601,
+      "grad_norm": 3.755697250366211,
+      "learning_rate": 5.147750799924713e-05,
+      "loss": 0.5165,
       "step": 23830
     },
     {
       "epoch": 4.487107095802748,
-      "grad_norm": 11.042409896850586,
-      "learning_rate": 3.0773574251835122e-06,
-      "loss": 0.5312,
+      "grad_norm": 2.821866512298584,
+      "learning_rate": 5.1289290419725205e-05,
+      "loss": 0.6264,
       "step": 23840
     },
     {
       "epoch": 4.488989271597967,
-      "grad_norm": 2.8644516468048096,
-      "learning_rate": 3.0660643704121966e-06,
-      "loss": 0.5416,
+      "grad_norm": 1.1792480945587158,
+      "learning_rate": 5.1101072840203274e-05,
+      "loss": 0.5714,
       "step": 23850
     },
     {
       "epoch": 4.490871447393187,
-      "grad_norm": 16.289228439331055,
-      "learning_rate": 3.0547713156408806e-06,
-      "loss": 0.37,
+      "grad_norm": 3.8462843894958496,
+      "learning_rate": 5.0912855260681344e-05,
+      "loss": 0.3337,
       "step": 23860
     },
     {
       "epoch": 4.492753623188406,
-      "grad_norm": 25.66763687133789,
-      "learning_rate": 3.0434782608695654e-06,
-      "loss": 0.9307,
+      "grad_norm": 4.147275924682617,
+      "learning_rate": 5.072463768115943e-05,
+      "loss": 1.012,
       "step": 23870
     },
     {
       "epoch": 4.494635798983625,
-      "grad_norm": 3.9312744140625,
-      "learning_rate": 3.0321852060982498e-06,
-      "loss": 0.3846,
+      "grad_norm": 4.648754596710205,
+      "learning_rate": 5.05364201016375e-05,
+      "loss": 0.6537,
       "step": 23880
     },
     {
       "epoch": 4.496517974778844,
-      "grad_norm": 2.878774404525757,
-      "learning_rate": 3.020892151326934e-06,
-      "loss": 0.6847,
+      "grad_norm": 0.7671937346458435,
+      "learning_rate": 5.0348202522115566e-05,
+      "loss": 0.8298,
       "step": 23890
     },
     {
       "epoch": 4.498400150574064,
-      "grad_norm": 8.510382652282715,
-      "learning_rate": 3.0095990965556185e-06,
-      "loss": 0.4522,
+      "grad_norm": 1.893687129020691,
+      "learning_rate": 5.015998494259364e-05,
+      "loss": 0.4802,
       "step": 23900
     },
     {
       "epoch": 4.500282326369283,
-      "grad_norm": 9.131433486938477,
-      "learning_rate": 2.998306041784303e-06,
-      "loss": 0.3852,
+      "grad_norm": 2.864920139312744,
+      "learning_rate": 4.997176736307171e-05,
+      "loss": 0.4142,
       "step": 23910
     },
     {
       "epoch": 4.5021645021645025,
-      "grad_norm": 14.799365043640137,
-      "learning_rate": 2.9870129870129873e-06,
-      "loss": 0.3988,
+      "grad_norm": 2.016402244567871,
+      "learning_rate": 4.978354978354978e-05,
+      "loss": 0.5811,
       "step": 23920
     },
     {
       "epoch": 4.504046677959721,
-      "grad_norm": 25.332460403442383,
-      "learning_rate": 2.9757199322416712e-06,
-      "loss": 0.5088,
+      "grad_norm": 3.925957441329956,
+      "learning_rate": 4.959533220402786e-05,
+      "loss": 0.6436,
       "step": 23930
     },
     {
       "epoch": 4.5059288537549405,
-      "grad_norm": 15.210441589355469,
-      "learning_rate": 2.9644268774703556e-06,
-      "loss": 0.5733,
+      "grad_norm": 6.499224662780762,
+      "learning_rate": 4.940711462450593e-05,
+      "loss": 0.6918,
       "step": 23940
     },
     {
       "epoch": 4.50781102955016,
-      "grad_norm": 17.5853271484375,
-      "learning_rate": 2.95313382269904e-06,
-      "loss": 0.8435,
+      "grad_norm": 2.7809550762176514,
+      "learning_rate": 4.9218897044984e-05,
+      "loss": 0.8341,
       "step": 23950
     },
     {
       "epoch": 4.509693205345379,
-      "grad_norm": 5.9127020835876465,
-      "learning_rate": 2.9418407679277244e-06,
-      "loss": 0.405,
+      "grad_norm": 2.4689996242523193,
+      "learning_rate": 4.9030679465462074e-05,
+      "loss": 0.5343,
       "step": 23960
     },
     {
       "epoch": 4.511575381140599,
-      "grad_norm": 15.873945236206055,
-      "learning_rate": 2.9305477131564088e-06,
-      "loss": 0.3017,
+      "grad_norm": 2.489816188812256,
+      "learning_rate": 4.8842461885940144e-05,
+      "loss": 0.3929,
       "step": 23970
     },
     {
       "epoch": 4.513457556935818,
-      "grad_norm": 6.168259620666504,
-      "learning_rate": 2.919254658385093e-06,
-      "loss": 0.8213,
+      "grad_norm": 2.4629080295562744,
+      "learning_rate": 4.865424430641822e-05,
+      "loss": 0.8677,
       "step": 23980
     },
     {
       "epoch": 4.515339732731037,
-      "grad_norm": 14.95897102355957,
-      "learning_rate": 2.907961603613778e-06,
-      "loss": 0.7194,
+      "grad_norm": 3.8149843215942383,
+      "learning_rate": 4.84660267268963e-05,
+      "loss": 0.6299,
       "step": 23990
     },
     {
       "epoch": 4.517221908526256,
-      "grad_norm": 9.933831214904785,
-      "learning_rate": 2.896668548842462e-06,
-      "loss": 0.4408,
+      "grad_norm": 1.8469231128692627,
+      "learning_rate": 4.8277809147374366e-05,
+      "loss": 0.5836,
       "step": 24000
     },
     {
       "epoch": 4.519104084321476,
-      "grad_norm": 0.1559751182794571,
-      "learning_rate": 2.8853754940711463e-06,
-      "loss": 0.5496,
+      "grad_norm": 0.25325506925582886,
+      "learning_rate": 4.808959156785244e-05,
+      "loss": 0.5975,
       "step": 24010
     },
     {
       "epoch": 4.520986260116695,
-      "grad_norm": 16.345056533813477,
-      "learning_rate": 2.8740824392998307e-06,
-      "loss": 0.5864,
+      "grad_norm": 2.1251931190490723,
+      "learning_rate": 4.790137398833051e-05,
+      "loss": 0.5914,
       "step": 24020
     },
     {
       "epoch": 4.5228684359119145,
-      "grad_norm": 28.054203033447266,
-      "learning_rate": 2.862789384528515e-06,
-      "loss": 0.7297,
+      "grad_norm": 4.482444763183594,
+      "learning_rate": 4.771315640880858e-05,
+      "loss": 0.6149,
       "step": 24030
     },
     {
       "epoch": 4.524750611707134,
-      "grad_norm": 14.966489791870117,
-      "learning_rate": 2.8514963297571994e-06,
-      "loss": 0.3968,
+      "grad_norm": 0.7547290921211243,
+      "learning_rate": 4.752493882928666e-05,
+      "loss": 0.4447,
       "step": 24040
     },
     {
       "epoch": 4.526632787502352,
-      "grad_norm": 20.097759246826172,
-      "learning_rate": 2.840203274985884e-06,
-      "loss": 0.7483,
+      "grad_norm": 3.609314441680908,
+      "learning_rate": 4.733672124976473e-05,
+      "loss": 0.8325,
       "step": 24050
     },
     {
       "epoch": 4.528514963297572,
-      "grad_norm": 19.95496940612793,
-      "learning_rate": 2.828910220214568e-06,
-      "loss": 0.6654,
+      "grad_norm": 2.2269513607025146,
+      "learning_rate": 4.71485036702428e-05,
+      "loss": 0.6928,
       "step": 24060
     },
     {
       "epoch": 4.530397139092791,
-      "grad_norm": 8.588127136230469,
-      "learning_rate": 2.8176171654432526e-06,
-      "loss": 0.2225,
+      "grad_norm": 2.025987148284912,
+      "learning_rate": 4.6960286090720874e-05,
+      "loss": 0.3431,
       "step": 24070
     },
     {
       "epoch": 4.532279314888011,
-      "grad_norm": 10.798761367797852,
-      "learning_rate": 2.8063241106719366e-06,
-      "loss": 0.5675,
+      "grad_norm": 1.6247036457061768,
+      "learning_rate": 4.6772068511198944e-05,
+      "loss": 0.6616,
       "step": 24080
     },
     {
       "epoch": 4.53416149068323,
-      "grad_norm": 0.21323099732398987,
-      "learning_rate": 2.795031055900621e-06,
-      "loss": 0.4654,
+      "grad_norm": 3.7297098636627197,
+      "learning_rate": 4.658385093167702e-05,
+      "loss": 0.5445,
       "step": 24090
     },
     {
       "epoch": 4.536043666478449,
-      "grad_norm": 12.417448997497559,
-      "learning_rate": 2.7837380011293053e-06,
-      "loss": 0.7423,
+      "grad_norm": 2.0358946323394775,
+      "learning_rate": 4.639563335215509e-05,
+      "loss": 0.8047,
       "step": 24100
     },
     {
       "epoch": 4.537925842273668,
-      "grad_norm": 8.83974838256836,
-      "learning_rate": 2.7724449463579897e-06,
-      "loss": 0.3925,
+      "grad_norm": 2.806361675262451,
+      "learning_rate": 4.620741577263316e-05,
+      "loss": 0.5435,
       "step": 24110
     },
     {
       "epoch": 4.539808018068888,
-      "grad_norm": 28.08986473083496,
-      "learning_rate": 2.7611518915866745e-06,
-      "loss": 0.5835,
+      "grad_norm": 4.876021385192871,
+      "learning_rate": 4.601919819311124e-05,
+      "loss": 0.6177,
       "step": 24120
     },
     {
       "epoch": 4.541690193864107,
-      "grad_norm": 2.058382034301758,
-      "learning_rate": 2.749858836815359e-06,
-      "loss": 0.5558,
+      "grad_norm": 0.8642858862876892,
+      "learning_rate": 4.583098061358931e-05,
+      "loss": 0.5793,
       "step": 24130
     },
     {
       "epoch": 4.543572369659326,
-      "grad_norm": 11.856590270996094,
-      "learning_rate": 2.7385657820440433e-06,
-      "loss": 0.6624,
+      "grad_norm": 3.2075188159942627,
+      "learning_rate": 4.564276303406738e-05,
+      "loss": 0.6995,
       "step": 24140
     },
     {
       "epoch": 4.545454545454545,
-      "grad_norm": 28.79555320739746,
-      "learning_rate": 2.7272727272727272e-06,
-      "loss": 0.4069,
+      "grad_norm": 2.1683106422424316,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.4836,
       "step": 24150
     },
     {
       "epoch": 4.547336721249764,
-      "grad_norm": 0.156916081905365,
-      "learning_rate": 2.7159796725014116e-06,
-      "loss": 0.3982,
+      "grad_norm": 8.566926002502441,
+      "learning_rate": 4.526632787502353e-05,
+      "loss": 0.5407,
       "step": 24160
     },
     {
       "epoch": 4.549218897044984,
-      "grad_norm": 22.804506301879883,
-      "learning_rate": 2.704686617730096e-06,
-      "loss": 0.5942,
+      "grad_norm": 3.3746559619903564,
+      "learning_rate": 4.50781102955016e-05,
+      "loss": 0.524,
       "step": 24170
     },
     {
       "epoch": 4.551101072840203,
-      "grad_norm": 12.148880004882812,
