diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.2954551577568054,
+  "best_metric": 0.2850923240184784,
   "best_model_checkpoint": "./cifar100_outputs/checkpoint-26565",
   "epoch": 5.0,
   "eval_steps": 500,
@@ -10,18649 +10,18649 @@
   "log_history": [
     {
       "epoch": 0.0018821757952192735,
-      "grad_norm": 2.8234074115753174,
-      "learning_rate": 1.9992471296819125e-05,
-      "loss": 4.6312,
+      "grad_norm": 3.259775161743164,
+      "learning_rate": 2.9988706945228687e-05,
+      "loss": 4.6305,
       "step": 10
     },
     {
       "epoch": 0.003764351590438547,
-      "grad_norm": 2.5866012573242188,
-      "learning_rate": 1.9984942593638247e-05,
-      "loss": 4.5949,
+      "grad_norm": 2.9150309562683105,
+      "learning_rate": 2.997741389045737e-05,
+      "loss": 4.5916,
       "step": 20
     },
     {
       "epoch": 0.00564652738565782,
-      "grad_norm": 2.6120123863220215,
-      "learning_rate": 1.997741389045737e-05,
-      "loss": 4.6003,
+      "grad_norm": 2.639889717102051,
+      "learning_rate": 2.9966120835686056e-05,
+      "loss": 4.5977,
       "step": 30
     },
     {
       "epoch": 0.007528703180877094,
-      "grad_norm": 2.558011293411255,
-      "learning_rate": 1.9969885187276493e-05,
-      "loss": 4.6029,
+      "grad_norm": 2.729921340942383,
+      "learning_rate": 2.9954827780914738e-05,
+      "loss": 4.5964,
       "step": 40
     },
     {
       "epoch": 0.009410878976096368,
-      "grad_norm": 2.4474682807922363,
-      "learning_rate": 1.9962356484095616e-05,
-      "loss": 4.6158,
+      "grad_norm": 2.3783774375915527,
+      "learning_rate": 2.994353472614342e-05,
+      "loss": 4.6022,
       "step": 50
     },
     {
       "epoch": 0.01129305477131564,
-      "grad_norm": 2.758237838745117,
-      "learning_rate": 1.995482778091474e-05,
-      "loss": 4.6014,
+      "grad_norm": 2.656496524810791,
+      "learning_rate": 2.9932241671372107e-05,
+      "loss": 4.5908,
       "step": 60
     },
     {
       "epoch": 0.013175230566534914,
-      "grad_norm": 4.711178779602051,
-      "learning_rate": 1.9947299077733862e-05,
-      "loss": 4.5894,
+      "grad_norm": 2.692168712615967,
+      "learning_rate": 2.992094861660079e-05,
+      "loss": 4.5829,
       "step": 70
     },
     {
       "epoch": 0.015057406361754188,
-      "grad_norm": 2.3408522605895996,
-      "learning_rate": 1.9939770374552985e-05,
-      "loss": 4.5783,
+      "grad_norm": 2.8096225261688232,
+      "learning_rate": 2.9909655561829476e-05,
+      "loss": 4.5645,
       "step": 80
     },
     {
       "epoch": 0.01693958215697346,
-      "grad_norm": 2.6499340534210205,
-      "learning_rate": 1.9932241671372108e-05,
-      "loss": 4.5679,
+      "grad_norm": 2.7067952156066895,
+      "learning_rate": 2.989836250705816e-05,
+      "loss": 4.5448,
       "step": 90
     },
     {
       "epoch": 0.018821757952192736,
-      "grad_norm": 2.65826678276062,
-      "learning_rate": 1.992471296819123e-05,
-      "loss": 4.5715,
+      "grad_norm": 2.5358433723449707,
+      "learning_rate": 2.9887069452286845e-05,
+      "loss": 4.5471,
       "step": 100
     },
     {
       "epoch": 0.020703933747412008,
-      "grad_norm": 2.484968662261963,
-      "learning_rate": 1.9917184265010354e-05,
-      "loss": 4.5451,
+      "grad_norm": 2.4757063388824463,
+      "learning_rate": 2.987577639751553e-05,
+      "loss": 4.5546,
       "step": 110
     },
     {
       "epoch": 0.02258610954263128,
-      "grad_norm": 2.5142550468444824,
-      "learning_rate": 1.9909655561829477e-05,
-      "loss": 4.5675,
+      "grad_norm": 2.8501641750335693,
+      "learning_rate": 2.9864483342744213e-05,
+      "loss": 4.5471,
       "step": 120
     },
     {
       "epoch": 0.024468285337850556,
-      "grad_norm": 2.3911235332489014,
-      "learning_rate": 1.99021268586486e-05,
-      "loss": 4.5623,
+      "grad_norm": 2.506326198577881,
+      "learning_rate": 2.98531902879729e-05,
+      "loss": 4.5289,
       "step": 130
     },
     {
       "epoch": 0.026350461133069828,
-      "grad_norm": 2.7664413452148438,
-      "learning_rate": 1.9894598155467723e-05,
-      "loss": 4.5548,
+      "grad_norm": 2.7096266746520996,
+      "learning_rate": 2.9841897233201582e-05,
+      "loss": 4.5213,
       "step": 140
     },
     {
       "epoch": 0.028232636928289104,
-      "grad_norm": 2.7500317096710205,
-      "learning_rate": 1.9887069452286845e-05,
-      "loss": 4.5317,
+      "grad_norm": 2.9105842113494873,
+      "learning_rate": 2.9830604178430265e-05,
+      "loss": 4.5017,
       "step": 150
     },
     {
       "epoch": 0.030114812723508376,
-      "grad_norm": 2.9022419452667236,
-      "learning_rate": 1.987954074910597e-05,
-      "loss": 4.5442,
+      "grad_norm": 2.713897466659546,
+      "learning_rate": 2.981931112365895e-05,
+      "loss": 4.5014,
       "step": 160
     },
     {
       "epoch": 0.03199698851872765,
-      "grad_norm": 3.4607958793640137,
-      "learning_rate": 1.987201204592509e-05,
-      "loss": 4.5409,
+      "grad_norm": 3.520843744277954,
+      "learning_rate": 2.9808018068887634e-05,
+      "loss": 4.482,
       "step": 170
     },
     {
       "epoch": 0.03387916431394692,
-      "grad_norm": 2.7307426929473877,
-      "learning_rate": 1.9864483342744214e-05,
-      "loss": 4.5272,
+      "grad_norm": 3.0594747066497803,
+      "learning_rate": 2.979672501411632e-05,
+      "loss": 4.4706,
       "step": 180
     },
     {
       "epoch": 0.0357613401091662,
-      "grad_norm": 2.797895669937134,
-      "learning_rate": 1.9856954639563337e-05,
-      "loss": 4.5225,
+      "grad_norm": 2.884718418121338,
+      "learning_rate": 2.9785431959345006e-05,
+      "loss": 4.4648,
       "step": 190
     },
     {
       "epoch": 0.03764351590438547,
-      "grad_norm": 2.6407244205474854,
-      "learning_rate": 1.984942593638246e-05,
-      "loss": 4.505,
+      "grad_norm": 2.8616080284118652,
+      "learning_rate": 2.977413890457369e-05,
+      "loss": 4.4629,
       "step": 200
     },
     {
       "epoch": 0.039525691699604744,
-      "grad_norm": 2.401859760284424,
-      "learning_rate": 1.9841897233201583e-05,
-      "loss": 4.5198,
+      "grad_norm": 2.6512672901153564,
+      "learning_rate": 2.9762845849802374e-05,
+      "loss": 4.4555,
       "step": 210
     },
     {
       "epoch": 0.041407867494824016,
-      "grad_norm": 2.945293664932251,
-      "learning_rate": 1.9834368530020706e-05,
-      "loss": 4.5059,
+      "grad_norm": 3.5559914112091064,
+      "learning_rate": 2.9751552795031054e-05,
+      "loss": 4.4335,
       "step": 220
     },
     {
       "epoch": 0.04329004329004329,
-      "grad_norm": 2.723217248916626,
-      "learning_rate": 1.982683982683983e-05,
-      "loss": 4.4505,
+      "grad_norm": 2.8921027183532715,
+      "learning_rate": 2.974025974025974e-05,
+      "loss": 4.3462,
       "step": 230
     },
     {
       "epoch": 0.04517221908526256,
-      "grad_norm": 3.357862710952759,
-      "learning_rate": 1.9819311123658952e-05,
-      "loss": 4.4769,
+      "grad_norm": 2.7680187225341797,
+      "learning_rate": 2.9728966685488426e-05,
+      "loss": 4.4017,
       "step": 240
     },
     {
       "epoch": 0.04705439488048184,
-      "grad_norm": 3.170531749725342,
-      "learning_rate": 1.9811782420478075e-05,
-      "loss": 4.47,
+      "grad_norm": 2.8839056491851807,
+      "learning_rate": 2.971767363071711e-05,
+      "loss": 4.4098,
       "step": 250
     },
     {
       "epoch": 0.04893657067570111,
-      "grad_norm": 2.830183744430542,
-      "learning_rate": 1.9804253717297198e-05,
-      "loss": 4.4742,
+      "grad_norm": 2.8372817039489746,
+      "learning_rate": 2.9706380575945795e-05,
+      "loss": 4.3837,
       "step": 260
     },
     {
       "epoch": 0.050818746470920384,
-      "grad_norm": 2.7911784648895264,
-      "learning_rate": 1.979672501411632e-05,
-      "loss": 4.4557,
+      "grad_norm": 2.884737968444824,
+      "learning_rate": 2.9695087521174477e-05,
+      "loss": 4.3769,
       "step": 270
     },
     {
       "epoch": 0.052700922266139656,
-      "grad_norm": 2.965761423110962,
-      "learning_rate": 1.9789196310935443e-05,
-      "loss": 4.4504,
+      "grad_norm": 2.899841070175171,
+      "learning_rate": 2.9683794466403163e-05,
+      "loss": 4.3447,
       "step": 280
     },
     {
       "epoch": 0.05458309806135893,
-      "grad_norm": 3.0982887744903564,
-      "learning_rate": 1.9781667607754566e-05,
-      "loss": 4.4378,
+      "grad_norm": 3.1840929985046387,
+      "learning_rate": 2.967250141163185e-05,
+      "loss": 4.3174,
       "step": 290
     },
     {
       "epoch": 0.05646527385657821,
-      "grad_norm": 3.3339033126831055,
-      "learning_rate": 1.977413890457369e-05,
-      "loss": 4.4213,
+      "grad_norm": 2.9409921169281006,
+      "learning_rate": 2.9661208356860532e-05,
+      "loss": 4.3072,
       "step": 300
     },
     {
       "epoch": 0.05834744965179748,
-      "grad_norm": 2.5607752799987793,
-      "learning_rate": 1.9766610201392812e-05,
-      "loss": 4.4296,
+      "grad_norm": 3.342649459838867,
+      "learning_rate": 2.9649915302089215e-05,
+      "loss": 4.3291,
       "step": 310
     },
     {
       "epoch": 0.06022962544701675,
-      "grad_norm": 2.7375972270965576,
-      "learning_rate": 1.9759081498211935e-05,
-      "loss": 4.4418,
+      "grad_norm": 2.746373176574707,
+      "learning_rate": 2.9638622247317898e-05,
+      "loss": 4.3288,
       "step": 320
     },
     {
       "epoch": 0.062111801242236024,
-      "grad_norm": 2.757209539413452,
-      "learning_rate": 1.9751552795031058e-05,
-      "loss": 4.3921,
+      "grad_norm": 3.7221550941467285,
+      "learning_rate": 2.9627329192546584e-05,
+      "loss": 4.2612,
       "step": 330
     },
     {
       "epoch": 0.0639939770374553,
-      "grad_norm": 3.423086404800415,
-      "learning_rate": 1.974402409185018e-05,
-      "loss": 4.408,
+      "grad_norm": 2.8289361000061035,
+      "learning_rate": 2.961603613777527e-05,
+      "loss": 4.2885,
       "step": 340
     },
     {
       "epoch": 0.06587615283267458,
-      "grad_norm": 3.178119421005249,
-      "learning_rate": 1.9736495388669304e-05,
-      "loss": 4.3862,
+      "grad_norm": 3.2310261726379395,
+      "learning_rate": 2.9604743083003952e-05,
+      "loss": 4.2502,
       "step": 350
     },
     {
       "epoch": 0.06775832862789384,
-      "grad_norm": 2.718017816543579,
-      "learning_rate": 1.9728966685488427e-05,
-      "loss": 4.3867,
+      "grad_norm": 2.873511791229248,
+      "learning_rate": 2.959345002823264e-05,
+      "loss": 4.2713,
       "step": 360
     },
     {
       "epoch": 0.06964050442311312,
-      "grad_norm": 2.780374765396118,
-      "learning_rate": 1.972143798230755e-05,
-      "loss": 4.3705,
+      "grad_norm": 2.8159632682800293,
+      "learning_rate": 2.958215697346132e-05,
+      "loss": 4.237,
       "step": 370
     },
     {
       "epoch": 0.0715226802183324,
-      "grad_norm": 3.7188100814819336,
-      "learning_rate": 1.9713909279126673e-05,
-      "loss": 4.3587,
+      "grad_norm": 4.2501220703125,
+      "learning_rate": 2.9570863918690007e-05,
+      "loss": 4.2377,
       "step": 380
     },
     {
       "epoch": 0.07340485601355166,
-      "grad_norm": 2.7755348682403564,
-      "learning_rate": 1.9706380575945796e-05,
-      "loss": 4.38,
+      "grad_norm": 3.2059030532836914,
+      "learning_rate": 2.9559570863918693e-05,
+      "loss": 4.245,
       "step": 390
     },
     {
       "epoch": 0.07528703180877094,
-      "grad_norm": 3.0622341632843018,
-      "learning_rate": 1.969885187276492e-05,
-      "loss": 4.3634,
+      "grad_norm": 3.3538947105407715,
+      "learning_rate": 2.9548277809147373e-05,
+      "loss": 4.2061,
       "step": 400
     },
     {
       "epoch": 0.07716920760399021,
-      "grad_norm": 3.146724224090576,
-      "learning_rate": 1.969132316958404e-05,
-      "loss": 4.3677,
+      "grad_norm": 3.506152391433716,
+      "learning_rate": 2.953698475437606e-05,
+      "loss": 4.1917,
       "step": 410
     },
     {
       "epoch": 0.07905138339920949,