-      "learning_rate": 2.6933935629587804e-06,
-      "loss": 0.6395,
+      "grad_norm": 2.927900791168213,
+      "learning_rate": 4.4889892715979674e-05,
+      "loss": 0.7662,
       "step": 24180
     },
     {
       "epoch": 4.552983248635423,
-      "grad_norm": 17.041955947875977,
-      "learning_rate": 2.6821005081874648e-06,
-      "loss": 0.349,
+      "grad_norm": 1.0152127742767334,
+      "learning_rate": 4.4701675136457744e-05,
+      "loss": 0.3934,
       "step": 24190
     },
     {
       "epoch": 4.554865424430642,
-      "grad_norm": 15.154557228088379,
-      "learning_rate": 2.670807453416149e-06,
-      "loss": 0.5323,
+      "grad_norm": 5.355255603790283,
+      "learning_rate": 4.451345755693582e-05,
+      "loss": 0.5864,
       "step": 24200
     },
     {
       "epoch": 4.556747600225862,
-      "grad_norm": 5.193390369415283,
-      "learning_rate": 2.6595143986448335e-06,
-      "loss": 0.306,
+      "grad_norm": 2.3401451110839844,
+      "learning_rate": 4.432523997741389e-05,
+      "loss": 0.3616,
       "step": 24210
     },
     {
       "epoch": 4.55862977602108,
-      "grad_norm": 14.101144790649414,
-      "learning_rate": 2.648221343873518e-06,
-      "loss": 0.5806,
+      "grad_norm": 2.2497141361236572,
+      "learning_rate": 4.413702239789196e-05,
+      "loss": 0.6165,
       "step": 24220
     },
     {
       "epoch": 4.5605119518162995,
-      "grad_norm": 4.696810722351074,
-      "learning_rate": 2.636928289102202e-06,
-      "loss": 0.4525,
+      "grad_norm": 1.6242727041244507,
+      "learning_rate": 4.3948804818370036e-05,
+      "loss": 0.5554,
       "step": 24230
     },
     {
       "epoch": 4.562394127611519,
-      "grad_norm": 2.615445613861084,
-      "learning_rate": 2.6256352343308867e-06,
-      "loss": 0.4967,
+      "grad_norm": 2.4595236778259277,
+      "learning_rate": 4.376058723884811e-05,
+      "loss": 0.5662,
       "step": 24240
     },
     {
       "epoch": 4.564276303406738,
-      "grad_norm": 35.27506637573242,
-      "learning_rate": 2.614342179559571e-06,
-      "loss": 0.4971,
+      "grad_norm": 0.9235624074935913,
+      "learning_rate": 4.357236965932618e-05,
+      "loss": 0.4612,
       "step": 24250
     },
     {
       "epoch": 4.566158479201958,
-      "grad_norm": 8.616056442260742,
-      "learning_rate": 2.6030491247882554e-06,
-      "loss": 0.3323,
+      "grad_norm": 1.4337669610977173,
+      "learning_rate": 4.338415207980426e-05,
+      "loss": 0.428,
       "step": 24260
     },
     {
       "epoch": 4.568040654997176,
-      "grad_norm": 1.696595311164856,
-      "learning_rate": 2.59175607001694e-06,
-      "loss": 0.343,
+      "grad_norm": 2.5631136894226074,
+      "learning_rate": 4.319593450028233e-05,
+      "loss": 0.3682,
       "step": 24270
     },
     {
       "epoch": 4.569922830792396,
-      "grad_norm": 8.711554527282715,
-      "learning_rate": 2.580463015245624e-06,
-      "loss": 0.5896,
+      "grad_norm": 2.581136703491211,
+      "learning_rate": 4.30077169207604e-05,
+      "loss": 0.7191,
       "step": 24280
     },
     {
       "epoch": 4.571805006587615,
-      "grad_norm": 15.512261390686035,
-      "learning_rate": 2.5691699604743086e-06,
-      "loss": 0.4962,
+      "grad_norm": 1.4882745742797852,
+      "learning_rate": 4.2819499341238474e-05,
+      "loss": 0.5948,
       "step": 24290
     },
     {
       "epoch": 4.573687182382835,
-      "grad_norm": 10.31464672088623,
-      "learning_rate": 2.5578769057029925e-06,
-      "loss": 0.5446,
+      "grad_norm": 2.5179505348205566,
+      "learning_rate": 4.2631281761716544e-05,
+      "loss": 0.639,
       "step": 24300
     },
     {
       "epoch": 4.575569358178054,
-      "grad_norm": 68.28598022460938,
-      "learning_rate": 2.546583850931677e-06,
-      "loss": 0.5234,
+      "grad_norm": 6.4887919425964355,
+      "learning_rate": 4.244306418219462e-05,
+      "loss": 0.6493,
       "step": 24310
     },
     {
       "epoch": 4.577451533973273,
-      "grad_norm": 23.362018585205078,
-      "learning_rate": 2.5352907961603613e-06,
-      "loss": 0.4662,
+      "grad_norm": 2.8530259132385254,
+      "learning_rate": 4.225484660267269e-05,
+      "loss": 0.4943,
       "step": 24320
     },
     {
       "epoch": 4.579333709768492,
-      "grad_norm": 0.38361719250679016,
-      "learning_rate": 2.5239977413890457e-06,
-      "loss": 0.3885,
+      "grad_norm": 0.3246712386608124,
+      "learning_rate": 4.206662902315076e-05,
+      "loss": 0.3812,
       "step": 24330
     },
     {
       "epoch": 4.5812158855637115,
-      "grad_norm": 13.636258125305176,
-      "learning_rate": 2.51270468661773e-06,
-      "loss": 0.1951,
+      "grad_norm": 2.248939037322998,
+      "learning_rate": 4.1878411443628836e-05,
+      "loss": 0.3863,
       "step": 24340
     },
     {
       "epoch": 4.583098061358931,
-      "grad_norm": 13.915003776550293,
-      "learning_rate": 2.5014116318464144e-06,
-      "loss": 0.2864,
+      "grad_norm": 3.7241125106811523,
+      "learning_rate": 4.1690193864106906e-05,
+      "loss": 0.4399,
       "step": 24350
     },
     {
       "epoch": 4.58498023715415,
-      "grad_norm": 28.40807342529297,
-      "learning_rate": 2.4901185770750993e-06,
-      "loss": 0.4974,
+      "grad_norm": 7.332564353942871,
+      "learning_rate": 4.150197628458498e-05,
+      "loss": 0.5854,
       "step": 24360
     },
     {
       "epoch": 4.58686241294937,
-      "grad_norm": 21.473955154418945,
-      "learning_rate": 2.478825522303783e-06,
-      "loss": 0.6872,
+      "grad_norm": 4.252923965454102,
+      "learning_rate": 4.131375870506306e-05,
+      "loss": 0.6474,
       "step": 24370
     },
     {
       "epoch": 4.588744588744589,
-      "grad_norm": 39.9091911315918,
-      "learning_rate": 2.4675324675324676e-06,
-      "loss": 0.4134,
+      "grad_norm": 5.477579593658447,
+      "learning_rate": 4.112554112554113e-05,
+      "loss": 0.5945,
       "step": 24380
     },
     {
       "epoch": 4.590626764539808,
-      "grad_norm": 18.155563354492188,
-      "learning_rate": 2.456239412761152e-06,
-      "loss": 0.6036,
+      "grad_norm": 4.131031513214111,
+      "learning_rate": 4.09373235460192e-05,
+      "loss": 0.6526,
       "step": 24390
     },
     {
       "epoch": 4.592508940335027,
-      "grad_norm": 9.114380836486816,
-      "learning_rate": 2.4449463579898364e-06,
-      "loss": 0.6799,
+      "grad_norm": 1.5530714988708496,
+      "learning_rate": 4.0749105966497274e-05,
+      "loss": 0.6654,
       "step": 24400
     },
     {
       "epoch": 4.594391116130247,
-      "grad_norm": 6.3043904304504395,
-      "learning_rate": 2.4336533032185207e-06,
-      "loss": 0.626,
+      "grad_norm": 2.8524997234344482,
+      "learning_rate": 4.0560888386975344e-05,
+      "loss": 0.5783,
       "step": 24410
     },
     {
       "epoch": 4.596273291925466,
-      "grad_norm": 16.44568634033203,
-      "learning_rate": 2.422360248447205e-06,
-      "loss": 0.3922,
+      "grad_norm": 2.7015812397003174,
+      "learning_rate": 4.0372670807453414e-05,
+      "loss": 0.4753,
       "step": 24420
     },
     {
       "epoch": 4.5981554677206855,
-      "grad_norm": 9.025197982788086,
-      "learning_rate": 2.4110671936758895e-06,
-      "loss": 0.7261,
+      "grad_norm": 1.2602338790893555,
+      "learning_rate": 4.018445322793149e-05,
+      "loss": 0.7536,
       "step": 24430
     },
     {
       "epoch": 4.600037643515904,
-      "grad_norm": 19.083044052124023,
-      "learning_rate": 2.399774138904574e-06,
-      "loss": 0.4588,
+      "grad_norm": 3.109779119491577,
+      "learning_rate": 3.999623564840956e-05,
+      "loss": 0.5292,
       "step": 24440
     },
     {
       "epoch": 4.6019198193111235,
-      "grad_norm": 16.03182029724121,
-      "learning_rate": 2.388481084133258e-06,
-      "loss": 0.4318,
+      "grad_norm": 3.305142402648926,
+      "learning_rate": 3.9808018068887636e-05,
+      "loss": 0.44,
       "step": 24450
     },
     {
       "epoch": 4.603801995106343,
-      "grad_norm": 1.4416394233703613,
-      "learning_rate": 2.3771880293619422e-06,
-      "loss": 0.658,
+      "grad_norm": 0.8609833717346191,
+      "learning_rate": 3.9619800489365706e-05,
+      "loss": 0.592,
       "step": 24460
     },
     {
       "epoch": 4.605684170901562,
-      "grad_norm": 7.5059075355529785,
-      "learning_rate": 2.3658949745906266e-06,
-      "loss": 0.4453,
+      "grad_norm": 1.3746150732040405,
+      "learning_rate": 3.9431582909843775e-05,
+      "loss": 0.5808,
       "step": 24470
     },
     {
       "epoch": 4.607566346696782,
-      "grad_norm": 4.021427631378174,
-      "learning_rate": 2.3546019198193114e-06,
-      "loss": 0.5636,
+      "grad_norm": 1.0084211826324463,
+      "learning_rate": 3.924336533032186e-05,
+      "loss": 0.7987,
       "step": 24480
     },
     {
       "epoch": 4.609448522492,
-      "grad_norm": 5.8382744789123535,
-      "learning_rate": 2.343308865047996e-06,
-      "loss": 0.2773,
+      "grad_norm": 2.227627992630005,
+      "learning_rate": 3.905514775079993e-05,
+      "loss": 0.4238,
       "step": 24490
     },
     {
       "epoch": 4.61133069828722,
-      "grad_norm": 10.830999374389648,
-      "learning_rate": 2.33201581027668e-06,
-      "loss": 0.5919,
+      "grad_norm": 3.0300769805908203,
+      "learning_rate": 3.8866930171278e-05,
+      "loss": 0.4748,
       "step": 24500
     },
     {
       "epoch": 4.613212874082439,
-      "grad_norm": 30.35820198059082,
-      "learning_rate": 2.3207227555053646e-06,
-      "loss": 0.3376,
+      "grad_norm": 3.6097495555877686,
+      "learning_rate": 3.8678712591756074e-05,
+      "loss": 0.3743,
       "step": 24510
     },
     {
       "epoch": 4.615095049877659,
-      "grad_norm": 17.739852905273438,
-      "learning_rate": 2.3094297007340485e-06,
-      "loss": 0.9373,
+      "grad_norm": 5.164699077606201,
+      "learning_rate": 3.8490495012234144e-05,
+      "loss": 0.8332,