-      "grad_norm": 4.22658634185791,
-      "learning_rate": 1.9683794466403164e-05,
-      "loss": 4.3087,
+      "grad_norm": 3.823164701461792,
+      "learning_rate": 2.9525691699604745e-05,
+      "loss": 4.1214,
       "step": 420
     },
     {
       "epoch": 0.08093355919442875,
-      "grad_norm": 3.4560253620147705,
-      "learning_rate": 1.9676265763222287e-05,
-      "loss": 4.3527,
+      "grad_norm": 3.458685874938965,
+      "learning_rate": 2.9514398644833427e-05,
+      "loss": 4.1866,
       "step": 430
     },
     {
       "epoch": 0.08281573498964803,
-      "grad_norm": 2.921839714050293,
-      "learning_rate": 1.966873706004141e-05,
-      "loss": 4.3253,
+      "grad_norm": 3.0740559101104736,
+      "learning_rate": 2.9503105590062114e-05,
+      "loss": 4.1349,
       "step": 440
     },
     {
       "epoch": 0.08469791078486731,
-      "grad_norm": 2.827158212661743,
-      "learning_rate": 1.9661208356860533e-05,
-      "loss": 4.3189,
+      "grad_norm": 4.038434982299805,
+      "learning_rate": 2.9491812535290796e-05,
+      "loss": 4.1513,
       "step": 450
     },
     {
       "epoch": 0.08658008658008658,
-      "grad_norm": 2.8546581268310547,
-      "learning_rate": 1.9653679653679656e-05,
-      "loss": 4.2886,
+      "grad_norm": 3.2392990589141846,
+      "learning_rate": 2.9480519480519482e-05,
+      "loss": 4.0751,
       "step": 460
     },
     {
       "epoch": 0.08846226237530586,
-      "grad_norm": 3.17734694480896,
-      "learning_rate": 1.964615095049878e-05,
-      "loss": 4.3264,
+      "grad_norm": 2.7104597091674805,
+      "learning_rate": 2.946922642574817e-05,
+      "loss": 4.1389,
       "step": 470
     },
     {
       "epoch": 0.09034443817052512,
-      "grad_norm": 3.0701353549957275,
-      "learning_rate": 1.96386222473179e-05,
-      "loss": 4.2996,
+      "grad_norm": 3.526334762573242,
+      "learning_rate": 2.9457933370976848e-05,
+      "loss": 4.0984,
       "step": 480
     },
     {
       "epoch": 0.0922266139657444,
-      "grad_norm": 2.914555549621582,
-      "learning_rate": 1.9631093544137025e-05,
-      "loss": 4.2455,
+      "grad_norm": 4.827049732208252,
+      "learning_rate": 2.9446640316205534e-05,
+      "loss": 4.065,
       "step": 490
     },
     {
       "epoch": 0.09410878976096368,
-      "grad_norm": 2.8697192668914795,
-      "learning_rate": 1.9623564840956148e-05,
-      "loss": 4.2583,
+      "grad_norm": 3.1481475830078125,
+      "learning_rate": 2.9435347261434216e-05,
+      "loss": 4.0429,
       "step": 500
     },
     {
       "epoch": 0.09599096555618294,
-      "grad_norm": 2.944056272506714,
-      "learning_rate": 1.961603613777527e-05,
-      "loss": 4.2599,
+      "grad_norm": 3.47776198387146,
+      "learning_rate": 2.9424054206662903e-05,
+      "loss": 4.0432,
       "step": 510
     },
     {
       "epoch": 0.09787314135140222,
-      "grad_norm": 3.0705983638763428,
-      "learning_rate": 1.9608507434594394e-05,
-      "loss": 4.2685,
+      "grad_norm": 2.9574878215789795,
+      "learning_rate": 2.941276115189159e-05,
+      "loss": 4.0486,
       "step": 520
     },
     {
       "epoch": 0.09975531714662149,
-      "grad_norm": 3.1787703037261963,
-      "learning_rate": 1.9600978731413516e-05,
-      "loss": 4.2272,
+      "grad_norm": 3.398369073867798,
+      "learning_rate": 2.940146809712027e-05,
+      "loss": 3.9975,
       "step": 530
     },
     {
       "epoch": 0.10163749294184077,
-      "grad_norm": 3.236412763595581,
-      "learning_rate": 1.959345002823264e-05,
-      "loss": 4.251,
+      "grad_norm": 3.2423532009124756,
+      "learning_rate": 2.9390175042348957e-05,
+      "loss": 4.0189,
       "step": 540
     },
     {
       "epoch": 0.10351966873706005,
-      "grad_norm": 3.3419723510742188,
-      "learning_rate": 1.9585921325051762e-05,
-      "loss": 4.2056,
+      "grad_norm": 4.2174153327941895,
+      "learning_rate": 2.937888198757764e-05,
+      "loss": 3.9679,
       "step": 550
     },
     {
       "epoch": 0.10540184453227931,
-      "grad_norm": 3.7011070251464844,
-      "learning_rate": 1.9578392621870885e-05,
-      "loss": 4.2403,
+      "grad_norm": 4.3004045486450195,
+      "learning_rate": 2.9367588932806326e-05,
+      "loss": 4.0162,
       "step": 560
     },
     {
       "epoch": 0.10728402032749859,
-      "grad_norm": 3.077976703643799,
-      "learning_rate": 1.9570863918690005e-05,
-      "loss": 4.2163,
+      "grad_norm": 3.0636956691741943,
+      "learning_rate": 2.935629587803501e-05,
+      "loss": 3.9957,
       "step": 570
     },
     {
       "epoch": 0.10916619612271786,
-      "grad_norm": 3.373072624206543,
-      "learning_rate": 1.956333521550913e-05,
-      "loss": 4.2113,
+      "grad_norm": 3.1375842094421387,
+      "learning_rate": 2.934500282326369e-05,
+      "loss": 4.0078,
       "step": 580
     },
     {
       "epoch": 0.11104837191793714,
-      "grad_norm": 3.0897161960601807,
-      "learning_rate": 1.9555806512328254e-05,
-      "loss": 4.1671,
+      "grad_norm": 3.439136505126953,
+      "learning_rate": 2.9333709768492378e-05,
+      "loss": 3.9473,
       "step": 590
     },
     {
       "epoch": 0.11293054771315642,
-      "grad_norm": 3.42901873588562,
-      "learning_rate": 1.9548277809147377e-05,
-      "loss": 4.1616,
+      "grad_norm": 3.252802610397339,
+      "learning_rate": 2.932241671372106e-05,
+      "loss": 3.9155,
       "step": 600
     },
     {
       "epoch": 0.11481272350837568,
-      "grad_norm": 3.4786670207977295,
-      "learning_rate": 1.95407491059665e-05,
-      "loss": 4.1452,
+      "grad_norm": 3.3427743911743164,
+      "learning_rate": 2.9311123658949746e-05,
+      "loss": 3.9003,
       "step": 610
     },
     {
       "epoch": 0.11669489930359496,
-      "grad_norm": 2.832836151123047,
-      "learning_rate": 1.9533220402785623e-05,
-      "loss": 4.1742,
+      "grad_norm": 3.1991214752197266,
+      "learning_rate": 2.9299830604178432e-05,
+      "loss": 3.8892,
       "step": 620
     },
     {
       "epoch": 0.11857707509881422,
-      "grad_norm": 3.0124964714050293,
-      "learning_rate": 1.9525691699604746e-05,
-      "loss": 4.1412,
+      "grad_norm": 3.2154698371887207,
+      "learning_rate": 2.9288537549407115e-05,
+      "loss": 3.8717,
       "step": 630
     },
     {
       "epoch": 0.1204592508940335,
-      "grad_norm": 2.7018187046051025,
-      "learning_rate": 1.951816299642387e-05,
-      "loss": 4.1027,
+      "grad_norm": 2.7783098220825195,
+      "learning_rate": 2.92772444946358e-05,
+      "loss": 3.8236,
       "step": 640
     },
     {
       "epoch": 0.12234142668925278,
-      "grad_norm": 3.9656338691711426,
-      "learning_rate": 1.951063429324299e-05,
-      "loss": 4.1837,
+      "grad_norm": 4.598132610321045,
+      "learning_rate": 2.9265951439864487e-05,
+      "loss": 3.9406,
       "step": 650
     },
     {
       "epoch": 0.12422360248447205,
-      "grad_norm": 3.147698402404785,
-      "learning_rate": 1.950310559006211e-05,
-      "loss": 4.0989,
+      "grad_norm": 3.633305549621582,
+      "learning_rate": 2.9254658385093167e-05,
+      "loss": 3.7923,
       "step": 660
     },
     {
       "epoch": 0.12610577827969133,
-      "grad_norm": 3.232612371444702,
-      "learning_rate": 1.9495576886881234e-05,
-      "loss": 4.0819,
+      "grad_norm": 3.794126272201538,
+      "learning_rate": 2.9243365330321853e-05,
+      "loss": 3.7865,
       "step": 670
     },
     {
       "epoch": 0.1279879540749106,
-      "grad_norm": 3.327009916305542,
-      "learning_rate": 1.948804818370036e-05,
-      "loss": 4.0768,
+      "grad_norm": 3.1769659519195557,
+      "learning_rate": 2.9232072275550535e-05,
+      "loss": 3.7679,
       "step": 680
     },
     {
       "epoch": 0.12987012987012986,
-      "grad_norm": 3.508767604827881,
-      "learning_rate": 1.9480519480519483e-05,
-      "loss": 4.0784,
+      "grad_norm": 3.30849289894104,
+      "learning_rate": 2.922077922077922e-05,
+      "loss": 3.8356,
       "step": 690
     },
     {
       "epoch": 0.13175230566534915,
-      "grad_norm": 3.3220221996307373,
-      "learning_rate": 1.9472990777338606e-05,
-      "loss": 4.0878,
+      "grad_norm": 4.426156997680664,
+      "learning_rate": 2.9209486166007908e-05,
+      "loss": 3.7757,
       "step": 700
     },
     {
       "epoch": 0.13363448146056842,
-      "grad_norm": 2.8315317630767822,
-      "learning_rate": 1.946546207415773e-05,
-      "loss": 4.0156,
+      "grad_norm": 3.004460096359253,
+      "learning_rate": 2.919819311123659e-05,
+      "loss": 3.6859,
       "step": 710
     },
     {
       "epoch": 0.13551665725578768,
-      "grad_norm": 3.6923279762268066,
-      "learning_rate": 1.9457933370976852e-05,
-      "loss": 4.0251,
+      "grad_norm": 3.6174192428588867,
+      "learning_rate": 2.9186900056465276e-05,
+      "loss": 3.745,
       "step": 720
     },
     {
       "epoch": 0.13739883305100697,
-      "grad_norm": 2.8323168754577637,
-      "learning_rate": 1.9450404667795975e-05,
-      "loss": 4.0082,
+      "grad_norm": 3.1848292350769043,
+      "learning_rate": 2.917560700169396e-05,
+      "loss": 3.6902,
       "step": 730
     },
     {
       "epoch": 0.13928100884622624,
-      "grad_norm": 3.0776381492614746,
-      "learning_rate": 1.9442875964615098e-05,
-      "loss": 4.0494,
+      "grad_norm": 3.1406095027923584,
+      "learning_rate": 2.9164313946922645e-05,
+      "loss": 3.7553,
       "step": 740
     },
     {
       "epoch": 0.1411631846414455,
-      "grad_norm": 3.2516698837280273,
-      "learning_rate": 1.9435347261434217e-05,
-      "loss": 4.0256,
+      "grad_norm": 3.086644411087036,
+      "learning_rate": 2.9153020892151328e-05,
+      "loss": 3.736,
       "step": 750
     },
     {
       "epoch": 0.1430453604366648,
-      "grad_norm": 5.407994747161865,
-      "learning_rate": 1.942781855825334e-05,
-      "loss": 3.9929,
+      "grad_norm": 5.149982452392578,
+      "learning_rate": 2.914172783738001e-05,
+      "loss": 3.6852,
       "step": 760
     },
     {
       "epoch": 0.14492753623188406,
-      "grad_norm": 3.1175644397735596,
-      "learning_rate": 1.9420289855072467e-05,
-      "loss": 3.999,
+      "grad_norm": 4.200068950653076,
+      "learning_rate": 2.9130434782608696e-05,
+      "loss": 3.6916,
       "step": 770
     },
     {
       "epoch": 0.14680971202710333,
-      "grad_norm": 3.38260817527771,
-      "learning_rate": 1.941276115189159e-05,
-      "loss": 3.9711,
+      "grad_norm": 3.5384104251861572,
+      "learning_rate": 2.911914172783738e-05,
+      "loss": 3.6724,
       "step": 780
     },
     {
       "epoch": 0.1486918878223226,
-      "grad_norm": 3.698014259338379,
-      "learning_rate": 1.9405232448710712e-05,
-      "loss": 4.0025,
+      "grad_norm": 4.555196285247803,
+      "learning_rate": 2.9107848673066065e-05,
+      "loss": 3.6608,
       "step": 790
     },
     {
       "epoch": 0.1505740636175419,
-      "grad_norm": 3.224360466003418,
-      "learning_rate": 1.9397703745529835e-05,
-      "loss": 3.9694,
+      "grad_norm": 4.264336585998535,
+      "learning_rate": 2.909655561829475e-05,
+      "loss": 3.6538,
       "step": 800
     },
     {
       "epoch": 0.15245623941276115,
-      "grad_norm": 3.315228223800659,
-      "learning_rate": 1.9390175042348958e-05,
-      "loss": 3.9376,
+      "grad_norm": 4.192948818206787,
+      "learning_rate": 2.9085262563523434e-05,
+      "loss": 3.5788,
       "step": 810
     },
     {
       "epoch": 0.15433841520798042,
-      "grad_norm": 4.7839531898498535,
-      "learning_rate": 1.938264633916808e-05,
-      "loss": 3.9115,
+      "grad_norm": 4.293764114379883,
+      "learning_rate": 2.907396950875212e-05,
+      "loss": 3.6034,
       "step": 820
     },
     {
       "epoch": 0.1562205910031997,
-      "grad_norm": 3.185082197189331,
-      "learning_rate": 1.93751176359872e-05,
-      "loss": 3.93,
+      "grad_norm": 3.505387544631958,
+      "learning_rate": 2.90626764539808e-05,
+      "loss": 3.6135,
       "step": 830
     },
     {
       "epoch": 0.15810276679841898,
-      "grad_norm": 3.3744490146636963,