       "step": 24520
     },
     {
       "epoch": 4.616977225672878,
-      "grad_norm": 9.197490692138672,
-      "learning_rate": 2.298136645962733e-06,
-      "loss": 0.3972,
+      "grad_norm": 1.3957386016845703,
+      "learning_rate": 3.8302277432712214e-05,
+      "loss": 0.532,
       "step": 24530
     },
     {
       "epoch": 4.6188594014680975,
-      "grad_norm": 16.6794490814209,
-      "learning_rate": 2.2868435911914173e-06,
-      "loss": 0.3707,
+      "grad_norm": 6.435964107513428,
+      "learning_rate": 3.811405985319029e-05,
+      "loss": 0.5485,
       "step": 24540
     },
     {
       "epoch": 4.620741577263316,
-      "grad_norm": 0.1481352001428604,
-      "learning_rate": 2.2755505364201017e-06,
-      "loss": 0.4161,
+      "grad_norm": 0.23536163568496704,
+      "learning_rate": 3.792584227366836e-05,
+      "loss": 0.5219,
       "step": 24550
     },
     {
       "epoch": 4.6226237530585355,
-      "grad_norm": 0.12366284430027008,
-      "learning_rate": 2.264257481648786e-06,
-      "loss": 0.4102,
+      "grad_norm": 0.12256874889135361,
+      "learning_rate": 3.7737624694146436e-05,
+      "loss": 0.3599,
       "step": 24560
     },
     {
       "epoch": 4.624505928853755,
-      "grad_norm": 3.985370397567749,
-      "learning_rate": 2.2529644268774704e-06,
-      "loss": 0.6824,
+      "grad_norm": 0.9305835366249084,
+      "learning_rate": 3.7549407114624506e-05,
+      "loss": 0.6705,
       "step": 24570
     },
     {
       "epoch": 4.626388104648974,
-      "grad_norm": 4.121525764465332,
-      "learning_rate": 2.241671372106155e-06,
-      "loss": 0.4749,
+      "grad_norm": 1.9154576063156128,
+      "learning_rate": 3.7361189535102575e-05,
+      "loss": 0.5016,
       "step": 24580
     },
     {
       "epoch": 4.628270280444194,
-      "grad_norm": 3.955786943435669,
-      "learning_rate": 2.2303783173348388e-06,
-      "loss": 0.4139,
+      "grad_norm": 2.5845377445220947,
+      "learning_rate": 3.717297195558065e-05,
+      "loss": 0.4827,
       "step": 24590
     },
     {
       "epoch": 4.630152456239413,
-      "grad_norm": 17.2208251953125,
-      "learning_rate": 2.219085262563523e-06,
-      "loss": 0.7047,
+      "grad_norm": 2.6821815967559814,
+      "learning_rate": 3.698475437605872e-05,
+      "loss": 0.7094,
       "step": 24600
     },
     {
       "epoch": 4.632034632034632,
-      "grad_norm": 7.985387325286865,
-      "learning_rate": 2.207792207792208e-06,
-      "loss": 0.4245,
+      "grad_norm": 1.9128459692001343,
+      "learning_rate": 3.67965367965368e-05,
+      "loss": 0.4997,
       "step": 24610
     },
     {
       "epoch": 4.633916807829851,
-      "grad_norm": 10.840561866760254,
-      "learning_rate": 2.1964991530208923e-06,
-      "loss": 0.451,
+      "grad_norm": 4.187811851501465,
+      "learning_rate": 3.6608319217014874e-05,
+      "loss": 0.6142,
       "step": 24620
     },
     {
       "epoch": 4.635798983625071,
-      "grad_norm": 1.2460243701934814,
-      "learning_rate": 2.1852060982495767e-06,
-      "loss": 0.5374,
+      "grad_norm": 0.5653007626533508,
+      "learning_rate": 3.6420101637492944e-05,
+      "loss": 0.5527,
       "step": 24630
     },
     {
       "epoch": 4.63768115942029,
-      "grad_norm": 7.391758918762207,
-      "learning_rate": 2.173913043478261e-06,
-      "loss": 0.5306,
+      "grad_norm": 4.011232852935791,
+      "learning_rate": 3.6231884057971014e-05,
+      "loss": 0.5676,
       "step": 24640
     },
     {
       "epoch": 4.6395633352155095,
-      "grad_norm": 17.164552688598633,
-      "learning_rate": 2.1626199887069455e-06,
-      "loss": 0.4736,
+      "grad_norm": 1.5307304859161377,
+      "learning_rate": 3.604366647844909e-05,
+      "loss": 0.4625,
       "step": 24650
     },
     {
       "epoch": 4.641445511010728,
-      "grad_norm": 15.955860137939453,
-      "learning_rate": 2.1513269339356294e-06,
-      "loss": 0.7017,
+      "grad_norm": 3.410989761352539,
+      "learning_rate": 3.585544889892716e-05,
+      "loss": 0.7372,
       "step": 24660
     },
     {
       "epoch": 4.643327686805947,
-      "grad_norm": 6.390231609344482,
-      "learning_rate": 2.140033879164314e-06,
-      "loss": 0.6922,
+      "grad_norm": 4.534970283508301,
+      "learning_rate": 3.5667231319405236e-05,
+      "loss": 0.668,
       "step": 24670
     },
     {
       "epoch": 4.645209862601167,
-      "grad_norm": 35.61813735961914,
-      "learning_rate": 2.128740824392998e-06,
-      "loss": 0.7417,
+      "grad_norm": 5.051069259643555,
+      "learning_rate": 3.5479013739883306e-05,
+      "loss": 0.6868,
       "step": 24680
     },
     {
       "epoch": 4.647092038396386,
-      "grad_norm": 23.050209045410156,
-      "learning_rate": 2.1174477696216826e-06,
-      "loss": 0.6244,
+      "grad_norm": 2.785597085952759,
+      "learning_rate": 3.5290796160361375e-05,
+      "loss": 0.6126,
       "step": 24690
     },
     {
       "epoch": 4.648974214191606,
-      "grad_norm": 8.97560977935791,
-      "learning_rate": 2.106154714850367e-06,
-      "loss": 0.6843,
+      "grad_norm": 2.1288540363311768,
+      "learning_rate": 3.510257858083945e-05,
+      "loss": 0.7734,
       "step": 24700
     },
     {
       "epoch": 4.650856389986825,
-      "grad_norm": 5.241304874420166,
-      "learning_rate": 2.0948616600790514e-06,
-      "loss": 0.4725,
+      "grad_norm": 1.6902096271514893,
+      "learning_rate": 3.491436100131752e-05,
+      "loss": 0.5791,
       "step": 24710
     },
     {
       "epoch": 4.652738565782044,
-      "grad_norm": 15.338078498840332,
-      "learning_rate": 2.0835686053077357e-06,
-      "loss": 0.6333,
+      "grad_norm": 4.554157257080078,
+      "learning_rate": 3.472614342179559e-05,
+      "loss": 0.7908,
       "step": 24720
     },
     {
       "epoch": 4.654620741577263,
-      "grad_norm": 31.047271728515625,
-      "learning_rate": 2.0722755505364205e-06,
-      "loss": 0.2574,
+      "grad_norm": 2.3936281204223633,
+      "learning_rate": 3.4537925842273674e-05,
+      "loss": 0.3783,
       "step": 24730
     },
     {
       "epoch": 4.656502917372483,
-      "grad_norm": 39.24232864379883,
-      "learning_rate": 2.0609824957651045e-06,
-      "loss": 0.3953,
+      "grad_norm": 3.950031280517578,
+      "learning_rate": 3.4349708262751744e-05,
+      "loss": 0.4945,
       "step": 24740
     },
     {
       "epoch": 4.658385093167702,
-      "grad_norm": 6.512187957763672,
-      "learning_rate": 2.049689440993789e-06,
-      "loss": 0.398,
+      "grad_norm": 2.500994920730591,
+      "learning_rate": 3.4161490683229814e-05,
+      "loss": 0.637,
       "step": 24750
     },
     {
       "epoch": 4.660267268962921,
-      "grad_norm": 10.04133415222168,
-      "learning_rate": 2.0383963862224733e-06,
-      "loss": 0.6215,
+      "grad_norm": 2.361637592315674,
+      "learning_rate": 3.397327310370789e-05,
+      "loss": 0.6903,
       "step": 24760
     },
     {
       "epoch": 4.662149444758141,
-      "grad_norm": 20.212570190429688,
-      "learning_rate": 2.0271033314511576e-06,
-      "loss": 0.8138,
+      "grad_norm": 2.957148551940918,
+      "learning_rate": 3.378505552418596e-05,
+      "loss": 0.8531,
       "step": 24770
     },
     {
       "epoch": 4.664031620553359,
-      "grad_norm": 8.476882934570312,
-      "learning_rate": 2.015810276679842e-06,
-      "loss": 0.7214,
+      "grad_norm": 2.1123647689819336,
+      "learning_rate": 3.359683794466403e-05,
+      "loss": 0.7265,
       "step": 24780
     },
     {
       "epoch": 4.665913796348579,
-      "grad_norm": 11.082260131835938,
-      "learning_rate": 2.0045172219085264e-06,
-      "loss": 0.4978,
+      "grad_norm": 2.273115396499634,
+      "learning_rate": 3.3408620365142106e-05,
+      "loss": 0.651,
       "step": 24790
     },
     {
       "epoch": 4.667795972143798,
-      "grad_norm": 0.5434314012527466,
-      "learning_rate": 1.993224167137211e-06,
-      "loss": 0.72,
+      "grad_norm": 0.9176938533782959,
+      "learning_rate": 3.3220402785620175e-05,
+      "loss": 0.8676,
       "step": 24800
     },
     {
       "epoch": 4.669678147939018,
-      "grad_norm": 28.41150665283203,
-      "learning_rate": 1.9819311123658948e-06,
-      "loss": 0.6336,
+      "grad_norm": 5.138895511627197,
+      "learning_rate": 3.303218520609825e-05,
+      "loss": 0.6918,
       "step": 24810
     },
     {
       "epoch": 4.671560323734237,
-      "grad_norm": 1.0246213674545288,
-      "learning_rate": 1.970638057594579e-06,
-      "loss": 0.888,
+      "grad_norm": 1.913735270500183,
+      "learning_rate": 3.284396762657632e-05,
+      "loss": 0.8698,
       "step": 24820
     },
     {
       "epoch": 4.673442499529456,
-      "grad_norm": 15.647318840026855,
-      "learning_rate": 1.9593450028232635e-06,
-      "loss": 0.6784,
+      "grad_norm": 3.0344340801239014,
+      "learning_rate": 3.265575004705439e-05,
+      "loss": 0.7489,
       "step": 24830
     },
     {
       "epoch": 4.675324675324675,
-      "grad_norm": 16.091299057006836,
-      "learning_rate": 1.948051948051948e-06,
-      "loss": 0.2994,
+      "grad_norm": 3.871687889099121,
+      "learning_rate": 3.246753246753247e-05,
+      "loss": 0.3914,
       "step": 24840
     },
     {
       "epoch": 4.6772068511198945,
-      "grad_norm": 20.402158737182617,
-      "learning_rate": 1.9367588932806327e-06,
-      "loss": 0.4222,
+      "grad_norm": 7.366399765014648,
+      "learning_rate": 3.2279314888010544e-05,
+      "loss": 0.6409,
       "step": 24850
     },
     {
       "epoch": 4.679089026915114,
-      "grad_norm": 29.887250900268555,
-      "learning_rate": 1.925465838509317e-06,
-      "loss": 0.5632,
+      "grad_norm": 6.943544864654541,
+      "learning_rate": 3.2091097308488614e-05,
+      "loss": 0.6329,
       "step": 24860
     },
     {
       "epoch": 4.680971202710333,
-      "grad_norm": 0.39280015230178833,
-      "learning_rate": 1.9141727837380015e-06,
-      "loss": 0.3391,