-      "learning_rate": 1.9367588932806324e-05,
-      "loss": 3.9689,
+      "grad_norm": 3.4899723529815674,
+      "learning_rate": 2.9051383399209485e-05,
+      "loss": 3.7187,
       "step": 840
     },
     {
       "epoch": 0.15998494259363824,
-      "grad_norm": 3.343932867050171,
-      "learning_rate": 1.9360060229625447e-05,
-      "loss": 3.9531,
+      "grad_norm": 5.375092029571533,
+      "learning_rate": 2.904009034443817e-05,
+      "loss": 3.6486,
       "step": 850
     },
     {
       "epoch": 0.1618671183888575,
-      "grad_norm": 3.245112895965576,
-      "learning_rate": 1.9352531526444573e-05,
-      "loss": 4.0125,
+      "grad_norm": 4.234030246734619,
+      "learning_rate": 2.9028797289666854e-05,
+      "loss": 3.7097,
       "step": 860
     },
     {
       "epoch": 0.1637492941840768,
-      "grad_norm": 3.1948068141937256,
-      "learning_rate": 1.9345002823263696e-05,
-      "loss": 3.7966,
+      "grad_norm": 3.9143476486206055,
+      "learning_rate": 2.901750423489554e-05,
+      "loss": 3.4356,
       "step": 870
     },
     {
       "epoch": 0.16563146997929606,
-      "grad_norm": 3.6105539798736572,
-      "learning_rate": 1.933747412008282e-05,
-      "loss": 3.8701,
+      "grad_norm": 3.578145742416382,
+      "learning_rate": 2.9006211180124223e-05,
+      "loss": 3.4969,
       "step": 880
     },
     {
       "epoch": 0.16751364577451533,
-      "grad_norm": 3.940016031265259,
-      "learning_rate": 1.932994541690194e-05,
-      "loss": 3.8866,
+      "grad_norm": 4.751120567321777,
+      "learning_rate": 2.899491812535291e-05,
+      "loss": 3.5853,
       "step": 890
     },
     {
       "epoch": 0.16939582156973462,
-      "grad_norm": 4.30803918838501,
-      "learning_rate": 1.9322416713721065e-05,
-      "loss": 3.9068,
+      "grad_norm": 4.476664066314697,
+      "learning_rate": 2.8983625070581595e-05,
+      "loss": 3.5424,
       "step": 900
     },
     {
       "epoch": 0.1712779973649539,
-      "grad_norm": 3.2661314010620117,
-      "learning_rate": 1.9314888010540187e-05,
-      "loss": 3.8431,
+      "grad_norm": 3.4918501377105713,
+      "learning_rate": 2.8972332015810278e-05,
+      "loss": 3.5004,
       "step": 910
     },
     {
       "epoch": 0.17316017316017315,
-      "grad_norm": 4.278161525726318,
-      "learning_rate": 1.9307359307359307e-05,
-      "loss": 3.8407,
+      "grad_norm": 9.847784996032715,
+      "learning_rate": 2.896103896103896e-05,
+      "loss": 3.5159,
       "step": 920
     },
     {
       "epoch": 0.17504234895539245,
-      "grad_norm": 3.765481472015381,
-      "learning_rate": 1.929983060417843e-05,
-      "loss": 3.8293,
+      "grad_norm": 3.551976442337036,
+      "learning_rate": 2.8949745906267647e-05,
+      "loss": 3.475,
       "step": 930
     },
     {
       "epoch": 0.1769245247506117,
-      "grad_norm": 4.025678634643555,
-      "learning_rate": 1.9292301900997553e-05,
-      "loss": 3.8125,
+      "grad_norm": 4.220102787017822,
+      "learning_rate": 2.893845285149633e-05,
+      "loss": 3.4396,
       "step": 940
     },
     {
       "epoch": 0.17880670054583098,
-      "grad_norm": 3.7228808403015137,
-      "learning_rate": 1.9284773197816676e-05,
-      "loss": 3.7891,
+      "grad_norm": 6.077229976654053,
+      "learning_rate": 2.8927159796725015e-05,
+      "loss": 3.448,
       "step": 950
     },
     {
       "epoch": 0.18068887634105024,
-      "grad_norm": 3.49621844291687,
-      "learning_rate": 1.9277244494635802e-05,
-      "loss": 3.7859,
+      "grad_norm": 4.203821659088135,
+      "learning_rate": 2.8915866741953698e-05,
+      "loss": 3.4436,
       "step": 960
     },
     {
       "epoch": 0.18257105213626953,
-      "grad_norm": 3.645171880722046,
-      "learning_rate": 1.9269715791454925e-05,
-      "loss": 3.7994,
+      "grad_norm": 3.7120420932769775,
+      "learning_rate": 2.8904573687182384e-05,
+      "loss": 3.416,
       "step": 970
     },
     {
       "epoch": 0.1844532279314888,
-      "grad_norm": 4.006845951080322,
-      "learning_rate": 1.9262187088274048e-05,
-      "loss": 3.7183,
+      "grad_norm": 4.179491996765137,
+      "learning_rate": 2.889328063241107e-05,
+      "loss": 3.3503,
       "step": 980
     },
     {
       "epoch": 0.18633540372670807,
-      "grad_norm": 3.9820406436920166,
-      "learning_rate": 1.925465838509317e-05,
-      "loss": 3.8167,
+      "grad_norm": 3.9055488109588623,
+      "learning_rate": 2.8881987577639753e-05,
+      "loss": 3.4424,
       "step": 990
     },
     {
       "epoch": 0.18821757952192736,
-      "grad_norm": 3.6761341094970703,
-      "learning_rate": 1.9247129681912294e-05,
-      "loss": 3.7398,
+      "grad_norm": 3.6015703678131104,
+      "learning_rate": 2.887069452286844e-05,
+      "loss": 3.3619,
       "step": 1000
     },
     {
       "epoch": 0.19009975531714662,
-      "grad_norm": 4.603309154510498,
-      "learning_rate": 1.9239600978731413e-05,
-      "loss": 3.7817,
+      "grad_norm": 5.409115314483643,
+      "learning_rate": 2.8859401468097118e-05,
+      "loss": 3.4609,
       "step": 1010
     },
     {
       "epoch": 0.1919819311123659,
-      "grad_norm": 3.3203389644622803,
-      "learning_rate": 1.9232072275550536e-05,
-      "loss": 3.7976,
+      "grad_norm": 3.458120584487915,
+      "learning_rate": 2.8848108413325804e-05,
+      "loss": 3.4654,
       "step": 1020
     },
     {
       "epoch": 0.19386410690758518,
-      "grad_norm": 3.973386287689209,
-      "learning_rate": 1.922454357236966e-05,
-      "loss": 3.746,
+      "grad_norm": 3.861100673675537,
+      "learning_rate": 2.883681535855449e-05,
+      "loss": 3.3997,
       "step": 1030
     },
     {
       "epoch": 0.19574628270280445,
-      "grad_norm": 5.21986722946167,
-      "learning_rate": 1.9217014869188782e-05,
-      "loss": 3.6609,
+      "grad_norm": 8.975924491882324,
+      "learning_rate": 2.8825522303783173e-05,
+      "loss": 3.297,
       "step": 1040
     },
     {
       "epoch": 0.1976284584980237,
-      "grad_norm": 8.304903030395508,
-      "learning_rate": 1.920948616600791e-05,
-      "loss": 3.693,
+      "grad_norm": 9.686234474182129,
+      "learning_rate": 2.881422924901186e-05,
+      "loss": 3.3386,
       "step": 1050
     },
     {
       "epoch": 0.19951063429324298,
-      "grad_norm": 3.812041759490967,
-      "learning_rate": 1.920195746282703e-05,
-      "loss": 3.748,
+      "grad_norm": 6.7616658210754395,
+      "learning_rate": 2.8802936194240542e-05,
+      "loss": 3.3721,
       "step": 1060
     },
     {
       "epoch": 0.20139281008846227,
-      "grad_norm": 3.4189064502716064,
-      "learning_rate": 1.9194428759646154e-05,
-      "loss": 3.7635,
+      "grad_norm": 3.126573324203491,
+      "learning_rate": 2.8791643139469228e-05,
+      "loss": 3.4284,
       "step": 1070
     },
     {
       "epoch": 0.20327498588368154,
-      "grad_norm": 3.6516013145446777,
-      "learning_rate": 1.9186900056465277e-05,
-      "loss": 3.6519,
+      "grad_norm": 4.771273136138916,
+      "learning_rate": 2.8780350084697914e-05,
+      "loss": 3.2652,
       "step": 1080
     },
     {
       "epoch": 0.2051571616789008,
-      "grad_norm": 3.0502750873565674,
-      "learning_rate": 1.9179371353284397e-05,
-      "loss": 3.7235,
+      "grad_norm": 3.69577956199646,
+      "learning_rate": 2.8769057029926593e-05,
+      "loss": 3.3577,
       "step": 1090
     },
     {
       "epoch": 0.2070393374741201,
-      "grad_norm": 7.05238151550293,
-      "learning_rate": 1.917184265010352e-05,
-      "loss": 3.7494,
+      "grad_norm": 4.15922212600708,
+      "learning_rate": 2.875776397515528e-05,
+      "loss": 3.4093,
       "step": 1100
     },
     {
       "epoch": 0.20892151326933936,
-      "grad_norm": 5.699125289916992,
-      "learning_rate": 1.9164313946922643e-05,
-      "loss": 3.6823,
+      "grad_norm": 15.523391723632812,
+      "learning_rate": 2.8746470920383962e-05,
+      "loss": 3.2856,
       "step": 1110
     },
     {
       "epoch": 0.21080368906455862,
-      "grad_norm": 5.011223793029785,
-      "learning_rate": 1.9156785243741765e-05,
-      "loss": 3.6604,
+      "grad_norm": 3.744535207748413,
+      "learning_rate": 2.8735177865612648e-05,
+      "loss": 3.344,
       "step": 1120
     },
     {
       "epoch": 0.2126858648597779,
-      "grad_norm": 4.8337788581848145,
-      "learning_rate": 1.914925654056089e-05,
-      "loss": 3.5912,
+      "grad_norm": 4.708155632019043,
+      "learning_rate": 2.8723884810841334e-05,
+      "loss": 3.2039,
       "step": 1130
     },
     {
       "epoch": 0.21456804065499718,
-      "grad_norm": 3.9982621669769287,
-      "learning_rate": 1.9141727837380015e-05,
-      "loss": 3.636,
+      "grad_norm": 3.991323232650757,
+      "learning_rate": 2.8712591756070017e-05,
+      "loss": 3.276,
       "step": 1140
     },
     {
       "epoch": 0.21645021645021645,
-      "grad_norm": 6.422266006469727,
-      "learning_rate": 1.9134199134199138e-05,
-      "loss": 3.6918,
+      "grad_norm": 6.3713154792785645,
+      "learning_rate": 2.8701298701298703e-05,
+      "loss": 3.3719,
       "step": 1150
     },
     {
       "epoch": 0.2183323922454357,
-      "grad_norm": 3.5577023029327393,
-      "learning_rate": 1.912667043101826e-05,
-      "loss": 3.645,
+      "grad_norm": 3.9468255043029785,
+      "learning_rate": 2.869000564652739e-05,
+      "loss": 3.2762,
       "step": 1160
     },
     {
       "epoch": 0.220214568040655,
-      "grad_norm": 3.5199527740478516,
-      "learning_rate": 1.9119141727837383e-05,
-      "loss": 3.5279,
+      "grad_norm": 3.5731687545776367,
+      "learning_rate": 2.8678712591756072e-05,
+      "loss": 3.1112,
       "step": 1170
     },
     {
       "epoch": 0.22209674383587427,
-      "grad_norm": 3.93874192237854,
-      "learning_rate": 1.9111613024656503e-05,
-      "loss": 3.591,
+      "grad_norm": 5.240946292877197,
+      "learning_rate": 2.8667419536984754e-05,
+      "loss": 3.1546,
       "step": 1180
     },
     {
       "epoch": 0.22397891963109354,
-      "grad_norm": 3.5050225257873535,
-      "learning_rate": 1.9104084321475626e-05,
-      "loss": 3.6323,
+      "grad_norm": 3.877687931060791,
+      "learning_rate": 2.8656126482213437e-05,
+      "loss": 3.2343,
       "step": 1190
     },
     {
       "epoch": 0.22586109542631283,
-      "grad_norm": 3.3720874786376953,
-      "learning_rate": 1.909655561829475e-05,
-      "loss": 3.6192,
+      "grad_norm": 4.6856584548950195,
+      "learning_rate": 2.8644833427442123e-05,
+      "loss": 3.2184,
       "step": 1200
     },
     {
       "epoch": 0.2277432712215321,
-      "grad_norm": 2.833101272583008,
-      "learning_rate": 1.9089026915113872e-05,
-      "loss": 3.544,
+      "grad_norm": 2.8222122192382812,
+      "learning_rate": 2.863354037267081e-05,
+      "loss": 3.1481,
       "step": 1210
     },
     {
       "epoch": 0.22962544701675136,
-      "grad_norm": 3.4263570308685303,
-      "learning_rate": 1.9081498211932995e-05,
-      "loss": 3.606,
+      "grad_norm": 2.7809884548187256,
+      "learning_rate": 2.8622247317899492e-05,
+      "loss": 3.21,
       "step": 1220
     },
     {
       "epoch": 0.23150762281197063,
-      "grad_norm": 4.502038955688477,
-      "learning_rate": 1.907396950875212e-05,
-      "loss": 3.5418,
+      "grad_norm": 4.449065208435059,
+      "learning_rate": 2.8610954263128178e-05,
+      "loss": 3.1756,
       "step": 1230
     },
     {
       "epoch": 0.23338979860718992,
-      "grad_norm": 4.862933158874512,
-      "learning_rate": 1.9066440805571244e-05,
-      "loss": 3.5316,
+      "grad_norm": 5.27800989151001,
+      "learning_rate": 2.859966120835686e-05,
+      "loss": 3.1003,
       "step": 1240
     },
     {
       "epoch": 0.23527197440240918,
-      "grad_norm": 7.199465751647949,
-      "learning_rate": 1.9058912102390367e-05,
-      "loss": 3.6576,
+      "grad_norm": 6.26723051071167,
+      "learning_rate": 2.8588368153585547e-05,
+      "loss": 3.3139,
       "step": 1250
     },
     {
       "epoch": 0.23715415019762845,
-      "grad_norm": 8.997977256774902,
-      "learning_rate": 1.905138339920949e-05,
-      "loss": 3.3866,
+      "grad_norm": 10.930907249450684,
+      "learning_rate": 2.8577075098814233e-05,
+      "loss": 2.9757,
       "step": 1260
     },
     {
       "epoch": 0.23903632599284774,
-      "grad_norm": 3.7024660110473633,
-      "learning_rate": 1.904385469602861e-05,
-      "loss": 3.633,
+      "grad_norm": 8.510289192199707,
+      "learning_rate": 2.8565782044042912e-05,
+      "loss": 3.2473,
       "step": 1270
     },
     {
       "epoch": 0.240918501788067,
-      "grad_norm": 4.383315563201904,
-      "learning_rate": 1.9036325992847732e-05,
-      "loss": 3.5204,
+      "grad_norm": 9.550591468811035,
+      "learning_rate": 2.8554488989271598e-05,
+      "loss": 3.0974,
       "step": 1280
     },
     {
       "epoch": 0.24280067758328627,
-      "grad_norm": 3.845693588256836,
-      "learning_rate": 1.9028797289666855e-05,
-      "loss": 3.4319,
+      "grad_norm": 3.702929735183716,
+      "learning_rate": 2.854319593450028e-05,
+      "loss": 3.0416,
       "step": 1290
     },
     {
       "epoch": 0.24468285337850557,
-      "grad_norm": 5.626176357269287,
-      "learning_rate": 1.9021268586485978e-05,
-      "loss": 3.6167,
+      "grad_norm": 3.1774191856384277,
+      "learning_rate": 2.8531902879728967e-05,
+      "loss": 3.2329,
       "step": 1300
     },
     {
       "epoch": 0.24656502917372483,
-      "grad_norm": 3.270461320877075,
-      "learning_rate": 1.90137398833051e-05,
-      "loss": 3.5568,
+      "grad_norm": 3.9982802867889404,
+      "learning_rate": 2.8520609824957653e-05,
+      "loss": 3.145,
       "step": 1310
     },
     {
       "epoch": 0.2484472049689441,
-      "grad_norm": 6.196926116943359,
-      "learning_rate": 1.9006211180124224e-05,
-      "loss": 3.5659,
+      "grad_norm": 10.488585472106934,
+      "learning_rate": 2.8509316770186336e-05,
+      "loss": 3.1751,
       "step": 1320
     },
     {
       "epoch": 0.2503293807641634,
-      "grad_norm": 3.670217514038086,
-      "learning_rate": 1.899868247694335e-05,
-      "loss": 3.5009,
+      "grad_norm": 4.030777454376221,
+      "learning_rate": 2.8498023715415022e-05,
+      "loss": 3.0312,
       "step": 1330
     },
     {
       "epoch": 0.25221155655938265,
-      "grad_norm": 3.7540853023529053,
-      "learning_rate": 1.8991153773762473e-05,
-      "loss": 3.4929,
+      "grad_norm": 5.116481304168701,
+      "learning_rate": 2.8486730660643705e-05,
+      "loss": 3.0881,
       "step": 1340
     },
     {
       "epoch": 0.2540937323546019,
-      "grad_norm": 5.680825710296631,
-      "learning_rate": 1.8983625070581596e-05,
-      "loss": 3.4018,
+      "grad_norm": 6.651054382324219,
+      "learning_rate": 2.847543760587239e-05,
+      "loss": 2.9569,
       "step": 1350
     },
     {
       "epoch": 0.2559759081498212,
-      "grad_norm": 5.842689514160156,
-      "learning_rate": 1.8976096367400716e-05,
-      "loss": 3.4934,
+      "grad_norm": 7.519700050354004,
+      "learning_rate": 2.8464144551101073e-05,
+      "loss": 3.1106,
       "step": 1360
     },
     {
       "epoch": 0.25785808394504045,
-      "grad_norm": 3.461512804031372,
-      "learning_rate": 1.896856766421984e-05,
-      "loss": 3.526,
+      "grad_norm": 5.690998554229736,
+      "learning_rate": 2.8452851496329756e-05,
+      "loss": 3.176,
       "step": 1370
     },
     {
       "epoch": 0.2597402597402597,
-      "grad_norm": 4.65621280670166,
-      "learning_rate": 1.896103896103896e-05,
-      "loss": 3.4837,
+      "grad_norm": 7.928246021270752,
+      "learning_rate": 2.8441558441558442e-05,
+      "loss": 3.0882,
       "step": 1380
     },
     {
       "epoch": 0.26162243553547904,
-      "grad_norm": 3.209920644760132,
-      "learning_rate": 1.8953510257858084e-05,
-      "loss": 3.475,
+      "grad_norm": 4.824883460998535,
+      "learning_rate": 2.8430265386787128e-05,
+      "loss": 3.0282,
       "step": 1390
     },
     {
       "epoch": 0.2635046113306983,
-      "grad_norm": 4.2360615730285645,
-      "learning_rate": 1.8945981554677207e-05,
-      "loss": 3.3463,
+      "grad_norm": 5.6374640464782715,
+      "learning_rate": 2.841897233201581e-05,
+      "loss": 2.8817,
       "step": 1400
     },
     {
       "epoch": 0.26538678712591757,
-      "grad_norm": 3.977550506591797,
-      "learning_rate": 1.893845285149633e-05,
-      "loss": 3.4359,
+      "grad_norm": 5.54218053817749,
+      "learning_rate": 2.8407679277244497e-05,
+      "loss": 3.0196,
       "step": 1410
     },
     {
       "epoch": 0.26726896292113683,
-      "grad_norm": 3.533844470977783,
-      "learning_rate": 1.8930924148315456e-05,
-      "loss": 3.3868,
+      "grad_norm": 3.4347853660583496,
+      "learning_rate": 2.839638622247318e-05,
+      "loss": 2.9811,
       "step": 1420
     },
     {
       "epoch": 0.2691511387163561,
-      "grad_norm": 4.788166522979736,
-      "learning_rate": 1.892339544513458e-05,
-      "loss": 3.3756,
+      "grad_norm": 3.8664019107818604,
+      "learning_rate": 2.8385093167701866e-05,
+      "loss": 2.9376,
       "step": 1430
     },
     {
       "epoch": 0.27103331451157536,
-      "grad_norm": 4.711440086364746,
-      "learning_rate": 1.89158667419537e-05,
-      "loss": 3.5177,
+      "grad_norm": 5.806457996368408,
+      "learning_rate": 2.837380011293055e-05,
+      "loss": 3.1184,
       "step": 1440
     },
     {
       "epoch": 0.27291549030679463,
-      "grad_norm": 3.7958436012268066,
-      "learning_rate": 1.8908338038772822e-05,
-      "loss": 3.3974,
+      "grad_norm": 4.753952980041504,
+      "learning_rate": 2.836250705815923e-05,
+      "loss": 2.9645,
       "step": 1450
     },
     {
       "epoch": 0.27479766610201395,
-      "grad_norm": 5.754987716674805,
-      "learning_rate": 1.8900809335591945e-05,
-      "loss": 3.3473,
+      "grad_norm": 4.03913688659668,
+      "learning_rate": 2.8351214003387917e-05,
+      "loss": 2.8864,
       "step": 1460
     },
     {
       "epoch": 0.2766798418972332,
-      "grad_norm": 6.3976311683654785,
-      "learning_rate": 1.8893280632411068e-05,
-      "loss": 3.4912,
+      "grad_norm": 8.83630657196045,
+      "learning_rate": 2.83399209486166e-05,
+      "loss": 3.0768,
       "step": 1470
     },
     {
       "epoch": 0.2785620176924525,
-      "grad_norm": 6.161326885223389,
-      "learning_rate": 1.888575192923019e-05,
-      "loss": 3.3726,
+      "grad_norm": 7.6859540939331055,
+      "learning_rate": 2.8328627893845286e-05,
+      "loss": 2.915,
       "step": 1480
     },
     {
       "epoch": 0.28044419348767174,
-      "grad_norm": 4.482143878936768,
-      "learning_rate": 1.8878223226049314e-05,
-      "loss": 3.3857,
+      "grad_norm": 4.7729811668396,
+      "learning_rate": 2.8317334839073972e-05,
+      "loss": 2.9354,
       "step": 1490
     },
     {
       "epoch": 0.282326369282891,
-      "grad_norm": 6.511773586273193,
-      "learning_rate": 1.8870694522868436e-05,
-      "loss": 3.3012,
+      "grad_norm": 6.917285442352295,
+      "learning_rate": 2.8306041784302655e-05,
+      "loss": 2.8416,
       "step": 1500
     },
     {
       "epoch": 0.2842085450781103,
-      "grad_norm": 3.7000627517700195,
-      "learning_rate": 1.8863165819687563e-05,
-      "loss": 3.4734,
+      "grad_norm": 4.168449878692627,
+      "learning_rate": 2.829474872953134e-05,
+      "loss": 3.0573,
       "step": 1510
     },
     {
       "epoch": 0.2860907208733296,
-      "grad_norm": 3.221539258956909,
-      "learning_rate": 1.8855637116506686e-05,
-      "loss": 3.3935,
+      "grad_norm": 3.04876708984375,
+      "learning_rate": 2.8283455674760023e-05,
+      "loss": 2.9579,
       "step": 1520
     },
     {
       "epoch": 0.28797289666854886,
-      "grad_norm": 4.660919189453125,
-      "learning_rate": 1.8848108413325805e-05,
-      "loss": 3.4218,
+      "grad_norm": 6.365512847900391,
+      "learning_rate": 2.8272162619988706e-05,
+      "loss": 2.9621,
       "step": 1530
     },
     {
       "epoch": 0.2898550724637681,
-      "grad_norm": 3.7305450439453125,
-      "learning_rate": 1.8840579710144928e-05,
-      "loss": 3.3499,
+      "grad_norm": 4.556066989898682,
+      "learning_rate": 2.8260869565217392e-05,
+      "loss": 2.9296,
       "step": 1540
     },
     {
       "epoch": 0.2917372482589874,
-      "grad_norm": 4.458515644073486,
-      "learning_rate": 1.883305100696405e-05,
-      "loss": 3.243,
+      "grad_norm": 2.86582088470459,
+      "learning_rate": 2.8249576510446075e-05,
+      "loss": 2.7293,
       "step": 1550
     },
     {
       "epoch": 0.29361942405420666,
-      "grad_norm": 4.746747970581055,
-      "learning_rate": 1.8825522303783174e-05,
-      "loss": 3.4016,
+      "grad_norm": 4.107886791229248,
+      "learning_rate": 2.823828345567476e-05,
+      "loss": 3.0,
       "step": 1560
     },
     {
       "epoch": 0.2955015998494259,
-      "grad_norm": 4.314861297607422,
-      "learning_rate": 1.8817993600602297e-05,
-      "loss": 3.3974,
+      "grad_norm": 5.17814826965332,
+      "learning_rate": 2.8226990400903444e-05,
+      "loss": 2.9396,
       "step": 1570
     },
     {
       "epoch": 0.2973837756446452,
-      "grad_norm": 3.778343677520752,
-      "learning_rate": 1.881046489742142e-05,
-      "loss": 3.3342,
+      "grad_norm": 3.8069369792938232,
+      "learning_rate": 2.821569734613213e-05,
+      "loss": 2.8673,
       "step": 1580
     },
     {
       "epoch": 0.2992659514398645,
-      "grad_norm": 3.7303268909454346,
-      "learning_rate": 1.8802936194240543e-05,
-      "loss": 3.3313,
+      "grad_norm": 4.464517593383789,
+      "learning_rate": 2.8204404291360816e-05,
+      "loss": 2.8787,
       "step": 1590
     },
     {
       "epoch": 0.3011481272350838,
-      "grad_norm": 3.9831275939941406,
-      "learning_rate": 1.8795407491059666e-05,
-      "loss": 3.214,
+      "grad_norm": 6.4331536293029785,
+      "learning_rate": 2.81931112365895e-05,
+      "loss": 2.7073,
       "step": 1600
     },
     {
       "epoch": 0.30303030303030304,
-      "grad_norm": 4.6280975341796875,
-      "learning_rate": 1.8787878787878792e-05,
-      "loss": 3.2715,
+      "grad_norm": 5.987295627593994,
+      "learning_rate": 2.8181818181818185e-05,
+      "loss": 2.7962,
       "step": 1610
     },
     {
       "epoch": 0.3049124788255223,
-      "grad_norm": 5.433199405670166,
-      "learning_rate": 1.878035008469791e-05,
-      "loss": 3.4141,
+      "grad_norm": 5.483144760131836,
+      "learning_rate": 2.8170525127046864e-05,
+      "loss": 2.9668,
       "step": 1620
     },
     {
       "epoch": 0.30679465462074157,
-      "grad_norm": 3.2537083625793457,
-      "learning_rate": 1.8772821381517034e-05,
-      "loss": 3.2985,
+      "grad_norm": 3.3265535831451416,
+      "learning_rate": 2.815923207227555e-05,
+      "loss": 2.8663,
       "step": 1630
     },
     {
       "epoch": 0.30867683041596083,
-      "grad_norm": 6.475714683532715,
-      "learning_rate": 1.8765292678336157e-05,
-      "loss": 3.3312,
+      "grad_norm": 6.472630500793457,
+      "learning_rate": 2.8147939017504236e-05,
+      "loss": 2.8595,
       "step": 1640
     },
     {
       "epoch": 0.3105590062111801,
-      "grad_norm": 3.690768003463745,
-      "learning_rate": 1.875776397515528e-05,
-      "loss": 3.4284,
+      "grad_norm": 4.674870491027832,
+      "learning_rate": 2.813664596273292e-05,
+      "loss": 2.9178,
       "step": 1650
     },
     {
       "epoch": 0.3124411820063994,
-      "grad_norm": 4.372195243835449,
-      "learning_rate": 1.8750235271974403e-05,
-      "loss": 3.2672,
+      "grad_norm": 4.7341628074646,
+      "learning_rate": 2.8125352907961605e-05,
+      "loss": 2.7904,
       "step": 1660
     },
     {
       "epoch": 0.3143233578016187,
-      "grad_norm": 3.0955049991607666,
-      "learning_rate": 1.8742706568793526e-05,
-      "loss": 3.2831,
+      "grad_norm": 3.6063005924224854,
+      "learning_rate": 2.811405985319029e-05,
+      "loss": 2.8127,
       "step": 1670
     },
     {
       "epoch": 0.31620553359683795,
-      "grad_norm": 5.1684112548828125,
-      "learning_rate": 1.873517786561265e-05,
-      "loss": 3.2791,