+      "grad_norm": 0.23021714389324188,
+      "learning_rate": 3.190287972896669e-05,
+      "loss": 0.3462,
       "step": 24870
     },
     {
       "epoch": 4.682853378505552,
-      "grad_norm": 1.1423434019088745,
-      "learning_rate": 1.9028797289666856e-06,
-      "loss": 0.4052,
+      "grad_norm": 3.1440484523773193,
+      "learning_rate": 3.171466214944476e-05,
+      "loss": 0.5147,
       "step": 24880
     },
     {
       "epoch": 4.684735554300771,
-      "grad_norm": 0.18918712437152863,
-      "learning_rate": 1.89158667419537e-06,
-      "loss": 0.6111,
+      "grad_norm": 0.49845820665359497,
+      "learning_rate": 3.152644456992283e-05,
+      "loss": 0.8418,
       "step": 24890
     },
     {
       "epoch": 4.686617730095991,
-      "grad_norm": 6.821197986602783,
-      "learning_rate": 1.8802936194240542e-06,
-      "loss": 0.3858,
+      "grad_norm": 2.562613010406494,
+      "learning_rate": 3.1338226990400906e-05,
+      "loss": 0.4522,
       "step": 24900
     },
     {
       "epoch": 4.68849990589121,
-      "grad_norm": 18.688823699951172,
-      "learning_rate": 1.8690005646527386e-06,
-      "loss": 0.5372,
+      "grad_norm": 1.2562291622161865,
+      "learning_rate": 3.1150009410878975e-05,
+      "loss": 0.5723,
       "step": 24910
     },
     {
       "epoch": 4.69038208168643,
-      "grad_norm": 22.61110496520996,
-      "learning_rate": 1.857707509881423e-06,
-      "loss": 0.5846,
+      "grad_norm": 3.6861045360565186,
+      "learning_rate": 3.096179183135705e-05,
+      "loss": 0.6128,
       "step": 24920
     },
     {
       "epoch": 4.692264257481649,
-      "grad_norm": 14.270989418029785,
-      "learning_rate": 1.8464144551101073e-06,
-      "loss": 0.3677,
+      "grad_norm": 2.452535629272461,
+      "learning_rate": 3.077357425183512e-05,
+      "loss": 0.5187,
       "step": 24930
     },
     {
       "epoch": 4.6941464332768685,
-      "grad_norm": 20.81678581237793,
-      "learning_rate": 1.8351214003387917e-06,
-      "loss": 0.3929,
+      "grad_norm": 5.4972944259643555,
+      "learning_rate": 3.05853566723132e-05,
+      "loss": 0.4926,
       "step": 24940
     },
     {
       "epoch": 4.696028609072087,
-      "grad_norm": 0.7753129601478577,
-      "learning_rate": 1.823828345567476e-06,
-      "loss": 0.512,
+      "grad_norm": 0.3770606815814972,
+      "learning_rate": 3.0397139092791268e-05,
+      "loss": 0.5036,
       "step": 24950
     },
     {
       "epoch": 4.6979107848673065,
-      "grad_norm": 7.8035664558410645,
-      "learning_rate": 1.8125352907961605e-06,
-      "loss": 0.8553,
+      "grad_norm": 1.6760812997817993,
+      "learning_rate": 3.020892151326934e-05,
+      "loss": 0.7949,
       "step": 24960
     },
     {
       "epoch": 4.699792960662526,
-      "grad_norm": 9.686589241027832,
-      "learning_rate": 1.8012422360248449e-06,
-      "loss": 0.7138,
+      "grad_norm": 1.8957998752593994,
+      "learning_rate": 3.0020703933747414e-05,
+      "loss": 0.891,
       "step": 24970
     },
     {
       "epoch": 4.701675136457745,
-      "grad_norm": 0.7475610375404358,
-      "learning_rate": 1.789949181253529e-06,
-      "loss": 0.6173,
+      "grad_norm": 0.5944697260856628,
+      "learning_rate": 2.9832486354225483e-05,
+      "loss": 0.5906,
       "step": 24980
     },
     {
       "epoch": 4.703557312252965,
-      "grad_norm": 11.084076881408691,
-      "learning_rate": 1.7786561264822134e-06,
-      "loss": 0.353,
+      "grad_norm": 4.481252670288086,
+      "learning_rate": 2.9644268774703556e-05,
+      "loss": 0.4426,
       "step": 24990
     },
     {
       "epoch": 4.705439488048183,
-      "grad_norm": 33.30714416503906,
-      "learning_rate": 1.7673630717108978e-06,
-      "loss": 0.578,
+      "grad_norm": 4.642335891723633,
+      "learning_rate": 2.945605119518163e-05,
+      "loss": 0.5466,
       "step": 25000
     },
     {
       "epoch": 4.707321663843403,
-      "grad_norm": 35.408416748046875,
-      "learning_rate": 1.7560700169395822e-06,
-      "loss": 0.4082,
+      "grad_norm": 3.7608211040496826,
+      "learning_rate": 2.9267833615659706e-05,
+      "loss": 0.5645,
       "step": 25010
     },
     {
       "epoch": 4.709203839638622,
-      "grad_norm": 37.653106689453125,
-      "learning_rate": 1.7447769621682666e-06,
-      "loss": 0.5041,
+      "grad_norm": 4.015994071960449,
+      "learning_rate": 2.9079616036137775e-05,
+      "loss": 0.5362,
       "step": 25020
     },
     {
       "epoch": 4.711086015433842,
-      "grad_norm": 0.38172101974487305,
-      "learning_rate": 1.733483907396951e-06,
-      "loss": 0.6022,
+      "grad_norm": 0.06955814361572266,
+      "learning_rate": 2.889139845661585e-05,
+      "loss": 0.7009,
       "step": 25030
     },
     {
       "epoch": 4.712968191229061,
-      "grad_norm": 0.1132197305560112,
-      "learning_rate": 1.7221908526256353e-06,
-      "loss": 0.4891,
+      "grad_norm": 0.20994648337364197,
+      "learning_rate": 2.870318087709392e-05,
+      "loss": 0.4559,
       "step": 25040
     },
     {
       "epoch": 4.71485036702428,
-      "grad_norm": 24.087400436401367,
-      "learning_rate": 1.7108977978543195e-06,
-      "loss": 0.4906,
+      "grad_norm": 0.5177777409553528,
+      "learning_rate": 2.851496329757199e-05,
+      "loss": 0.5917,
       "step": 25050
     },
     {
       "epoch": 4.716732542819499,
-      "grad_norm": 9.231995582580566,
-      "learning_rate": 1.6996047430830039e-06,
-      "loss": 0.6238,
+      "grad_norm": 1.8041309118270874,
+      "learning_rate": 2.8326745718050064e-05,
+      "loss": 0.4951,
       "step": 25060
     },
     {
       "epoch": 4.7186147186147185,
-      "grad_norm": 22.86023712158203,
-      "learning_rate": 1.6883116883116885e-06,
-      "loss": 0.9478,
+      "grad_norm": 3.804542303085327,
+      "learning_rate": 2.813852813852814e-05,
+      "loss": 1.0139,
       "step": 25070
     },
     {
       "epoch": 4.720496894409938,
-      "grad_norm": 7.785012245178223,
-      "learning_rate": 1.6770186335403729e-06,
-      "loss": 0.5008,
+      "grad_norm": 3.876880168914795,
+      "learning_rate": 2.7950310559006214e-05,
+      "loss": 0.5698,
       "step": 25080
     },
     {
       "epoch": 4.722379070205157,
-      "grad_norm": 2.3250386714935303,
-      "learning_rate": 1.665725578769057e-06,
-      "loss": 0.7925,
+      "grad_norm": 3.0488855838775635,
+      "learning_rate": 2.7762092979484283e-05,
+      "loss": 0.9231,
       "step": 25090
     },
     {
       "epoch": 4.724261246000377,
-      "grad_norm": 12.125810623168945,
-      "learning_rate": 1.6544325239977414e-06,
-      "loss": 0.3744,
+      "grad_norm": 2.9211177825927734,
+      "learning_rate": 2.7573875399962356e-05,
+      "loss": 0.4906,
       "step": 25100
     },
     {
       "epoch": 4.726143421795595,
-      "grad_norm": 14.144658088684082,
-      "learning_rate": 1.6431394692264258e-06,
-      "loss": 0.4073,
+      "grad_norm": 2.8851497173309326,
+      "learning_rate": 2.738565782044043e-05,
+      "loss": 0.4443,
       "step": 25110
     },
     {
       "epoch": 4.728025597590815,
-      "grad_norm": 4.601728916168213,
-      "learning_rate": 1.63184641445511e-06,
-      "loss": 0.3743,
+      "grad_norm": 2.138963460922241,
+      "learning_rate": 2.7197440240918502e-05,
+      "loss": 0.3601,
       "step": 25120
     },
     {
       "epoch": 4.729907773386034,
-      "grad_norm": 9.717355728149414,
-      "learning_rate": 1.6205533596837946e-06,
-      "loss": 0.5983,
+      "grad_norm": 2.7832868099212646,
+      "learning_rate": 2.7009222661396575e-05,
+      "loss": 0.669,
       "step": 25130
     },
     {
       "epoch": 4.731789949181254,
-      "grad_norm": 13.010869026184082,
-      "learning_rate": 1.609260304912479e-06,
-      "loss": 0.4777,
+      "grad_norm": 3.739305019378662,
+      "learning_rate": 2.682100508187465e-05,
+      "loss": 0.5055,
       "step": 25140
     },
     {
       "epoch": 4.733672124976473,
-      "grad_norm": 10.841914176940918,
-      "learning_rate": 1.5979672501411633e-06,
-      "loss": 0.7914,
+      "grad_norm": 2.2495994567871094,
+      "learning_rate": 2.663278750235272e-05,
+      "loss": 0.8668,
       "step": 25150
     },
     {
       "epoch": 4.7355543007716925,
-      "grad_norm": 11.253789901733398,
-      "learning_rate": 1.5866741953698475e-06,
-      "loss": 0.5007,
+      "grad_norm": 2.521467924118042,
+      "learning_rate": 2.644456992283079e-05,
+      "loss": 0.5573,
       "step": 25160
     },
     {
       "epoch": 4.737436476566911,
-      "grad_norm": 20.718843460083008,
-      "learning_rate": 1.5753811405985319e-06,
-      "loss": 0.4821,
+      "grad_norm": 2.1538898944854736,
+      "learning_rate": 2.6256352343308864e-05,
+      "loss": 0.5449,
       "step": 25170
     },
     {
       "epoch": 4.7393186523621305,
-      "grad_norm": 15.470338821411133,
-      "learning_rate": 1.5640880858272163e-06,
-      "loss": 0.3737,
+      "grad_norm": 5.599278926849365,
+      "learning_rate": 2.6068134763786937e-05,
+      "loss": 0.4944,
       "step": 25180
     },
     {
       "epoch": 4.74120082815735,
-      "grad_norm": 0.15901845693588257,
-      "learning_rate": 1.5527950310559008e-06,
-      "loss": 0.1535,
+      "grad_norm": 1.8122804164886475,
+      "learning_rate": 2.5879917184265014e-05,
+      "loss": 0.3517,
       "step": 25190
     },
     {
       "epoch": 4.743083003952569,
-      "grad_norm": 17.537002563476562,
-      "learning_rate": 1.541501976284585e-06,
-      "loss": 0.3773,
+      "grad_norm": 3.613555669784546,
+      "learning_rate": 2.5691699604743083e-05,
+      "loss": 0.5506,
       "step": 25200
     },
     {
       "epoch": 4.744965179747789,
-      "grad_norm": 6.956587314605713,
-      "learning_rate": 1.5302089215132694e-06,
-      "loss": 0.3261,
+      "grad_norm": 3.1564066410064697,
+      "learning_rate": 2.5503482025221156e-05,
+      "loss": 0.3377,
       "step": 25210
     },
     {
       "epoch": 4.746847355543007,