+      "grad_norm": 10.121143341064453,
+      "learning_rate": 2.8102766798418974e-05,
+      "loss": 2.8588,
       "step": 1680
     },
     {
       "epoch": 0.3180877093920572,
-      "grad_norm": 3.559154510498047,
-      "learning_rate": 1.8727649162431772e-05,
-      "loss": 3.2338,
+      "grad_norm": 5.036760330200195,
+      "learning_rate": 2.809147374364766e-05,
+      "loss": 2.7812,
       "step": 1690
     },
     {
       "epoch": 0.3199698851872765,
-      "grad_norm": 6.47905158996582,
-      "learning_rate": 1.8720120459250895e-05,
-      "loss": 3.2949,
+      "grad_norm": 8.621694564819336,
+      "learning_rate": 2.808018068887634e-05,
+      "loss": 2.8555,
       "step": 1700
     },
     {
       "epoch": 0.32185206098249575,
-      "grad_norm": 4.15554141998291,
-      "learning_rate": 1.8712591756070018e-05,
-      "loss": 3.2074,
+      "grad_norm": 6.541132926940918,
+      "learning_rate": 2.8068887634105025e-05,
+      "loss": 2.794,
       "step": 1710
     },
     {
       "epoch": 0.323734236777715,
-      "grad_norm": 5.843398094177246,
-      "learning_rate": 1.870506305288914e-05,
-      "loss": 3.1397,
+      "grad_norm": 5.8154425621032715,
+      "learning_rate": 2.805759457933371e-05,
+      "loss": 2.6459,
       "step": 1720
     },
     {
       "epoch": 0.32561641257293433,
-      "grad_norm": 4.45064640045166,
-      "learning_rate": 1.8697534349708264e-05,
-      "loss": 3.2493,
+      "grad_norm": 3.9110445976257324,
+      "learning_rate": 2.8046301524562394e-05,
+      "loss": 2.7375,
       "step": 1730
     },
     {
       "epoch": 0.3274985883681536,
-      "grad_norm": 3.1222758293151855,
-      "learning_rate": 1.8690005646527387e-05,
-      "loss": 3.2675,
+      "grad_norm": 3.47908878326416,
+      "learning_rate": 2.803500846979108e-05,
+      "loss": 2.8092,
       "step": 1740
     },
     {
       "epoch": 0.32938076416337286,
-      "grad_norm": 4.332676410675049,
-      "learning_rate": 1.868247694334651e-05,
-      "loss": 3.2176,
+      "grad_norm": 4.073140621185303,
+      "learning_rate": 2.8023715415019763e-05,
+      "loss": 2.7205,
       "step": 1750
     },
     {
       "epoch": 0.33126293995859213,
-      "grad_norm": 6.784658908843994,
-      "learning_rate": 1.8674948240165632e-05,
-      "loss": 3.194,
+      "grad_norm": 9.20858097076416,
+      "learning_rate": 2.801242236024845e-05,
+      "loss": 2.6537,
       "step": 1760
     },
     {
       "epoch": 0.3331451157538114,
-      "grad_norm": 3.7953319549560547,
-      "learning_rate": 1.8667419536984755e-05,
-      "loss": 3.2735,
+      "grad_norm": 3.811323881149292,
+      "learning_rate": 2.8001129305477135e-05,
+      "loss": 2.7925,
       "step": 1770
     },
     {
       "epoch": 0.33502729154903066,
-      "grad_norm": 3.5565590858459473,
-      "learning_rate": 1.8659890833803878e-05,
-      "loss": 3.2094,
+      "grad_norm": 3.6044814586639404,
+      "learning_rate": 2.7989836250705817e-05,
+      "loss": 2.7287,
       "step": 1780
     },
     {
       "epoch": 0.33690946734425,
-      "grad_norm": 3.9369053840637207,
-      "learning_rate": 1.8652362130623e-05,
-      "loss": 3.0813,
+      "grad_norm": 5.667727470397949,
+      "learning_rate": 2.79785431959345e-05,
+      "loss": 2.5265,
       "step": 1790
     },
     {
       "epoch": 0.33879164313946925,
-      "grad_norm": 7.771551609039307,
-      "learning_rate": 1.8644833427442124e-05,
-      "loss": 3.1139,
+      "grad_norm": 4.497561931610107,
+      "learning_rate": 2.7967250141163183e-05,
+      "loss": 2.5919,
       "step": 1800
     },
     {
       "epoch": 0.3406738189346885,
-      "grad_norm": 4.863523960113525,
-      "learning_rate": 1.8637304724261247e-05,
-      "loss": 3.2591,
+      "grad_norm": 6.864144802093506,
+      "learning_rate": 2.795595708639187e-05,
+      "loss": 2.7344,
       "step": 1810
     },
     {
       "epoch": 0.3425559947299078,
-      "grad_norm": 7.337231636047363,
-      "learning_rate": 1.862977602108037e-05,
-      "loss": 3.1937,
+      "grad_norm": 5.193475723266602,
+      "learning_rate": 2.7944664031620555e-05,
+      "loss": 2.7341,
       "step": 1820
     },
     {
       "epoch": 0.34443817052512704,
-      "grad_norm": 4.414863109588623,
-      "learning_rate": 1.8622247317899493e-05,
-      "loss": 3.1138,
+      "grad_norm": 10.060043334960938,
+      "learning_rate": 2.7933370976849238e-05,
+      "loss": 2.6031,
       "step": 1830
     },
     {
       "epoch": 0.3463203463203463,
-      "grad_norm": 5.118075370788574,
-      "learning_rate": 1.8614718614718616e-05,
-      "loss": 3.1822,
+      "grad_norm": 7.921087265014648,
+      "learning_rate": 2.7922077922077924e-05,
+      "loss": 2.6902,
       "step": 1840
     },
     {
       "epoch": 0.34820252211556557,
-      "grad_norm": 5.352917194366455,
-      "learning_rate": 1.860718991153774e-05,
-      "loss": 3.1788,
+      "grad_norm": 5.25139856338501,
+      "learning_rate": 2.7910784867306606e-05,
+      "loss": 2.7383,
       "step": 1850
     },
     {
       "epoch": 0.3500846979107849,
-      "grad_norm": 8.42982006072998,
-      "learning_rate": 1.859966120835686e-05,
-      "loss": 3.1031,
+      "grad_norm": 9.80843734741211,
+      "learning_rate": 2.7899491812535292e-05,
+      "loss": 2.5868,
       "step": 1860
     },
     {
       "epoch": 0.35196687370600416,
-      "grad_norm": 7.573580741882324,
-      "learning_rate": 1.8592132505175985e-05,
-      "loss": 3.1666,
+      "grad_norm": 8.038129806518555,
+      "learning_rate": 2.788819875776398e-05,
+      "loss": 2.6574,
       "step": 1870
     },
     {
       "epoch": 0.3538490495012234,
-      "grad_norm": 3.9582431316375732,
-      "learning_rate": 1.8584603801995107e-05,
-      "loss": 3.1233,
+      "grad_norm": 5.235549449920654,
+      "learning_rate": 2.7876905702992658e-05,
+      "loss": 2.623,
       "step": 1880
     },
     {
       "epoch": 0.3557312252964427,
-      "grad_norm": 3.29245924949646,
-      "learning_rate": 1.857707509881423e-05,
-      "loss": 3.1,
+      "grad_norm": 3.0741400718688965,
+      "learning_rate": 2.7865612648221344e-05,
+      "loss": 2.6216,
       "step": 1890
     },
     {
       "epoch": 0.35761340109166195,
-      "grad_norm": 4.551001071929932,
-      "learning_rate": 1.8569546395633353e-05,
-      "loss": 3.1246,
+      "grad_norm": 3.884474515914917,
+      "learning_rate": 2.785431959345003e-05,
+      "loss": 2.5854,
       "step": 1900
     },
     {
       "epoch": 0.3594955768868812,
-      "grad_norm": 3.6506519317626953,
-      "learning_rate": 1.8562017692452476e-05,
-      "loss": 2.9412,
+      "grad_norm": 3.7831873893737793,
+      "learning_rate": 2.7843026538678713e-05,
+      "loss": 2.3631,
       "step": 1910
     },
     {
       "epoch": 0.3613777526821005,
-      "grad_norm": 4.867851257324219,
-      "learning_rate": 1.85544889892716e-05,
-      "loss": 3.1142,
+      "grad_norm": 6.738157272338867,
+      "learning_rate": 2.78317334839074e-05,
+      "loss": 2.6551,
       "step": 1920
     },
     {
       "epoch": 0.3632599284773198,
-      "grad_norm": 6.288889408111572,
-      "learning_rate": 1.8546960286090722e-05,
-      "loss": 3.1336,
+      "grad_norm": 5.1181793212890625,
+      "learning_rate": 2.782044042913608e-05,
+      "loss": 2.6403,
       "step": 1930
     },
     {
       "epoch": 0.36514210427253907,
-      "grad_norm": 4.735827922821045,
-      "learning_rate": 1.8539431582909845e-05,
-      "loss": 2.9844,
+      "grad_norm": 8.148448944091797,
+      "learning_rate": 2.7809147374364768e-05,
+      "loss": 2.4368,
       "step": 1940
     },
     {
       "epoch": 0.36702428006775834,
-      "grad_norm": 9.145768165588379,
-      "learning_rate": 1.8531902879728968e-05,
-      "loss": 3.1812,
+      "grad_norm": 14.908823013305664,
+      "learning_rate": 2.7797854319593454e-05,
+      "loss": 2.7317,
       "step": 1950
     },
     {
       "epoch": 0.3689064558629776,
-      "grad_norm": 4.640198230743408,
-      "learning_rate": 1.852437417654809e-05,
-      "loss": 3.0759,
+      "grad_norm": 4.497227668762207,
+      "learning_rate": 2.7786561264822133e-05,
+      "loss": 2.5193,
       "step": 1960
     },
     {
       "epoch": 0.37078863165819687,
-      "grad_norm": 5.206761837005615,
-      "learning_rate": 1.8516845473367214e-05,
-      "loss": 3.1733,
+      "grad_norm": 7.139688491821289,
+      "learning_rate": 2.777526821005082e-05,
+      "loss": 2.6937,
       "step": 1970
     },
     {
       "epoch": 0.37267080745341613,
-      "grad_norm": 8.24525260925293,
-      "learning_rate": 1.8509316770186337e-05,
-      "loss": 2.8758,
+      "grad_norm": 11.658743858337402,
+      "learning_rate": 2.77639751552795e-05,
+      "loss": 2.3396,
       "step": 1980
     },
     {
       "epoch": 0.3745529832486354,
-      "grad_norm": 6.640933990478516,
-      "learning_rate": 1.850178806700546e-05,
-      "loss": 3.156,
+      "grad_norm": 11.21719741821289,
+      "learning_rate": 2.7752682100508188e-05,
+      "loss": 2.6433,
       "step": 1990
     },
     {
       "epoch": 0.3764351590438547,
-      "grad_norm": 3.1573383808135986,
-      "learning_rate": 1.8494259363824583e-05,
-      "loss": 3.0404,
+      "grad_norm": 3.650562047958374,
+      "learning_rate": 2.7741389045736874e-05,
+      "loss": 2.5423,
       "step": 2000
     },
     {
       "epoch": 0.378317334839074,
-      "grad_norm": 5.134864807128906,
-      "learning_rate": 1.8486730660643705e-05,
-      "loss": 2.9695,
+      "grad_norm": 6.978822231292725,
+      "learning_rate": 2.7730095990965556e-05,
+      "loss": 2.4149,
       "step": 2010
     },
     {
       "epoch": 0.38019951063429325,
-      "grad_norm": 7.871883392333984,
-      "learning_rate": 1.847920195746283e-05,
-      "loss": 3.1814,
+      "grad_norm": 12.115926742553711,
+      "learning_rate": 2.7718802936194243e-05,
+      "loss": 2.6864,
       "step": 2020
     },
     {
       "epoch": 0.3820816864295125,
-      "grad_norm": 12.194758415222168,
-      "learning_rate": 1.847167325428195e-05,
-      "loss": 3.0276,
+      "grad_norm": 12.849691390991211,
+      "learning_rate": 2.7707509881422925e-05,
+      "loss": 2.4976,
       "step": 2030
     },
     {
       "epoch": 0.3839638622247318,
-      "grad_norm": 6.0320820808410645,
-      "learning_rate": 1.8464144551101074e-05,
-      "loss": 2.9643,
+      "grad_norm": 7.091000556945801,
+      "learning_rate": 2.769621682665161e-05,
+      "loss": 2.3657,
       "step": 2040
     },
     {
       "epoch": 0.38584603801995104,
-      "grad_norm": 5.726856231689453,
-      "learning_rate": 1.8456615847920197e-05,
-      "loss": 2.9393,
+      "grad_norm": 7.3910746574401855,
+      "learning_rate": 2.7684923771880294e-05,
+      "loss": 2.4003,
       "step": 2050
     },
     {
       "epoch": 0.38772821381517036,
-      "grad_norm": 3.55429744720459,
-      "learning_rate": 1.844908714473932e-05,
-      "loss": 3.1167,
+      "grad_norm": 4.848266124725342,
+      "learning_rate": 2.7673630717108977e-05,
+      "loss": 2.6143,
       "step": 2060
     },
     {
       "epoch": 0.38961038961038963,
-      "grad_norm": 7.710501194000244,
-      "learning_rate": 1.8441558441558443e-05,
-      "loss": 3.0125,
+      "grad_norm": 5.94016170501709,
+      "learning_rate": 2.7662337662337663e-05,
+      "loss": 2.4341,
       "step": 2070
     },
     {
       "epoch": 0.3914925654056089,
-      "grad_norm": 6.08244514465332,
-      "learning_rate": 1.8434029738377566e-05,
-      "loss": 3.0516,
+      "grad_norm": 8.351221084594727,
+      "learning_rate": 2.7651044607566345e-05,
+      "loss": 2.4935,
       "step": 2080
     },
     {
       "epoch": 0.39337474120082816,
-      "grad_norm": 5.4757585525512695,
-      "learning_rate": 1.842650103519669e-05,
-      "loss": 2.8831,
+      "grad_norm": 8.837783813476562,
+      "learning_rate": 2.763975155279503e-05,
+      "loss": 2.3477,
       "step": 2090
     },
     {
       "epoch": 0.3952569169960474,
-      "grad_norm": 5.588348865509033,
-      "learning_rate": 1.8418972332015812e-05,
-      "loss": 2.963,
+      "grad_norm": 9.10621166229248,
+      "learning_rate": 2.7628458498023718e-05,
+      "loss": 2.4406,
       "step": 2100
     },
     {
       "epoch": 0.3971390927912667,
-      "grad_norm": 8.077258110046387,
-      "learning_rate": 1.8411443628834935e-05,
-      "loss": 3.085,
+      "grad_norm": 9.467292785644531,
+      "learning_rate": 2.76171654432524e-05,
+      "loss": 2.6356,
       "step": 2110
     },
     {
       "epoch": 0.39902126858648596,
-      "grad_norm": 3.8269598484039307,
-      "learning_rate": 1.8403914925654058e-05,
-      "loss": 2.9913,
+      "grad_norm": 5.230591297149658,
+      "learning_rate": 2.7605872388481086e-05,
+      "loss": 2.4345,
       "step": 2120
     },
     {
       "epoch": 0.4009034443817053,
-      "grad_norm": 3.7287516593933105,
-      "learning_rate": 1.839638622247318e-05,
-      "loss": 3.035,
+      "grad_norm": 5.1924591064453125,
+      "learning_rate": 2.759457933370977e-05,
+      "loss": 2.5366,
       "step": 2130
     },
     {
       "epoch": 0.40278562017692454,
-      "grad_norm": 5.027886390686035,
-      "learning_rate": 1.8388857519292303e-05,
-      "loss": 2.9356,
+      "grad_norm": 6.529165744781494,
+      "learning_rate": 2.7583286278938452e-05,
+      "loss": 2.3564,
       "step": 2140
     },
     {
       "epoch": 0.4046677959721438,
-      "grad_norm": 5.572052478790283,
-      "learning_rate": 1.8381328816111426e-05,
-      "loss": 2.8601,
+      "grad_norm": 7.249635219573975,
+      "learning_rate": 2.7571993224167138e-05,
+      "loss": 2.2548,
       "step": 2150
     },
     {
       "epoch": 0.40654997176736307,
-      "grad_norm": 6.4008355140686035,
-      "learning_rate": 1.837380011293055e-05,
-      "loss": 3.0857,
+      "grad_norm": 9.388668060302734,
+      "learning_rate": 2.756070016939582e-05,
+      "loss": 2.6006,
       "step": 2160
     },
     {
       "epoch": 0.40843214756258234,
-      "grad_norm": 4.075533866882324,
-      "learning_rate": 1.8366271409749672e-05,
-      "loss": 2.9145,
+      "grad_norm": 3.7543675899505615,
+      "learning_rate": 2.7549407114624507e-05,
+      "loss": 2.3548,
       "step": 2170
     },
     {
       "epoch": 0.4103143233578016,
-      "grad_norm": 7.462157249450684,
-      "learning_rate": 1.8358742706568795e-05,
-      "loss": 2.9953,
+      "grad_norm": 7.2959442138671875,
+      "learning_rate": 2.7538114059853193e-05,
+      "loss": 2.4884,
       "step": 2180
     },
     {
       "epoch": 0.41219649915302087,
-      "grad_norm": 4.182966232299805,
-      "learning_rate": 1.8351214003387918e-05,
-      "loss": 3.0412,
+      "grad_norm": 4.733274936676025,
+      "learning_rate": 2.7526821005081875e-05,
+      "loss": 2.5703,
       "step": 2190
     },
     {
       "epoch": 0.4140786749482402,
-      "grad_norm": 10.89699649810791,
-      "learning_rate": 1.834368530020704e-05,
-      "loss": 2.9925,
+      "grad_norm": 14.75526237487793,
+      "learning_rate": 2.751552795031056e-05,
+      "loss": 2.4666,
       "step": 2200
     },
     {
       "epoch": 0.41596085074345945,
-      "grad_norm": 5.336112022399902,
-      "learning_rate": 1.8336156597026164e-05,
-      "loss": 2.9516,
+      "grad_norm": 5.791693210601807,
+      "learning_rate": 2.7504234895539244e-05,
+      "loss": 2.4131,
       "step": 2210
     },
     {
       "epoch": 0.4178430265386787,
-      "grad_norm": 5.08997917175293,
-      "learning_rate": 1.8328627893845287e-05,
-      "loss": 2.8662,
+      "grad_norm": 5.518586158752441,
+      "learning_rate": 2.749294184076793e-05,
+      "loss": 2.2924,
       "step": 2220
     },
     {
       "epoch": 0.419725202333898,
-      "grad_norm": 8.328145027160645,
-      "learning_rate": 1.832109919066441e-05,
-      "loss": 2.844,
+      "grad_norm": 8.407965660095215,
+      "learning_rate": 2.7481648785996613e-05,
+      "loss": 2.2358,
       "step": 2230
     },
     {
       "epoch": 0.42160737812911725,
-      "grad_norm": 4.348377227783203,
-      "learning_rate": 1.8313570487483533e-05,
-      "loss": 2.9414,
+      "grad_norm": 6.686824798583984,
+      "learning_rate": 2.7470355731225296e-05,
+      "loss": 2.4671,
       "step": 2240
     },
     {
       "epoch": 0.4234895539243365,
-      "grad_norm": 5.964900493621826,
-      "learning_rate": 1.8306041784302656e-05,
-      "loss": 2.9333,
+      "grad_norm": 5.854467391967773,
+      "learning_rate": 2.745906267645398e-05,
+      "loss": 2.3917,
       "step": 2250
     },
     {
       "epoch": 0.4253717297195558,
-      "grad_norm": 8.116161346435547,
-      "learning_rate": 1.829851308112178e-05,
-      "loss": 2.8507,
+      "grad_norm": 9.136099815368652,
+      "learning_rate": 2.7447769621682664e-05,
+      "loss": 2.2994,
       "step": 2260
     },
     {
       "epoch": 0.4272539055147751,
-      "grad_norm": 5.363483428955078,
-      "learning_rate": 1.82909843779409e-05,
-      "loss": 2.7429,
+      "grad_norm": 4.834329128265381,
+      "learning_rate": 2.743647656691135e-05,
+      "loss": 2.1033,
       "step": 2270
     },
     {
       "epoch": 0.42913608130999437,
-      "grad_norm": 3.6393518447875977,
-      "learning_rate": 1.8283455674760024e-05,
-      "loss": 2.9543,
+      "grad_norm": 5.209042072296143,
+      "learning_rate": 2.7425183512140036e-05,
+      "loss": 2.4016,
       "step": 2280
     },
     {
       "epoch": 0.43101825710521363,
-      "grad_norm": 11.873992919921875,
-      "learning_rate": 1.8275926971579147e-05,
-      "loss": 2.9146,
+      "grad_norm": 11.149210929870605,
+      "learning_rate": 2.741389045736872e-05,
+      "loss": 2.3709,
       "step": 2290
     },
     {
       "epoch": 0.4329004329004329,
-      "grad_norm": 6.46327018737793,
-      "learning_rate": 1.826839826839827e-05,
-      "loss": 2.8802,
+      "grad_norm": 12.44275188446045,
+      "learning_rate": 2.7402597402597405e-05,
+      "loss": 2.3464,
       "step": 2300
     },
     {
       "epoch": 0.43478260869565216,
-      "grad_norm": 4.694522857666016,
-      "learning_rate": 1.8260869565217393e-05,
-      "loss": 2.7743,
+      "grad_norm": 12.738330841064453,
+      "learning_rate": 2.7391304347826085e-05,
+      "loss": 2.1641,
       "step": 2310
     },
     {
       "epoch": 0.4366647844908714,
-      "grad_norm": 5.260149002075195,
-      "learning_rate": 1.8253340862036516e-05,
-      "loss": 2.7807,
+      "grad_norm": 8.074936866760254,
+      "learning_rate": 2.738001129305477e-05,
+      "loss": 2.1561,
       "step": 2320
     },
     {
       "epoch": 0.43854696028609075,
-      "grad_norm": 5.388479709625244,
-      "learning_rate": 1.824581215885564e-05,
-      "loss": 2.7548,
+      "grad_norm": 5.971907615661621,
+      "learning_rate": 2.7368718238283457e-05,
+      "loss": 2.2271,
       "step": 2330
     },
     {
       "epoch": 0.44042913608131,
-      "grad_norm": 4.184902191162109,
-      "learning_rate": 1.8238283455674762e-05,
-      "loss": 2.8732,
+      "grad_norm": 3.5811924934387207,
+      "learning_rate": 2.735742518351214e-05,
+      "loss": 2.2939,
       "step": 2340
     },
     {
       "epoch": 0.4423113118765293,
-      "grad_norm": 6.423655986785889,
-      "learning_rate": 1.8230754752493885e-05,
-      "loss": 2.7849,
+      "grad_norm": 7.316919326782227,
+      "learning_rate": 2.7346132128740825e-05,
+      "loss": 2.2452,
       "step": 2350
     },
     {
       "epoch": 0.44419348767174854,
-      "grad_norm": 8.165117263793945,
-      "learning_rate": 1.8223226049313008e-05,
-      "loss": 2.7968,
+      "grad_norm": 12.880569458007812,
+      "learning_rate": 2.7334839073969508e-05,
+      "loss": 2.2113,
       "step": 2360
     },
     {
       "epoch": 0.4460756634669678,
-      "grad_norm": 4.645898342132568,
-      "learning_rate": 1.821569734613213e-05,
-      "loss": 2.9012,
+      "grad_norm": 7.582205295562744,
+      "learning_rate": 2.7323546019198194e-05,
+      "loss": 2.3516,
       "step": 2370
     },
     {
       "epoch": 0.4479578392621871,
-      "grad_norm": 5.929998397827148,
-      "learning_rate": 1.8208168642951254e-05,
-      "loss": 2.7395,
+      "grad_norm": 5.996733665466309,
+      "learning_rate": 2.731225296442688e-05,
+      "loss": 2.17,
       "step": 2380
     },
     {
       "epoch": 0.44984001505740634,
-      "grad_norm": 3.953486442565918,
-      "learning_rate": 1.8200639939770376e-05,
-      "loss": 2.8225,
+      "grad_norm": 4.608865737915039,
+      "learning_rate": 2.7300959909655563e-05,
+      "loss": 2.3382,
       "step": 2390
     },
     {
       "epoch": 0.45172219085262566,
-      "grad_norm": 6.943857192993164,
-      "learning_rate": 1.81931112365895e-05,
-      "loss": 2.9133,
+      "grad_norm": 8.942387580871582,
+      "learning_rate": 2.7289666854884246e-05,
+      "loss": 2.4042,
       "step": 2400
     },
     {
       "epoch": 0.4536043666478449,
-      "grad_norm": 6.260727405548096,
-      "learning_rate": 1.8185582533408622e-05,
-      "loss": 2.8366,
+      "grad_norm": 5.910847187042236,
+      "learning_rate": 2.7278373800112932e-05,
+      "loss": 2.2939,
       "step": 2410
     },
     {
       "epoch": 0.4554865424430642,
-      "grad_norm": 9.32252025604248,
-      "learning_rate": 1.8178053830227745e-05,
-      "loss": 2.7906,
+      "grad_norm": 7.828692436218262,
+      "learning_rate": 2.7267080745341614e-05,
+      "loss": 2.244,
       "step": 2420
     },
     {
       "epoch": 0.45736871823828346,
-      "grad_norm": 4.764776706695557,
-      "learning_rate": 1.8170525127046868e-05,
-      "loss": 2.7746,
+      "grad_norm": 6.50481653213501,
+      "learning_rate": 2.72557876905703e-05,
+      "loss": 2.2278,
       "step": 2430
     },
     {
       "epoch": 0.4592508940335027,
-      "grad_norm": 5.310938358306885,
-      "learning_rate": 1.816299642386599e-05,
-      "loss": 2.8342,
+      "grad_norm": 6.783820629119873,
+      "learning_rate": 2.7244494635798983e-05,
+      "loss": 2.2561,
       "step": 2440
     },
     {
       "epoch": 0.461133069828722,
-      "grad_norm": 4.960974216461182,
-      "learning_rate": 1.8155467720685114e-05,
-      "loss": 2.8173,
+      "grad_norm": 5.222927570343018,
+      "learning_rate": 2.723320158102767e-05,
+      "loss": 2.1948,
       "step": 2450
     },
     {
       "epoch": 0.46301524562394125,
-      "grad_norm": 10.526418685913086,
-      "learning_rate": 1.8147939017504237e-05,
-      "loss": 2.718,
+      "grad_norm": 8.41578197479248,
+      "learning_rate": 2.7221908526256355e-05,
+      "loss": 2.1365,
       "step": 2460
     },
     {
       "epoch": 0.4648974214191606,
-      "grad_norm": 3.5313024520874023,
-      "learning_rate": 1.814041031432336e-05,
-      "loss": 2.6789,
+      "grad_norm": 7.914490222930908,
+      "learning_rate": 2.7210615471485038e-05,
+      "loss": 2.0912,
       "step": 2470
     },
     {
       "epoch": 0.46677959721437984,
-      "grad_norm": 5.258660793304443,
-      "learning_rate": 1.8132881611142483e-05,
-      "loss": 2.7351,
+      "grad_norm": 10.265892028808594,
+      "learning_rate": 2.7199322416713724e-05,
+      "loss": 2.2027,
       "step": 2480
     },
     {
       "epoch": 0.4686617730095991,
-      "grad_norm": 3.7664971351623535,
-      "learning_rate": 1.8125352907961606e-05,
-      "loss": 2.7529,
+      "grad_norm": 5.603802680969238,
+      "learning_rate": 2.7188029361942403e-05,
+      "loss": 2.1609,
       "step": 2490
     },
     {
       "epoch": 0.47054394880481837,
-      "grad_norm": 9.83526611328125,
-      "learning_rate": 1.811782420478073e-05,
-      "loss": 2.7291,
+      "grad_norm": 11.390222549438477,
+      "learning_rate": 2.717673630717109e-05,
+      "loss": 2.1755,
       "step": 2500
     },
     {
       "epoch": 0.47242612460003763,
-      "grad_norm": 5.410508155822754,
-      "learning_rate": 1.811029550159985e-05,
-      "loss": 2.5376,
+      "grad_norm": 4.981803894042969,
+      "learning_rate": 2.7165443252399776e-05,
+      "loss": 1.873,
       "step": 2510
     },
     {
       "epoch": 0.4743083003952569,
-      "grad_norm": 6.245204925537109,
-      "learning_rate": 1.8102766798418974e-05,
-      "loss": 2.7001,
+      "grad_norm": 6.110774040222168,
+      "learning_rate": 2.7154150197628458e-05,
+      "loss": 2.0585,