-      "grad_norm": 5.067175388336182,
-      "learning_rate": 1.5189158667419538e-06,
-      "loss": 0.6587,
+      "grad_norm": 1.0180994272232056,
+      "learning_rate": 2.531526444569923e-05,
+      "loss": 0.7821,
       "step": 25220
     },
     {
       "epoch": 4.748729531338227,
-      "grad_norm": 25.866880416870117,
-      "learning_rate": 1.507622811970638e-06,
-      "loss": 0.6093,
+      "grad_norm": 1.7408175468444824,
+      "learning_rate": 2.5127046866177302e-05,
+      "loss": 0.6947,
       "step": 25230
     },
     {
       "epoch": 4.750611707133446,
-      "grad_norm": 2.593472957611084,
-      "learning_rate": 1.4963297571993223e-06,
-      "loss": 0.2919,
+      "grad_norm": 0.4839925467967987,
+      "learning_rate": 2.4938829286655372e-05,
+      "loss": 0.3432,
       "step": 25240
     },
     {
       "epoch": 4.752493882928666,
-      "grad_norm": 5.722588062286377,
-      "learning_rate": 1.485036702428007e-06,
-      "loss": 0.464,
+      "grad_norm": 3.0300328731536865,
+      "learning_rate": 2.475061170713345e-05,
+      "loss": 0.5659,
       "step": 25250
     },
     {
       "epoch": 4.754376058723885,
-      "grad_norm": 12.117194175720215,
-      "learning_rate": 1.4737436476566913e-06,
-      "loss": 0.2929,
+      "grad_norm": 1.7946454286575317,
+      "learning_rate": 2.456239412761152e-05,
+      "loss": 0.3518,
       "step": 25260
     },
     {
       "epoch": 4.7562582345191045,
-      "grad_norm": 18.254680633544922,
-      "learning_rate": 1.4624505928853755e-06,
-      "loss": 0.757,
+      "grad_norm": 2.8185391426086426,
+      "learning_rate": 2.437417654808959e-05,
+      "loss": 0.7663,
       "step": 25270
     },
     {
       "epoch": 4.758140410314323,
-      "grad_norm": 0.17077714204788208,
-      "learning_rate": 1.4511575381140599e-06,
-      "loss": 0.4409,
+      "grad_norm": 0.3781205713748932,
+      "learning_rate": 2.4185958968567664e-05,
+      "loss": 0.65,
       "step": 25280
     },
     {
       "epoch": 4.760022586109542,
-      "grad_norm": 11.023554801940918,
-      "learning_rate": 1.4398644833427442e-06,
-      "loss": 0.6109,
+      "grad_norm": 3.49873423576355,
+      "learning_rate": 2.3997741389045737e-05,
+      "loss": 0.6592,
       "step": 25290
     },
     {
       "epoch": 4.761904761904762,
-      "grad_norm": 11.318467140197754,
-      "learning_rate": 1.4285714285714286e-06,
-      "loss": 0.41,
+      "grad_norm": 2.9731030464172363,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.4748,
       "step": 25300
     },
     {
       "epoch": 4.763786937699981,
-      "grad_norm": 0.2806694507598877,
-      "learning_rate": 1.417278373800113e-06,
-      "loss": 0.3642,
+      "grad_norm": 1.2424784898757935,
+      "learning_rate": 2.3621306230001883e-05,
+      "loss": 0.3967,
       "step": 25310
     },
     {
       "epoch": 4.765669113495201,
-      "grad_norm": 3.844435930252075,
-      "learning_rate": 1.4059853190287974e-06,
-      "loss": 0.5849,
+      "grad_norm": 2.284749746322632,
+      "learning_rate": 2.3433088650479956e-05,
+      "loss": 0.6169,
       "step": 25320
     },
     {
       "epoch": 4.76755128929042,
-      "grad_norm": 6.41024923324585,
-      "learning_rate": 1.3946922642574818e-06,
-      "loss": 0.4466,
+      "grad_norm": 4.062412738800049,
+      "learning_rate": 2.324487107095803e-05,
+      "loss": 0.4828,
       "step": 25330
     },
     {
       "epoch": 4.769433465085639,
-      "grad_norm": 12.630584716796875,
-      "learning_rate": 1.383399209486166e-06,
-      "loss": 0.4274,
+      "grad_norm": 1.2760084867477417,
+      "learning_rate": 2.30566534914361e-05,
+      "loss": 0.4661,
       "step": 25340
     },
     {
       "epoch": 4.771315640880858,
-      "grad_norm": 14.851526260375977,
-      "learning_rate": 1.3721061547148503e-06,
-      "loss": 0.5182,
+      "grad_norm": 2.2040622234344482,
+      "learning_rate": 2.2868435911914172e-05,
+      "loss": 0.5224,
       "step": 25350
     },
     {
       "epoch": 4.773197816676078,
-      "grad_norm": 10.023591995239258,
-      "learning_rate": 1.3608130999435347e-06,
-      "loss": 0.7274,
+      "grad_norm": 2.973165273666382,
+      "learning_rate": 2.2680218332392245e-05,
+      "loss": 0.7409,
       "step": 25360
     },
     {
       "epoch": 4.775079992471297,
-      "grad_norm": 32.864810943603516,
-      "learning_rate": 1.3495200451722193e-06,
-      "loss": 0.464,
+      "grad_norm": 4.954020023345947,
+      "learning_rate": 2.249200075287032e-05,
+      "loss": 0.3633,
       "step": 25370
     },
     {
       "epoch": 4.776962168266516,
-      "grad_norm": 17.206785202026367,
-      "learning_rate": 1.3382269904009035e-06,
-      "loss": 0.4297,
+      "grad_norm": 3.838167905807495,
+      "learning_rate": 2.230378317334839e-05,
+      "loss": 0.5015,
       "step": 25380
     },
     {
       "epoch": 4.778844344061735,
-      "grad_norm": 11.58726978302002,
-      "learning_rate": 1.3269339356295879e-06,
-      "loss": 0.4302,
+      "grad_norm": 2.710631847381592,
+      "learning_rate": 2.2115565593826464e-05,
+      "loss": 0.6205,
       "step": 25390
     },
     {
       "epoch": 4.780726519856954,
-      "grad_norm": 5.545811176300049,
-      "learning_rate": 1.3156408808582722e-06,
-      "loss": 0.4928,
+      "grad_norm": 1.263107419013977,
+      "learning_rate": 2.1927348014304537e-05,
+      "loss": 0.5492,
       "step": 25400
     },
     {
       "epoch": 4.782608695652174,
-      "grad_norm": 14.361126899719238,
-      "learning_rate": 1.3043478260869566e-06,
-      "loss": 0.6301,
+      "grad_norm": 2.989341974258423,
+      "learning_rate": 2.173913043478261e-05,
+      "loss": 0.5812,
       "step": 25410
     },
     {
       "epoch": 4.784490871447393,
-      "grad_norm": 16.078994750976562,
-      "learning_rate": 1.2930547713156408e-06,
-      "loss": 0.8214,
+      "grad_norm": 1.7873269319534302,
+      "learning_rate": 2.155091285526068e-05,
+      "loss": 0.7841,
       "step": 25420
     },
     {
       "epoch": 4.786373047242613,
-      "grad_norm": 18.931100845336914,
-      "learning_rate": 1.2817617165443252e-06,
-      "loss": 0.8551,
+      "grad_norm": 28.162975311279297,
+      "learning_rate": 2.1362695275738753e-05,
+      "loss": 0.8946,
       "step": 25430
     },
     {
       "epoch": 4.788255223037831,
-      "grad_norm": 12.793909072875977,
-      "learning_rate": 1.2704686617730098e-06,
-      "loss": 0.8418,
+      "grad_norm": 2.6471946239471436,
+      "learning_rate": 2.117447769621683e-05,
+      "loss": 0.8679,
       "step": 25440
     },
     {
       "epoch": 4.790137398833051,
-      "grad_norm": 9.514110565185547,
-      "learning_rate": 1.259175607001694e-06,
-      "loss": 0.4481,
+      "grad_norm": 5.760558128356934,
+      "learning_rate": 2.09862601166949e-05,
+      "loss": 0.6001,
       "step": 25450
     },
     {
       "epoch": 4.79201957462827,
-      "grad_norm": 2.824892997741699,
-      "learning_rate": 1.2478825522303783e-06,
-      "loss": 0.7443,
+      "grad_norm": 2.5529234409332275,
+      "learning_rate": 2.0798042537172972e-05,
+      "loss": 0.8864,
       "step": 25460
     },
     {
       "epoch": 4.7939017504234895,
-      "grad_norm": 0.1960848569869995,
-      "learning_rate": 1.2365894974590627e-06,
-      "loss": 0.5792,
+      "grad_norm": 0.7746294140815735,
+      "learning_rate": 2.0609824957651045e-05,
+      "loss": 0.6143,
       "step": 25470
     },
     {
       "epoch": 4.795783926218709,
-      "grad_norm": 0.339618057012558,
-      "learning_rate": 1.225296442687747e-06,
-      "loss": 0.5427,
+      "grad_norm": 0.07354772090911865,
+      "learning_rate": 2.0421607378129118e-05,
+      "loss": 0.4895,
       "step": 25480
     },
     {
       "epoch": 4.797666102013928,
-      "grad_norm": 6.723775863647461,
-      "learning_rate": 1.2140033879164313e-06,
-      "loss": 0.3267,
+      "grad_norm": 1.4285778999328613,
+      "learning_rate": 2.0233389798607188e-05,
+      "loss": 0.3116,
       "step": 25490
     },
     {
       "epoch": 4.799548277809148,
-      "grad_norm": 1.1799036264419556,
-      "learning_rate": 1.2027103331451158e-06,
-      "loss": 0.3805,
+      "grad_norm": 1.7135233879089355,
+      "learning_rate": 2.0045172219085264e-05,
+      "loss": 0.4581,
       "step": 25500
     },
     {
       "epoch": 4.801430453604366,
-      "grad_norm": 3.111079216003418,
-      "learning_rate": 1.1914172783738002e-06,
-      "loss": 0.3117,
+      "grad_norm": 1.4032816886901855,
+      "learning_rate": 1.9856954639563337e-05,
+      "loss": 0.3102,
       "step": 25510
     },
     {
       "epoch": 4.803312629399586,
-      "grad_norm": 0.1729825735092163,
-      "learning_rate": 1.1801242236024846e-06,
-      "loss": 0.4701,
+      "grad_norm": 0.2566004693508148,
+      "learning_rate": 1.9668737060041407e-05,
+      "loss": 0.4983,
       "step": 25520
     },
     {
       "epoch": 4.805194805194805,
-      "grad_norm": 8.52800464630127,
-      "learning_rate": 1.1688311688311688e-06,
-      "loss": 0.4286,
+      "grad_norm": 1.9948326349258423,
+      "learning_rate": 1.948051948051948e-05,
+      "loss": 0.5406,
       "step": 25530
     },
     {
       "epoch": 4.807076980990025,
-      "grad_norm": 10.847475051879883,
-      "learning_rate": 1.1575381140598532e-06,
-      "loss": 0.3005,
+      "grad_norm": 1.6091513633728027,
+      "learning_rate": 1.9292301900997553e-05,
+      "loss": 0.3729,
       "step": 25540
     },
     {
       "epoch": 4.808959156785244,
-      "grad_norm": 11.9356689453125,
-      "learning_rate": 1.1462450592885375e-06,
-      "loss": 0.4045,
+      "grad_norm": 2.1368279457092285,
+      "learning_rate": 1.9104084321475626e-05,
+      "loss": 0.4786,
       "step": 25550
     },
     {
       "epoch": 4.810841332580463,
-      "grad_norm": 7.127353668212891,
-      "learning_rate": 1.134952004517222e-06,
-      "loss": 0.5741,
+      "grad_norm": 2.6382944583892822,
+      "learning_rate": 1.89158667419537e-05,
+      "loss": 0.5737,
       "step": 25560
     },
     {
       "epoch": 4.812723508375682,
-      "grad_norm": 16.197805404663086,
-      "learning_rate": 1.1236589497459063e-06,
-      "loss": 0.5069,
+      "grad_norm": 4.222015857696533,
+      "learning_rate": 1.8727649162431772e-05,
+      "loss": 0.5957,
       "step": 25570
     },
     {
       "epoch": 4.8146056841709015,
-      "grad_norm": 0.5227009057998657,
-      "learning_rate": 1.1123658949745907e-06,
-      "loss": 0.4081,
+      "grad_norm": 4.877696514129639,
+      "learning_rate": 1.8539431582909845e-05,
+      "loss": 0.5378,
       "step": 25580
     },
     {
       "epoch": 4.816487859966121,
-      "grad_norm": 16.148591995239258,
-      "learning_rate": 1.101072840203275e-06,
-      "loss": 0.5699,
+      "grad_norm": 2.3263192176818848,
+      "learning_rate": 1.8351214003387918e-05,
+      "loss": 0.6127,
       "step": 25590
     },
     {
       "epoch": 4.81837003576134,
-      "grad_norm": 1.3249871730804443,
-      "learning_rate": 1.0897797854319592e-06,
-      "loss": 0.4742,
+      "grad_norm": 5.47594690322876,
+      "learning_rate": 1.8162996423865988e-05,
+      "loss": 0.5668,
       "step": 25600
     },
     {
       "epoch": 4.820252211556559,
-      "grad_norm": 11.252976417541504,
-      "learning_rate": 1.0784867306606436e-06,
-      "loss": 0.5348,
+      "grad_norm": 0.9088131785392761,
+      "learning_rate": 1.797477884434406e-05,
+      "loss": 0.5909,
       "step": 25610
     },
     {
       "epoch": 4.822134387351778,
-      "grad_norm": 9.052772521972656,
-      "learning_rate": 1.0671936758893282e-06,
-      "loss": 0.5361,
+      "grad_norm": 3.0957508087158203,
+      "learning_rate": 1.7786561264822137e-05,
+      "loss": 0.6015,
       "step": 25620
     },
     {
       "epoch": 4.824016563146998,
-      "grad_norm": 34.760986328125,
-      "learning_rate": 1.0559006211180126e-06,
-      "loss": 0.4909,
+      "grad_norm": 6.333992004394531,
+      "learning_rate": 1.7598343685300207e-05,
+      "loss": 0.6555,
       "step": 25630
     },
     {
       "epoch": 4.825898738942217,
-      "grad_norm": 18.605308532714844,
-      "learning_rate": 1.0446075663466968e-06,
-      "loss": 0.3664,
+      "grad_norm": 2.639153480529785,
+      "learning_rate": 1.741012610577828e-05,
+      "loss": 0.3782,
       "step": 25640
     },
     {
       "epoch": 4.827780914737437,
-      "grad_norm": 1.8343195915222168,
-      "learning_rate": 1.0333145115753812e-06,
-      "loss": 0.2793,
+      "grad_norm": 0.8514479398727417,
+      "learning_rate": 1.7221908526256353e-05,
+      "loss": 0.2632,
       "step": 25650
     },
     {
       "epoch": 4.829663090532656,
-      "grad_norm": 6.831780433654785,
-      "learning_rate": 1.0220214568040655e-06,
-      "loss": 0.4601,
+      "grad_norm": 2.4760193824768066,
+      "learning_rate": 1.7033690946734426e-05,
+      "loss": 0.506,
       "step": 25660
     },
     {
       "epoch": 4.8315452663278755,
-      "grad_norm": 9.190847396850586,
-      "learning_rate": 1.0107284020327497e-06,
-      "loss": 0.3852,
+      "grad_norm": 1.6171988248825073,
+      "learning_rate": 1.6845473367212496e-05,
+      "loss": 0.4352,
       "step": 25670
     },
     {
       "epoch": 4.833427442123094,
-      "grad_norm": 10.883723258972168,
-      "learning_rate": 9.994353472614343e-07,
-      "loss": 0.3296,
+      "grad_norm": 3.2398006916046143,
+      "learning_rate": 1.6657255787690572e-05,
+      "loss": 0.4091,
       "step": 25680
     },
     {
       "epoch": 4.8353096179183135,
-      "grad_norm": 11.370898246765137,
-      "learning_rate": 9.881422924901187e-07,
-      "loss": 0.3896,
+      "grad_norm": 3.9767098426818848,
+      "learning_rate": 1.6469038208168645e-05,
+      "loss": 0.4436,
       "step": 25690
     },
     {
       "epoch": 4.837191793713533,
-      "grad_norm": 0.8723617792129517,
-      "learning_rate": 9.76849237718803e-07,
-      "loss": 0.4617,
+      "grad_norm": 0.5642136931419373,
+      "learning_rate": 1.6280820628646715e-05,
+      "loss": 0.4776,
       "step": 25700
     },
     {
       "epoch": 4.839073969508752,
-      "grad_norm": 2.9830756187438965,
-      "learning_rate": 9.655561829474872e-07,
-      "loss": 0.3264,
+      "grad_norm": 1.381508469581604,
+      "learning_rate": 1.6092603049124788e-05,
+      "loss": 0.412,
       "step": 25710
     },
     {
       "epoch": 4.840956145303972,
-      "grad_norm": 9.031828880310059,
-      "learning_rate": 9.542631281761716e-07,
-      "loss": 0.5838,
+      "grad_norm": 4.1336750984191895,
+      "learning_rate": 1.590438546960286e-05,
+      "loss": 0.6967,
       "step": 25720
     },
     {
       "epoch": 4.84283832109919,
-      "grad_norm": 37.80723571777344,
-      "learning_rate": 9.42970073404856e-07,
-      "loss": 0.2824,
+      "grad_norm": 4.109206199645996,
+      "learning_rate": 1.5716167890080934e-05,
+      "loss": 0.3631,
       "step": 25730
     },
     {
       "epoch": 4.84472049689441,
-      "grad_norm": 9.098519325256348,
-      "learning_rate": 9.316770186335404e-07,
-      "loss": 0.4279,
+      "grad_norm": 1.8715598583221436,
+      "learning_rate": 1.5527950310559007e-05,
+      "loss": 0.5251,
       "step": 25740
     },
     {
       "epoch": 4.846602672689629,
-      "grad_norm": 12.211234092712402,
-      "learning_rate": 9.203839638622247e-07,
-      "loss": 0.5721,
+      "grad_norm": 1.6625672578811646,
+      "learning_rate": 1.533973273103708e-05,
+      "loss": 0.4531,
       "step": 25750
     },
     {
       "epoch": 4.848484848484849,
-      "grad_norm": 1.9672722816467285,
-      "learning_rate": 9.090909090909091e-07,
-      "loss": 0.432,
+      "grad_norm": 1.0363481044769287,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.5578,
       "step": 25760
     },
     {
       "epoch": 4.850367024280068,
-      "grad_norm": 18.3715763092041,
-      "learning_rate": 8.977978543195934e-07,
-      "loss": 0.5275,
+      "grad_norm": 12.811112403869629,
+      "learning_rate": 1.4963297571993224e-05,
+      "loss": 0.6469,
       "step": 25770
     },
     {
       "epoch": 4.852249200075287,
-      "grad_norm": 6.322066307067871,
-      "learning_rate": 8.865047995482778e-07,
-      "loss": 0.4312,
+      "grad_norm": 2.2655227184295654,
+      "learning_rate": 1.4775079992471297e-05,
+      "loss": 0.4671,
       "step": 25780
     },
     {
       "epoch": 4.854131375870506,
-      "grad_norm": 13.303924560546875,
-      "learning_rate": 8.752117447769622e-07,
-      "loss": 0.5374,
+      "grad_norm": 1.3414403200149536,
+      "learning_rate": 1.458686241294937e-05,
+      "loss": 0.5239,
       "step": 25790
     },
     {
       "epoch": 4.8560135516657255,
-      "grad_norm": 18.152463912963867,
-      "learning_rate": 8.639186900056466e-07,
-      "loss": 0.4755,
+      "grad_norm": 3.1365103721618652,
+      "learning_rate": 1.4398644833427442e-05,
+      "loss": 0.4356,
       "step": 25800
     },
     {
       "epoch": 4.857895727460945,
-      "grad_norm": 1.1153311729431152,
-      "learning_rate": 8.526256352343308e-07,
-      "loss": 0.5785,
+      "grad_norm": 0.4507659673690796,
+      "learning_rate": 1.4210427253905515e-05,
+      "loss": 0.5114,
       "step": 25810
     },
     {
       "epoch": 4.859777903256164,
-      "grad_norm": 17.98719024658203,
-      "learning_rate": 8.413325804630153e-07,
-      "loss": 0.4314,
+      "grad_norm": 2.2164368629455566,
+      "learning_rate": 1.4022209674383588e-05,
+      "loss": 0.4448,
       "step": 25820
     },
     {
       "epoch": 4.861660079051384,
-      "grad_norm": 5.713428020477295,
-      "learning_rate": 8.300395256916996e-07,
-      "loss": 0.5436,
+      "grad_norm": 1.65787672996521,
+      "learning_rate": 1.383399209486166e-05,
+      "loss": 0.5972,
       "step": 25830
     },
     {
       "epoch": 4.863542254846602,
-      "grad_norm": 17.330768585205078,
-      "learning_rate": 8.18746470920384e-07,
-      "loss": 0.3292,
+      "grad_norm": 2.5747125148773193,
+      "learning_rate": 1.3645774515339732e-05,
+      "loss": 0.3413,
       "step": 25840
     },
     {
       "epoch": 4.865424430641822,
-      "grad_norm": 0.1503809541463852,
-      "learning_rate": 8.074534161490684e-07,
-      "loss": 0.575,
+      "grad_norm": 0.23792870342731476,
+      "learning_rate": 1.3457556935817807e-05,
+      "loss": 0.496,
       "step": 25850
     },
     {
       "epoch": 4.867306606437041,
-      "grad_norm": 29.405603408813477,
-      "learning_rate": 7.961603613777526e-07,
-      "loss": 0.7215,
+      "grad_norm": 8.1236572265625,
+      "learning_rate": 1.3269339356295878e-05,
+      "loss": 0.9485,
       "step": 25860
     },
     {
       "epoch": 4.869188782232261,
-      "grad_norm": 5.5133209228515625,
-      "learning_rate": 7.84867306606437e-07,
-      "loss": 0.2398,
+      "grad_norm": 2.197131633758545,
+      "learning_rate": 1.3081121776773951e-05,
+      "loss": 0.4668,
       "step": 25870
     },
     {
       "epoch": 4.87107095802748,
-      "grad_norm": 10.060201644897461,
-      "learning_rate": 7.735742518351214e-07,
-      "loss": 0.7002,
+      "grad_norm": 3.55192494392395,
+      "learning_rate": 1.2892904197252024e-05,
+      "loss": 0.7436,
       "step": 25880
     },
     {
       "epoch": 4.8729531338226995,
-      "grad_norm": 31.25115966796875,
-      "learning_rate": 7.622811970638058e-07,
-      "loss": 0.5119,
+      "grad_norm": 2.4934020042419434,
+      "learning_rate": 1.2704686617730096e-05,
+      "loss": 0.5262,
       "step": 25890
     },
     {
       "epoch": 4.874835309617918,
-      "grad_norm": 11.812898635864258,
-      "learning_rate": 7.509881422924901e-07,
-      "loss": 0.622,
+      "grad_norm": 3.092714786529541,
+      "learning_rate": 1.2516469038208169e-05,
+      "loss": 0.6179,
       "step": 25900
     },
     {
       "epoch": 4.876717485413137,
-      "grad_norm": 6.675577163696289,
-      "learning_rate": 7.396950875211746e-07,