       "step": 2520
     },
     {
       "epoch": 0.47619047619047616,
-      "grad_norm": 10.058247566223145,
-      "learning_rate": 1.8095238095238097e-05,
-      "loss": 2.5141,
+      "grad_norm": 6.286103248596191,
+      "learning_rate": 2.7142857142857144e-05,
+      "loss": 1.9113,
       "step": 2530
     },
     {
       "epoch": 0.4780726519856955,
-      "grad_norm": 3.719804286956787,
-      "learning_rate": 1.808770939205722e-05,
-      "loss": 2.6479,
+      "grad_norm": 5.199717044830322,
+      "learning_rate": 2.7131564088085827e-05,
+      "loss": 2.0643,
       "step": 2540
     },
     {
       "epoch": 0.47995482778091475,
-      "grad_norm": 6.557003498077393,
-      "learning_rate": 1.8080180688876343e-05,
-      "loss": 2.6833,
+      "grad_norm": 10.684744834899902,
+      "learning_rate": 2.7120271033314513e-05,
+      "loss": 2.1255,
       "step": 2550
     },
     {
       "epoch": 0.481837003576134,
-      "grad_norm": 4.292045593261719,
-      "learning_rate": 1.8072651985695466e-05,
-      "loss": 2.6688,
+      "grad_norm": 6.335547924041748,
+      "learning_rate": 2.71089779785432e-05,
+      "loss": 2.0588,
       "step": 2560
     },
     {
       "epoch": 0.4837191793713533,
-      "grad_norm": 5.613296031951904,
-      "learning_rate": 1.8065123282514586e-05,
-      "loss": 2.7906,
+      "grad_norm": 15.644701957702637,
+      "learning_rate": 2.709768492377188e-05,
+      "loss": 2.2138,
       "step": 2570
     },
     {
       "epoch": 0.48560135516657255,
-      "grad_norm": 10.547001838684082,
-      "learning_rate": 1.8057594579333712e-05,
-      "loss": 2.5276,
+      "grad_norm": 8.629146575927734,
+      "learning_rate": 2.7086391869000565e-05,
+      "loss": 1.8363,
       "step": 2580
     },
     {
       "epoch": 0.4874835309617918,
-      "grad_norm": 4.5769782066345215,
-      "learning_rate": 1.8050065876152835e-05,
-      "loss": 2.6816,
+      "grad_norm": 6.585087299346924,
+      "learning_rate": 2.7075098814229247e-05,
+      "loss": 2.096,
       "step": 2590
     },
     {
       "epoch": 0.48936570675701113,
-      "grad_norm": 6.050182342529297,
-      "learning_rate": 1.8042537172971958e-05,
-      "loss": 2.6342,
+      "grad_norm": 4.812081336975098,
+      "learning_rate": 2.7063805759457933e-05,
+      "loss": 2.0322,
       "step": 2600
     },
     {
       "epoch": 0.4912478825522304,
-      "grad_norm": 3.172720432281494,
-      "learning_rate": 1.803500846979108e-05,
-      "loss": 2.7868,
+      "grad_norm": 3.4937989711761475,
+      "learning_rate": 2.705251270468662e-05,
+      "loss": 2.2819,
       "step": 2610
     },
     {
       "epoch": 0.49313005834744966,
-      "grad_norm": 7.896243095397949,
-      "learning_rate": 1.8027479766610204e-05,
-      "loss": 2.6741,
+      "grad_norm": 13.015551567077637,
+      "learning_rate": 2.7041219649915302e-05,
+      "loss": 2.0931,
       "step": 2620
     },
     {
       "epoch": 0.4950122341426689,
-      "grad_norm": 6.55963659286499,
-      "learning_rate": 1.8019951063429327e-05,
-      "loss": 2.6136,
+      "grad_norm": 26.068567276000977,
+      "learning_rate": 2.7029926595143988e-05,
+      "loss": 2.1267,
       "step": 2630
     },
     {
       "epoch": 0.4968944099378882,
-      "grad_norm": 4.171892166137695,
-      "learning_rate": 1.801242236024845e-05,
-      "loss": 2.5573,
+      "grad_norm": 6.577176094055176,
+      "learning_rate": 2.701863354037267e-05,
+      "loss": 1.9937,
       "step": 2640
     },
     {
       "epoch": 0.49877658573310746,
-      "grad_norm": 8.071471214294434,
-      "learning_rate": 1.8004893657067572e-05,
-      "loss": 2.7103,
+      "grad_norm": 8.388608932495117,
+      "learning_rate": 2.7007340485601357e-05,
+      "loss": 2.1152,
       "step": 2650
     },
     {
       "epoch": 0.5006587615283268,
-      "grad_norm": 5.946764945983887,
-      "learning_rate": 1.7997364953886692e-05,
-      "loss": 2.6454,
+      "grad_norm": 5.307988166809082,
+      "learning_rate": 2.699604743083004e-05,
+      "loss": 2.0221,
       "step": 2660
     },
     {
       "epoch": 0.502540937323546,
-      "grad_norm": 7.082939624786377,
-      "learning_rate": 1.7989836250705818e-05,
-      "loss": 2.6345,
+      "grad_norm": 7.120198726654053,
+      "learning_rate": 2.6984754376058722e-05,
+      "loss": 2.0487,
       "step": 2670
     },
     {
       "epoch": 0.5044231131187653,
-      "grad_norm": 5.81397008895874,
-      "learning_rate": 1.798230754752494e-05,
-      "loss": 2.6742,
+      "grad_norm": 5.206243991851807,
+      "learning_rate": 2.697346132128741e-05,
+      "loss": 2.112,
       "step": 2680
     },
     {
       "epoch": 0.5063052889139845,
-      "grad_norm": 11.386290550231934,
-      "learning_rate": 1.7974778844344064e-05,
-      "loss": 2.6589,
+      "grad_norm": 12.726581573486328,
+      "learning_rate": 2.6962168266516094e-05,
+      "loss": 2.1619,
       "step": 2690
     },
     {
       "epoch": 0.5081874647092038,
-      "grad_norm": 8.060523986816406,
-      "learning_rate": 1.7967250141163187e-05,
-      "loss": 2.4929,
+      "grad_norm": 5.604727745056152,
+      "learning_rate": 2.6950875211744777e-05,
+      "loss": 1.8609,
       "step": 2700
     },
     {
       "epoch": 0.5100696405044232,
-      "grad_norm": 4.004101753234863,
-      "learning_rate": 1.795972143798231e-05,
-      "loss": 2.6427,
+      "grad_norm": 4.6965012550354,
+      "learning_rate": 2.6939582156973463e-05,
+      "loss": 2.095,
       "step": 2710
     },
     {
       "epoch": 0.5119518162996424,
-      "grad_norm": 4.872875213623047,
-      "learning_rate": 1.7952192734801433e-05,
-      "loss": 2.6538,
+      "grad_norm": 6.7473602294921875,
+      "learning_rate": 2.6928289102202146e-05,
+      "loss": 2.0439,
       "step": 2720
     },
     {
       "epoch": 0.5138339920948617,
-      "grad_norm": 5.859134197235107,
-      "learning_rate": 1.7944664031620556e-05,
-      "loss": 2.736,
+      "grad_norm": 9.199115753173828,
+      "learning_rate": 2.6916996047430832e-05,
+      "loss": 2.1664,
       "step": 2730
     },
     {
       "epoch": 0.5157161678900809,
-      "grad_norm": 12.146455764770508,
-      "learning_rate": 1.793713532843968e-05,
-      "loss": 2.5728,
+      "grad_norm": 8.569664001464844,
+      "learning_rate": 2.6905702992659518e-05,
+      "loss": 1.9945,
       "step": 2740
     },
     {
       "epoch": 0.5175983436853002,
-      "grad_norm": 7.305758476257324,
-      "learning_rate": 1.7929606625258798e-05,
-      "loss": 2.5778,
+      "grad_norm": 13.89288330078125,
+      "learning_rate": 2.6894409937888197e-05,
+      "loss": 1.9975,
       "step": 2750
     },
     {
       "epoch": 0.5194805194805194,
-      "grad_norm": 3.7814443111419678,
-      "learning_rate": 1.792207792207792e-05,
-      "loss": 2.5947,
+      "grad_norm": 3.2873876094818115,
+      "learning_rate": 2.6883116883116883e-05,
+      "loss": 2.0202,
       "step": 2760
     },
     {
       "epoch": 0.5213626952757388,
-      "grad_norm": 3.1260640621185303,
-      "learning_rate": 1.7914549218897047e-05,
-      "loss": 2.6592,
+      "grad_norm": 3.6722116470336914,
+      "learning_rate": 2.6871823828345566e-05,
+      "loss": 2.1103,
       "step": 2770
     },
     {
       "epoch": 0.5232448710709581,
-      "grad_norm": 5.482910633087158,
-      "learning_rate": 1.790702051571617e-05,
-      "loss": 2.546,
+      "grad_norm": 6.789108753204346,
+      "learning_rate": 2.6860530773574252e-05,
+      "loss": 2.0607,
       "step": 2780
     },
     {
       "epoch": 0.5251270468661773,
-      "grad_norm": 4.761525630950928,
-      "learning_rate": 1.7899491812535293e-05,
-      "loss": 2.5239,
+      "grad_norm": 9.947601318359375,
+      "learning_rate": 2.6849237718802938e-05,
+      "loss": 1.8943,
       "step": 2790
     },
     {
       "epoch": 0.5270092226613966,
-      "grad_norm": 10.941164016723633,
-      "learning_rate": 1.7891963109354416e-05,
-      "loss": 2.5527,
+      "grad_norm": 29.515361785888672,
+      "learning_rate": 2.683794466403162e-05,
+      "loss": 1.9724,
       "step": 2800
     },
     {
       "epoch": 0.5288913984566158,
-      "grad_norm": 8.885186195373535,
-      "learning_rate": 1.788443440617354e-05,
-      "loss": 2.5902,
+      "grad_norm": 14.963418960571289,
+      "learning_rate": 2.6826651609260307e-05,
+      "loss": 2.0618,
       "step": 2810
     },
     {
       "epoch": 0.5307735742518351,
-      "grad_norm": 6.169194221496582,
-      "learning_rate": 1.7876905702992662e-05,
-      "loss": 2.6758,
+      "grad_norm": 16.72760772705078,
+      "learning_rate": 2.681535855448899e-05,
+      "loss": 2.1478,
       "step": 2820
     },
     {
       "epoch": 0.5326557500470543,
-      "grad_norm": 11.396913528442383,
-      "learning_rate": 1.7869376999811785e-05,
-      "loss": 2.6403,
+      "grad_norm": 11.509770393371582,
+      "learning_rate": 2.6804065499717676e-05,
+      "loss": 2.0953,
       "step": 2830
     },
     {
       "epoch": 0.5345379258422737,
-      "grad_norm": 6.287395477294922,
-      "learning_rate": 1.7861848296630905e-05,
-      "loss": 2.5786,
+      "grad_norm": 13.363036155700684,
+      "learning_rate": 2.679277244494636e-05,
+      "loss": 2.0589,
       "step": 2840
     },
     {
       "epoch": 0.536420101637493,
-      "grad_norm": 4.667734146118164,
-      "learning_rate": 1.7854319593450027e-05,
-      "loss": 2.4285,
+      "grad_norm": 4.475217342376709,
+      "learning_rate": 2.678147939017504e-05,
+      "loss": 1.8262,
       "step": 2850
     },
     {
       "epoch": 0.5383022774327122,
-      "grad_norm": 7.646928787231445,
-      "learning_rate": 1.7846790890269154e-05,
-      "loss": 2.3599,
+      "grad_norm": 3.634226083755493,
+      "learning_rate": 2.6770186335403727e-05,
+      "loss": 1.7898,
       "step": 2860
     },
     {
       "epoch": 0.5401844532279315,
-      "grad_norm": 7.949150085449219,
-      "learning_rate": 1.7839262187088277e-05,
-      "loss": 2.4367,
+      "grad_norm": 22.259286880493164,
+      "learning_rate": 2.675889328063241e-05,
+      "loss": 1.8528,
       "step": 2870
     },
     {
       "epoch": 0.5420666290231507,
-      "grad_norm": 23.867076873779297,
-      "learning_rate": 1.78317334839074e-05,
-      "loss": 2.3823,
+      "grad_norm": 11.183906555175781,
+      "learning_rate": 2.6747600225861096e-05,
+      "loss": 1.7605,
       "step": 2880
     },
     {
       "epoch": 0.54394880481837,
-      "grad_norm": 5.78860330581665,
-      "learning_rate": 1.7824204780726523e-05,
-      "loss": 2.5005,
+      "grad_norm": 6.606888294219971,
+      "learning_rate": 2.6736307171089782e-05,
+      "loss": 1.8818,
       "step": 2890
     },
     {
       "epoch": 0.5458309806135893,
-      "grad_norm": 10.566971778869629,
-      "learning_rate": 1.7816676077545645e-05,
-      "loss": 2.4836,
+      "grad_norm": 8.689189910888672,
+      "learning_rate": 2.6725014116318465e-05,
+      "loss": 1.9096,
       "step": 2900
     },
     {
       "epoch": 0.5477131564088086,
-      "grad_norm": 4.010104656219482,
-      "learning_rate": 1.780914737436477e-05,
-      "loss": 2.3889,
+      "grad_norm": 6.653372287750244,
+      "learning_rate": 2.671372106154715e-05,
+      "loss": 1.7785,
       "step": 2910
     },
     {
       "epoch": 0.5495953322040279,
-      "grad_norm": 10.083136558532715,
-      "learning_rate": 1.7801618671183888e-05,
-      "loss": 2.4056,
+      "grad_norm": 11.017134666442871,
+      "learning_rate": 2.6702428006775834e-05,
+      "loss": 1.8423,
       "step": 2920
     },
     {
       "epoch": 0.5514775079992471,
-      "grad_norm": 4.860154151916504,
-      "learning_rate": 1.779408996800301e-05,
-      "loss": 2.5066,
+      "grad_norm": 6.855719089508057,
+      "learning_rate": 2.6691134952004516e-05,
+      "loss": 1.8001,
       "step": 2930
     },
     {
       "epoch": 0.5533596837944664,
-      "grad_norm": 6.0354905128479,
-      "learning_rate": 1.7786561264822134e-05,
-      "loss": 2.5239,
+      "grad_norm": 5.900759220123291,
+      "learning_rate": 2.6679841897233202e-05,
+      "loss": 1.987,
       "step": 2940
     },
     {