-      "loss": 0.3038,
+      "grad_norm": 2.5720813274383545,
+      "learning_rate": 1.2328251458686242e-05,
+      "loss": 0.4667,
       "step": 25910
     },
     {
       "epoch": 4.878599661208357,
-      "grad_norm": 9.983612060546875,
-      "learning_rate": 7.284020327498588e-07,
-      "loss": 0.7158,
+      "grad_norm": 5.265648365020752,
+      "learning_rate": 1.2140033879164315e-05,
+      "loss": 0.7364,
       "step": 25920
     },
     {
       "epoch": 4.880481837003576,
-      "grad_norm": 8.77939510345459,
-      "learning_rate": 7.171089779785432e-07,
-      "loss": 0.7233,
+      "grad_norm": 3.7148709297180176,
+      "learning_rate": 1.1951816299642386e-05,
+      "loss": 0.7846,
       "step": 25930
     },
     {
       "epoch": 4.882364012798796,
-      "grad_norm": 11.423147201538086,
-      "learning_rate": 7.058159232072276e-07,
-      "loss": 0.8275,
+      "grad_norm": 3.3122756481170654,
+      "learning_rate": 1.176359872012046e-05,
+      "loss": 0.8727,
       "step": 25940
     },
     {
       "epoch": 4.884246188594014,
-      "grad_norm": 11.282443046569824,
-      "learning_rate": 6.94522868435912e-07,
-      "loss": 0.6037,
+      "grad_norm": 3.0102932453155518,
+      "learning_rate": 1.1575381140598532e-05,
+      "loss": 0.6746,
       "step": 25950
     },
     {
       "epoch": 4.886128364389234,
-      "grad_norm": 34.89209747314453,
-      "learning_rate": 6.832298136645963e-07,
-      "loss": 0.6057,
+      "grad_norm": 3.3670566082000732,
+      "learning_rate": 1.1387163561076605e-05,
+      "loss": 0.5846,
       "step": 25960
     },
     {
       "epoch": 4.888010540184453,
-      "grad_norm": 13.334453582763672,
-      "learning_rate": 6.719367588932806e-07,
-      "loss": 0.5336,
+      "grad_norm": 3.1409409046173096,
+      "learning_rate": 1.1198945981554678e-05,
+      "loss": 0.5595,
       "step": 25970
     },
     {
       "epoch": 4.889892715979673,
-      "grad_norm": 43.71876525878906,
-      "learning_rate": 6.60643704121965e-07,
-      "loss": 0.4498,
+      "grad_norm": 5.394914627075195,
+      "learning_rate": 1.101072840203275e-05,
+      "loss": 0.4697,
       "step": 25980
     },
     {
       "epoch": 4.891774891774892,
-      "grad_norm": 4.845729351043701,
-      "learning_rate": 6.493506493506493e-07,
-      "loss": 0.4448,
+      "grad_norm": 1.916937232017517,
+      "learning_rate": 1.0822510822510823e-05,
+      "loss": 0.5682,
       "step": 25990
     },
     {
       "epoch": 4.893657067570111,
-      "grad_norm": 10.815642356872559,
-      "learning_rate": 6.380575945793338e-07,
-      "loss": 0.5838,
+      "grad_norm": 2.7022321224212646,
+      "learning_rate": 1.0634293242988896e-05,
+      "loss": 0.453,
       "step": 26000
     },
     {
       "epoch": 4.89553924336533,
-      "grad_norm": 17.238981246948242,
-      "learning_rate": 6.267645398080181e-07,
-      "loss": 0.444,
+      "grad_norm": 6.567044734954834,
+      "learning_rate": 1.0446075663466969e-05,
+      "loss": 0.5277,
       "step": 26010
     },
     {
       "epoch": 4.897421419160549,
-      "grad_norm": 10.10824966430664,
-      "learning_rate": 6.154714850367024e-07,
-      "loss": 0.6663,
+      "grad_norm": 2.0348784923553467,
+      "learning_rate": 1.025785808394504e-05,
+      "loss": 0.7916,
       "step": 26020
     },
     {
       "epoch": 4.899303594955769,
-      "grad_norm": 33.35478973388672,
-      "learning_rate": 6.041784302653868e-07,
-      "loss": 0.4297,
+      "grad_norm": 6.761394023895264,
+      "learning_rate": 1.0069640504423115e-05,
+      "loss": 0.4096,
       "step": 26030
     },
     {
       "epoch": 4.901185770750988,
-      "grad_norm": 0.8835696578025818,
-      "learning_rate": 5.928853754940712e-07,
-      "loss": 0.6774,
+      "grad_norm": 1.9423185586929321,
+      "learning_rate": 9.881422924901186e-06,
+      "loss": 0.77,
       "step": 26040
     },
     {
       "epoch": 4.903067946546208,
-      "grad_norm": 5.606777191162109,
-      "learning_rate": 5.815923207227555e-07,
-      "loss": 0.3105,
+      "grad_norm": 1.9975135326385498,
+      "learning_rate": 9.693205345379259e-06,
+      "loss": 0.2964,
       "step": 26050
     },
     {
       "epoch": 4.904950122341427,
-      "grad_norm": 13.368224143981934,
-      "learning_rate": 5.7029926595144e-07,
-      "loss": 0.6274,
+      "grad_norm": 2.800941228866577,
+      "learning_rate": 9.504987765857332e-06,
+      "loss": 0.7167,
       "step": 26060
     },
     {
       "epoch": 4.906832298136646,
-      "grad_norm": 17.68413543701172,
-      "learning_rate": 5.590062111801243e-07,
-      "loss": 0.5218,
+      "grad_norm": 3.057074546813965,
+      "learning_rate": 9.316770186335403e-06,
+      "loss": 0.6159,
       "step": 26070
     },
     {
       "epoch": 4.908714473931865,
-      "grad_norm": 31.081344604492188,
-      "learning_rate": 5.477131564088085e-07,
-      "loss": 0.5692,
+      "grad_norm": 7.213024616241455,
+      "learning_rate": 9.128552606813476e-06,
+      "loss": 0.4206,
       "step": 26080
     },
     {
       "epoch": 4.9105966497270845,
-      "grad_norm": 11.856854438781738,
-      "learning_rate": 5.36420101637493e-07,
-      "loss": 0.6386,
+      "grad_norm": 1.9982532262802124,
+      "learning_rate": 8.94033502729155e-06,
+      "loss": 0.8574,
       "step": 26090
     },
     {
       "epoch": 4.912478825522304,
-      "grad_norm": 6.689940929412842,
-      "learning_rate": 5.251270468661773e-07,
-      "loss": 0.3447,
+      "grad_norm": 2.2955069541931152,
+      "learning_rate": 8.752117447769623e-06,
+      "loss": 0.4433,
       "step": 26100
     },
     {
       "epoch": 4.914361001317523,
-      "grad_norm": 36.05409622192383,
-      "learning_rate": 5.138339920948617e-07,
-      "loss": 0.2413,
+      "grad_norm": 5.474578857421875,
+      "learning_rate": 8.563899868247694e-06,
+      "loss": 0.3454,
       "step": 26110
     },
     {
       "epoch": 4.916243177112742,
-      "grad_norm": 21.475088119506836,
-      "learning_rate": 5.02540937323546e-07,
-      "loss": 0.6389,
+      "grad_norm": 3.1911139488220215,
+      "learning_rate": 8.375682288725767e-06,
+      "loss": 0.5058,
       "step": 26120
     },
     {
       "epoch": 4.918125352907961,
-      "grad_norm": 1.5936756134033203,
-      "learning_rate": 4.912478825522304e-07,
-      "loss": 0.4807,
+      "grad_norm": 2.5135414600372314,
+      "learning_rate": 8.18746470920384e-06,
+      "loss": 0.6098,
       "step": 26130
     },
     {
       "epoch": 4.920007528703181,
-      "grad_norm": 28.940305709838867,
-      "learning_rate": 4.799548277809147e-07,
-      "loss": 0.4051,
+      "grad_norm": 4.524852275848389,
+      "learning_rate": 7.999247129681913e-06,
+      "loss": 0.4982,
       "step": 26140
     },
     {
       "epoch": 4.9218897044984,
-      "grad_norm": 5.989145755767822,
-      "learning_rate": 4.686617730095991e-07,
-      "loss": 0.442,
+      "grad_norm": 1.1532628536224365,
+      "learning_rate": 7.811029550159986e-06,
+      "loss": 0.4741,
       "step": 26150
     },
     {
       "epoch": 4.92377188029362,
-      "grad_norm": 11.877305030822754,
-      "learning_rate": 4.573687182382835e-07,
-      "loss": 0.6841,
+      "grad_norm": 2.0357584953308105,
+      "learning_rate": 7.622811970638058e-06,
+      "loss": 0.7255,
       "step": 26160
     },
     {
       "epoch": 4.925654056088838,
-      "grad_norm": 10.223360061645508,
-      "learning_rate": 4.460756634669678e-07,
-      "loss": 0.5501,
+      "grad_norm": 4.0735344886779785,
+      "learning_rate": 7.43459439111613e-06,
+      "loss": 0.706,
       "step": 26170
     },
     {
       "epoch": 4.927536231884058,
-      "grad_norm": 6.764021396636963,
-      "learning_rate": 4.347826086956522e-07,
-      "loss": 0.711,
+      "grad_norm": 2.2796082496643066,
+      "learning_rate": 7.246376811594203e-06,
+      "loss": 0.7243,
       "step": 26180
     },
     {
       "epoch": 4.929418407679277,
-      "grad_norm": 9.022865295410156,
-      "learning_rate": 4.2348955392433657e-07,
-      "loss": 0.5528,
+      "grad_norm": 1.2028344869613647,
+      "learning_rate": 7.0581592320722764e-06,
+      "loss": 0.628,
       "step": 26190
     },
     {
       "epoch": 4.9313005834744965,
-      "grad_norm": 0.19420573115348816,
-      "learning_rate": 4.121964991530209e-07,
-      "loss": 0.8545,
+      "grad_norm": 0.6326656937599182,
+      "learning_rate": 6.869941652550348e-06,
+      "loss": 0.8906,
       "step": 26200
     },
     {
       "epoch": 4.933182759269716,
-      "grad_norm": 32.37028884887695,
-      "learning_rate": 4.009034443817053e-07,
-      "loss": 0.6968,
+      "grad_norm": 14.477922439575195,
+      "learning_rate": 6.681724073028421e-06,
+      "loss": 0.8574,
       "step": 26210
     },
     {
       "epoch": 4.935064935064935,
-      "grad_norm": 5.03609561920166,
-      "learning_rate": 3.8961038961038966e-07,
-      "loss": 0.5307,
+      "grad_norm": 2.4440884590148926,
+      "learning_rate": 6.493506493506494e-06,
+      "loss": 0.4625,
       "step": 26220
     },
     {
       "epoch": 4.936947110860155,
-      "grad_norm": 7.052425861358643,
-      "learning_rate": 3.78317334839074e-07,
-      "loss": 0.8121,
+      "grad_norm": 1.841376543045044,
+      "learning_rate": 6.305288913984566e-06,
+      "loss": 0.7268,
       "step": 26230
     },
     {
       "epoch": 4.938829286655373,
-      "grad_norm": 0.8062214255332947,
-      "learning_rate": 3.670242800677583e-07,
-      "loss": 0.4944,
+      "grad_norm": 0.1650257259607315,
+      "learning_rate": 6.117071334462639e-06,
+      "loss": 0.4164,
       "step": 26240
     },
     {
       "epoch": 4.940711462450593,
-      "grad_norm": 10.001143455505371,
-      "learning_rate": 3.5573122529644265e-07,
-      "loss": 0.3683,
+      "grad_norm": 2.4682869911193848,
+      "learning_rate": 5.928853754940711e-06,
+      "loss": 0.4473,
       "step": 26250
     },
     {
       "epoch": 4.942593638245812,
-      "grad_norm": 1.526486873626709,
-      "learning_rate": 3.4443817052512703e-07,
-      "loss": 0.3605,
+      "grad_norm": 0.20925642549991608,
+      "learning_rate": 5.740636175418784e-06,
+      "loss": 0.4475,
       "step": 26260
     },
     {
       "epoch": 4.944475814041032,
-      "grad_norm": 11.505463600158691,
-      "learning_rate": 3.331451157538114e-07,
-      "loss": 0.498,
+      "grad_norm": 0.88543301820755,
+      "learning_rate": 5.552418595896857e-06,
+      "loss": 0.5341,
       "step": 26270
     },
     {
       "epoch": 4.946357989836251,
-      "grad_norm": 18.725603103637695,
-      "learning_rate": 3.2185206098249575e-07,
-      "loss": 0.3436,
+      "grad_norm": 3.832179307937622,
+      "learning_rate": 5.3642010163749295e-06,
+      "loss": 0.4383,
       "step": 26280
     },
     {
       "epoch": 4.94824016563147,
-      "grad_norm": 8.331537246704102,
-      "learning_rate": 3.1055900621118013e-07,
-      "loss": 0.5627,
+      "grad_norm": 1.4329594373703003,
+      "learning_rate": 5.175983436853002e-06,
+      "loss": 0.5517,
       "step": 26290
     },
     {
       "epoch": 4.950122341426689,
-      "grad_norm": 10.584824562072754,
-      "learning_rate": 2.992659514398645e-07,
-      "loss": 0.4182,
+      "grad_norm": 4.162771701812744,
+      "learning_rate": 4.987765857331075e-06,
+      "loss": 0.4524,
       "step": 26300
     },
     {
       "epoch": 4.9520045172219085,
-      "grad_norm": 0.15740695595741272,
-      "learning_rate": 2.8797289666854884e-07,
-      "loss": 0.5521,
+      "grad_norm": 0.10146909952163696,
+      "learning_rate": 4.799548277809147e-06,
+      "loss": 0.5414,
       "step": 26310
     },
     {
       "epoch": 4.953886693017128,
-      "grad_norm": 11.263009071350098,
-      "learning_rate": 2.766798418972332e-07,
-      "loss": 0.4944,
+      "grad_norm": 2.9978926181793213,
+      "learning_rate": 4.61133069828722e-06,
+      "loss": 0.5602,
       "step": 26320
     },
     {
       "epoch": 4.955768868812347,
-      "grad_norm": 7.119065284729004,
-      "learning_rate": 2.653867871259176e-07,
-      "loss": 0.7362,
+      "grad_norm": 3.2251737117767334,
+      "learning_rate": 4.423113118765293e-06,
+      "loss": 0.7076,
       "step": 26330
     },
     {
       "epoch": 4.957651044607566,
-      "grad_norm": 1.8257107734680176,
-      "learning_rate": 2.5409373235460193e-07,
-      "loss": 0.1585,
+      "grad_norm": 1.7196509838104248,
+      "learning_rate": 4.234895539243365e-06,
+      "loss": 0.2048,
       "step": 26340
     },
     {
       "epoch": 4.959533220402785,
-      "grad_norm": 33.10144805908203,
-      "learning_rate": 2.4280067758328626e-07,
-      "loss": 0.4476,
+      "grad_norm": 1.6203038692474365,
+      "learning_rate": 4.046677959721438e-06,
+      "loss": 0.4421,
       "step": 26350
     },
     {
       "epoch": 4.961415396198005,
-      "grad_norm": 2.7676703929901123,
-      "learning_rate": 2.3150762281197064e-07,
-      "loss": 0.366,
+      "grad_norm": 2.4939804077148438,
+      "learning_rate": 3.85846038019951e-06,
+      "loss": 0.5145,
       "step": 26360
     },
     {
       "epoch": 4.963297571993224,
-      "grad_norm": 0.12121053785085678,
-      "learning_rate": 2.2021456804065503e-07,
-      "loss": 0.4664,
+      "grad_norm": 0.217250794172287,
+      "learning_rate": 3.6702428006775834e-06,
+      "loss": 0.5779,
       "step": 26370
     },
     {
       "epoch": 4.965179747788444,
-      "grad_norm": 0.2563054859638214,
-      "learning_rate": 2.0892151326933935e-07,
-      "loss": 0.3163,
+      "grad_norm": 1.894600510597229,
+      "learning_rate": 3.482025221155656e-06,
+      "loss": 0.4681,
       "step": 26380
     },
     {
       "epoch": 4.967061923583663,
-      "grad_norm": 15.612191200256348,
-      "learning_rate": 1.976284584980237e-07,
-      "loss": 0.4692,
+      "grad_norm": 4.688291549682617,
+      "learning_rate": 3.2938076416337287e-06,
+      "loss": 0.6112,
       "step": 26390
     },
     {
       "epoch": 4.9689440993788825,
-      "grad_norm": 12.518272399902344,
-      "learning_rate": 1.8633540372670807e-07,
-      "loss": 0.5011,
+      "grad_norm": 3.2292239665985107,
+      "learning_rate": 3.1055900621118013e-06,
+      "loss": 0.5831,
       "step": 26400
     },
     {
       "epoch": 4.970826275174101,
-      "grad_norm": 8.736549377441406,
-      "learning_rate": 1.7504234895539245e-07,
-      "loss": 0.7325,
+      "grad_norm": 2.3493993282318115,
+      "learning_rate": 2.917372482589874e-06,
+      "loss": 0.5896,
       "step": 26410
     },
     {
       "epoch": 4.9727084509693205,
-      "grad_norm": 45.901878356933594,
-      "learning_rate": 1.637492941840768e-07,
-      "loss": 0.5119,
+      "grad_norm": 6.07112455368042,
+      "learning_rate": 2.7291549030679465e-06,
+      "loss": 0.6324,
       "step": 26420
     },
     {
       "epoch": 4.97459062676454,
-      "grad_norm": 15.459980010986328,
-      "learning_rate": 1.5245623941276113e-07,
-      "loss": 0.4057,
+      "grad_norm": 1.9999442100524902,
+      "learning_rate": 2.540937323546019e-06,
+      "loss": 0.3659,
       "step": 26430
     },
     {
       "epoch": 4.976472802559759,
-      "grad_norm": 12.372354507446289,
-      "learning_rate": 1.4116318464144551e-07,
-      "loss": 0.9019,
+      "grad_norm": 1.9524005651474,
+      "learning_rate": 2.352719744024092e-06,
+      "loss": 0.7785,
       "step": 26440
     },
     {
       "epoch": 4.978354978354979,
-      "grad_norm": 8.161259651184082,
-      "learning_rate": 1.2987012987012987e-07,
-      "loss": 0.3162,
+      "grad_norm": 1.6963441371917725,
+      "learning_rate": 2.1645021645021648e-06,
+      "loss": 0.3846,
       "step": 26450
     },
     {
       "epoch": 4.980237154150197,
-      "grad_norm": 2.586444854736328,
-      "learning_rate": 1.1857707509881423e-07,
-      "loss": 0.6493,
+      "grad_norm": 2.9382524490356445,
+      "learning_rate": 1.976284584980237e-06,
+      "loss": 0.6171,
       "step": 26460
     },
     {
       "epoch": 4.982119329945417,
-      "grad_norm": 0.4139041602611542,
-      "learning_rate": 1.0728402032749858e-07,
-      "loss": 0.3598,
+      "grad_norm": 0.4651086628437042,
+      "learning_rate": 1.7880670054583098e-06,
+      "loss": 0.3558,
       "step": 26470
     },
     {
       "epoch": 4.984001505740636,
-      "grad_norm": 0.5446306467056274,
-      "learning_rate": 9.599096555618295e-08,
-      "loss": 0.4985,
+      "grad_norm": 4.864592552185059,
+      "learning_rate": 1.5998494259363826e-06,
+      "loss": 0.5754,
       "step": 26480
     },
     {
       "epoch": 4.985883681535856,
-      "grad_norm": 14.608466148376465,
-      "learning_rate": 8.46979107848673e-08,
-      "loss": 0.3913,
+      "grad_norm": 2.7198994159698486,
+      "learning_rate": 1.411631846414455e-06,
+      "loss": 0.4226,
       "step": 26490
     },
     {
       "epoch": 4.987765857331075,
-      "grad_norm": 27.869840621948242,
-      "learning_rate": 7.340485601355166e-08,
-      "loss": 0.5136,
+      "grad_norm": 3.3242056369781494,
+      "learning_rate": 1.2234142668925278e-06,
+      "loss": 0.6177,
       "step": 26500
     },
     {
       "epoch": 4.989648033126294,
-      "grad_norm": 7.235048294067383,
-      "learning_rate": 6.211180124223603e-08,
-      "loss": 0.5271,
+      "grad_norm": 3.1476473808288574,
+      "learning_rate": 1.0351966873706004e-06,
+      "loss": 0.5798,
       "step": 26510
     },
     {
       "epoch": 4.991530208921513,
-      "grad_norm": 16.698986053466797,
-      "learning_rate": 5.0818746470920386e-08,
-      "loss": 0.6373,
+      "grad_norm": 2.154055118560791,
+      "learning_rate": 8.469791078486731e-07,
+      "loss": 0.7016,
       "step": 26520
     },
     {
       "epoch": 4.993412384716732,
-      "grad_norm": 7.239467620849609,
-      "learning_rate": 3.952569169960475e-08,
-      "loss": 0.4937,
+      "grad_norm": 9.276118278503418,
+      "learning_rate": 6.587615283267458e-07,
+      "loss": 0.4034,
       "step": 26530
     },
     {
       "epoch": 4.995294560511952,
-      "grad_norm": 33.553871154785156,
-      "learning_rate": 2.8232636928289104e-08,
-      "loss": 0.6561,
+      "grad_norm": 5.92098331451416,
+      "learning_rate": 4.7054394880481837e-07,
+      "loss": 0.7285,
       "step": 26540
     },
     {
       "epoch": 4.997176736307171,
-      "grad_norm": 11.409107208251953,
-      "learning_rate": 1.693958215697346e-08,
-      "loss": 0.5426,
+      "grad_norm": 1.5935169458389282,
+      "learning_rate": 2.8232636928289103e-07,
+      "loss": 0.6105,
       "step": 26550
     },
     {
       "epoch": 4.999058912102391,
-      "grad_norm": 26.004404067993164,
-      "learning_rate": 5.6465273856578205e-09,
-      "loss": 0.4763,
+      "grad_norm": 2.964055061340332,
+      "learning_rate": 9.410878976096367e-08,
+      "loss": 0.4711,
       "step": 26560
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.9252,
-      "eval_loss": 0.2850923240184784,
-      "eval_runtime": 277.3592,
-      "eval_samples_per_second": 27.041,
-      "eval_steps_per_second": 3.382,
+      "eval_accuracy": 0.9236,
+      "eval_loss": 0.2550007104873657,
+      "eval_runtime": 89.7951,
+      "eval_samples_per_second": 83.523,
+      "eval_steps_per_second": 10.446,
       "step": 26565
     },
     {
       "epoch": 5.0,
       "step": 26565,
-      "total_flos": 1.64815115092992e+19,
-      "train_loss": 1.0248974845253678,
-      "train_runtime": 12727.7584,
-      "train_samples_per_second": 16.696,
-      "train_steps_per_second": 2.087
+      "total_flos": 1.66660717676544e+19,
+      "train_loss": 0.7802943237124471,
+      "train_runtime": 11340.9032,
+      "train_samples_per_second": 18.737,
+      "train_steps_per_second": 2.342
     }
   ],
   "logging_steps": 10,
@@ -18672,7 +18672,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.64815115092992e+19,
+  "total_flos": 1.66660717676544e+19,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null