       "epoch": 0.5552418595896856,
-      "grad_norm": 5.251504898071289,
-      "learning_rate": 1.777903256164126e-05,
-      "loss": 2.4828,
+      "grad_norm": 5.474452495574951,
+      "learning_rate": 2.6668548842461885e-05,
+      "loss": 1.9144,
       "step": 2950
     },
     {
       "epoch": 0.557124035384905,
-      "grad_norm": 10.071287155151367,
-      "learning_rate": 1.7771503858460383e-05,
-      "loss": 2.6557,
+      "grad_norm": 13.115139961242676,
+      "learning_rate": 2.665725578769057e-05,
+      "loss": 2.0743,
       "step": 2960
     },
     {
       "epoch": 0.5590062111801242,
-      "grad_norm": 22.186641693115234,
-      "learning_rate": 1.7763975155279506e-05,
-      "loss": 2.4674,
+      "grad_norm": 8.339238166809082,
+      "learning_rate": 2.6645962732919257e-05,
+      "loss": 1.9056,
       "step": 2970
     },
     {
       "epoch": 0.5608883869753435,
-      "grad_norm": 7.2353291511535645,
-      "learning_rate": 1.775644645209863e-05,
-      "loss": 2.4926,
+      "grad_norm": 12.62999153137207,
+      "learning_rate": 2.663466967814794e-05,
+      "loss": 1.8391,
       "step": 2980
     },
     {
       "epoch": 0.5627705627705628,
-      "grad_norm": 7.9539794921875,
-      "learning_rate": 1.7748917748917752e-05,
-      "loss": 2.5513,
+      "grad_norm": 5.597290515899658,
+      "learning_rate": 2.6623376623376626e-05,
+      "loss": 1.9349,
       "step": 2990
     },
     {
       "epoch": 0.564652738565782,
-      "grad_norm": 2.906517267227173,
-      "learning_rate": 1.7741389045736875e-05,
-      "loss": 2.5281,
+      "grad_norm": 2.5785648822784424,
+      "learning_rate": 2.661208356860531e-05,
+      "loss": 1.9463,
       "step": 3000
     },
     {
       "epoch": 0.5665349143610013,
-      "grad_norm": 6.161061763763428,
-      "learning_rate": 1.7733860342555994e-05,
-      "loss": 2.3144,
+      "grad_norm": 13.423247337341309,
+      "learning_rate": 2.660079051383399e-05,
+      "loss": 1.7291,
       "step": 3010
     },
     {
       "epoch": 0.5684170901562206,
-      "grad_norm": 6.098898410797119,
-      "learning_rate": 1.7726331639375117e-05,
-      "loss": 2.4401,
+      "grad_norm": 7.054490089416504,
+      "learning_rate": 2.6589497459062677e-05,
+      "loss": 1.8435,
       "step": 3020
     },
     {
       "epoch": 0.5702992659514399,
-      "grad_norm": 6.252312183380127,
-      "learning_rate": 1.771880293619424e-05,
-      "loss": 2.3486,
+      "grad_norm": 7.4367523193359375,
+      "learning_rate": 2.657820440429136e-05,
+      "loss": 1.7105,
       "step": 3030
     },
     {
       "epoch": 0.5721814417466592,
-      "grad_norm": 5.018267631530762,
-      "learning_rate": 1.7711274233013366e-05,
-      "loss": 2.5058,
+      "grad_norm": 9.216171264648438,
+      "learning_rate": 2.6566911349520046e-05,
+      "loss": 1.9434,
       "step": 3040
     },
     {
       "epoch": 0.5740636175418784,
-      "grad_norm": 4.869485378265381,
-      "learning_rate": 1.770374552983249e-05,
-      "loss": 2.4141,
+      "grad_norm": 7.6077165603637695,
+      "learning_rate": 2.655561829474873e-05,
+      "loss": 1.8798,
       "step": 3050
     },
     {
       "epoch": 0.5759457933370977,
-      "grad_norm": 9.376824378967285,
-      "learning_rate": 1.7696216826651612e-05,
-      "loss": 2.3761,
+      "grad_norm": 11.647812843322754,
+      "learning_rate": 2.6544325239977415e-05,
+      "loss": 1.7522,
       "step": 3060
     },
     {
       "epoch": 0.5778279691323169,
-      "grad_norm": 9.775004386901855,
-      "learning_rate": 1.7688688123470735e-05,
-      "loss": 2.6541,
+      "grad_norm": 5.945573806762695,
+      "learning_rate": 2.65330321852061e-05,
+      "loss": 2.0416,
       "step": 3070
     },
     {
       "epoch": 0.5797101449275363,
-      "grad_norm": 5.287901401519775,
-      "learning_rate": 1.7681159420289858e-05,
-      "loss": 2.3716,
+      "grad_norm": 24.724990844726562,
+      "learning_rate": 2.6521739130434784e-05,
+      "loss": 1.8017,
       "step": 3080
     },
     {
       "epoch": 0.5815923207227555,
-      "grad_norm": 4.416796684265137,
-      "learning_rate": 1.767363071710898e-05,
-      "loss": 2.2715,
+      "grad_norm": 4.811154842376709,
+      "learning_rate": 2.651044607566347e-05,
+      "loss": 1.6813,
       "step": 3090
     },
     {
       "epoch": 0.5834744965179748,
-      "grad_norm": 7.947756767272949,
-      "learning_rate": 1.76661020139281e-05,
-      "loss": 2.3447,
+      "grad_norm": 11.191183090209961,
+      "learning_rate": 2.649915302089215e-05,
+      "loss": 1.7622,
       "step": 3100
     },
     {
       "epoch": 0.5853566723131941,
-      "grad_norm": 8.54676342010498,
-      "learning_rate": 1.7658573310747223e-05,
-      "loss": 2.4445,
+      "grad_norm": 12.16427230834961,
+      "learning_rate": 2.6487859966120835e-05,
+      "loss": 1.9663,
       "step": 3110
     },
     {
       "epoch": 0.5872388481084133,
-      "grad_norm": 9.31558609008789,
-      "learning_rate": 1.7651044607566346e-05,
-      "loss": 2.3065,
+      "grad_norm": 8.902604103088379,
+      "learning_rate": 2.647656691134952e-05,
+      "loss": 1.7131,
       "step": 3120
     },
     {
       "epoch": 0.5891210239036326,
-      "grad_norm": 6.117997646331787,
-      "learning_rate": 1.764351590438547e-05,
-      "loss": 2.3316,
+      "grad_norm": 7.6382904052734375,
+      "learning_rate": 2.6465273856578204e-05,
+      "loss": 1.7472,
       "step": 3130
     },
     {
       "epoch": 0.5910031996988518,
-      "grad_norm": 6.215322017669678,
-      "learning_rate": 1.7635987201204596e-05,
-      "loss": 2.286,
+      "grad_norm": 23.864648818969727,
+      "learning_rate": 2.645398080180689e-05,
+      "loss": 1.679,
       "step": 3140
     },
     {
       "epoch": 0.5928853754940712,
-      "grad_norm": 7.536943435668945,
-      "learning_rate": 1.762845849802372e-05,
-      "loss": 2.6575,
+      "grad_norm": 18.402881622314453,
+      "learning_rate": 2.6442687747035573e-05,
+      "loss": 2.1442,
       "step": 3150
     },
     {
       "epoch": 0.5947675512892904,
-      "grad_norm": 9.97204875946045,
-      "learning_rate": 1.762092979484284e-05,
-      "loss": 2.2224,
+      "grad_norm": 6.911153793334961,
+      "learning_rate": 2.643139469226426e-05,
+      "loss": 1.6756,
       "step": 3160
     },
     {
       "epoch": 0.5966497270845097,
-      "grad_norm": 11.99947452545166,
-      "learning_rate": 1.7613401091661964e-05,
-      "loss": 2.5265,
+      "grad_norm": 12.472163200378418,
+      "learning_rate": 2.6420101637492945e-05,
+      "loss": 1.9381,
       "step": 3170
     },
     {
       "epoch": 0.598531902879729,
-      "grad_norm": 5.427422046661377,
-      "learning_rate": 1.7605872388481084e-05,
-      "loss": 2.2315,
+      "grad_norm": 4.999609470367432,
+      "learning_rate": 2.6408808582721624e-05,
+      "loss": 1.5681,
       "step": 3180
     },
     {
       "epoch": 0.6004140786749482,
-      "grad_norm": 4.574037551879883,
-      "learning_rate": 1.7598343685300207e-05,
-      "loss": 2.372,
+      "grad_norm": 5.08629846572876,
+      "learning_rate": 2.639751552795031e-05,
+      "loss": 1.8375,
       "step": 3190
     },
     {
       "epoch": 0.6022962544701675,
-      "grad_norm": 7.115758419036865,
-      "learning_rate": 1.759081498211933e-05,
-      "loss": 2.5758,
+      "grad_norm": 5.805210590362549,
+      "learning_rate": 2.6386222473178996e-05,
+      "loss": 2.0463,
       "step": 3200
     },
     {
       "epoch": 0.6041784302653868,
-      "grad_norm": 8.453811645507812,
-      "learning_rate": 1.7583286278938453e-05,
-      "loss": 2.1461,
+      "grad_norm": 7.146366119384766,
+      "learning_rate": 2.637492941840768e-05,
+      "loss": 1.5877,
       "step": 3210
     },
     {
       "epoch": 0.6060606060606061,
-      "grad_norm": 8.490642547607422,
-      "learning_rate": 1.7575757575757576e-05,
-      "loss": 2.3582,
+      "grad_norm": 4.034036159515381,
+      "learning_rate": 2.6363636363636365e-05,
+      "loss": 1.835,
       "step": 3220
     },
     {
       "epoch": 0.6079427818558253,
-      "grad_norm": 19.479263305664062,
-      "learning_rate": 1.7568228872576702e-05,
-      "loss": 2.2744,
+      "grad_norm": 22.046175003051758,
+      "learning_rate": 2.6352343308865048e-05,
+      "loss": 1.702,
       "step": 3230
     },
     {
       "epoch": 0.6098249576510446,
-      "grad_norm": 14.469964981079102,
-      "learning_rate": 1.7560700169395825e-05,
-      "loss": 2.4447,
+      "grad_norm": 20.46267318725586,
+      "learning_rate": 2.6341050254093734e-05,
+      "loss": 1.8594,
       "step": 3240
     },
     {
       "epoch": 0.6117071334462639,
-      "grad_norm": 3.8202433586120605,
-      "learning_rate": 1.7553171466214948e-05,
-      "loss": 2.367,
+      "grad_norm": 6.352889537811279,
+      "learning_rate": 2.632975719932242e-05,
+      "loss": 1.78,
       "step": 3250
     },
     {
       "epoch": 0.6135893092414831,
-      "grad_norm": 5.609041213989258,
-      "learning_rate": 1.754564276303407e-05,
-      "loss": 2.4837,
+      "grad_norm": 5.597209453582764,
+      "learning_rate": 2.6318464144551103e-05,
+      "loss": 1.9062,
       "step": 3260
     },
     {
       "epoch": 0.6154714850367025,
-      "grad_norm": 4.065234661102295,
-      "learning_rate": 1.753811405985319e-05,
-      "loss": 2.3212,
+      "grad_norm": 8.404434204101562,
+      "learning_rate": 2.6307171089779785e-05,
+      "loss": 1.7729,
       "step": 3270
     },
     {
       "epoch": 0.6173536608319217,
-      "grad_norm": 8.558253288269043,
-      "learning_rate": 1.7530585356672313e-05,
-      "loss": 2.2335,
+      "grad_norm": 9.512775421142578,
+      "learning_rate": 2.6295878035008468e-05,
+      "loss": 1.6567,
       "step": 3280
     },
     {
       "epoch": 0.619235836627141,
-      "grad_norm": 6.554466247558594,
-      "learning_rate": 1.7523056653491436e-05,
-      "loss": 2.1559,
+      "grad_norm": 11.096758842468262,
+      "learning_rate": 2.6284584980237154e-05,
+      "loss": 1.5892,
       "step": 3290
     },
     {
       "epoch": 0.6211180124223602,
-      "grad_norm": 6.196263790130615,
-      "learning_rate": 1.751552795031056e-05,
-      "loss": 2.3493,
+      "grad_norm": 4.303831100463867,
+      "learning_rate": 2.627329192546584e-05,
+      "loss": 1.7157,
       "step": 3300
     },
     {
       "epoch": 0.6230001882175795,
-      "grad_norm": 14.270978927612305,
-      "learning_rate": 1.7507999247129682e-05,
-      "loss": 2.2956,
+      "grad_norm": 3.8192107677459717,
+      "learning_rate": 2.6261998870694523e-05,
+      "loss": 1.7188,
       "step": 3310
     },
     {
       "epoch": 0.6248823640127988,
-      "grad_norm": 22.012874603271484,
-      "learning_rate": 1.7500470543948808e-05,
-      "loss": 2.3193,
+      "grad_norm": 10.297846794128418,
+      "learning_rate": 2.625070581592321e-05,
+      "loss": 1.7473,
       "step": 3320
     },
     {
       "epoch": 0.626764539808018,
-      "grad_norm": 6.940454959869385,
-      "learning_rate": 1.749294184076793e-05,
-      "loss": 2.2292,
+      "grad_norm": 8.126511573791504,
+      "learning_rate": 2.623941276115189e-05,
+      "loss": 1.5694,
       "step": 3330
     },
     {
       "epoch": 0.6286467156032374,
-      "grad_norm": 5.154090881347656,
-      "learning_rate": 1.7485413137587054e-05,
-      "loss": 2.3149,
+      "grad_norm": 4.767712116241455,
+      "learning_rate": 2.6228119706380578e-05,
+      "loss": 1.7124,
       "step": 3340
     },
     {
       "epoch": 0.6305288913984566,
-      "grad_norm": 4.400321006774902,
-      "learning_rate": 1.7477884434406177e-05,
-      "loss": 2.4335,
+      "grad_norm": 4.750823020935059,
+      "learning_rate": 2.6216826651609264e-05,
+      "loss": 1.8946,
       "step": 3350
     },
     {
       "epoch": 0.6324110671936759,
-      "grad_norm": 8.82219123840332,
-      "learning_rate": 1.7470355731225296e-05,
-      "loss": 2.1911,
+      "grad_norm": 9.973583221435547,
+      "learning_rate": 2.6205533596837943e-05,
+      "loss": 1.6632,
       "step": 3360
     },
     {
       "epoch": 0.6342932429888951,
-      "grad_norm": 7.568221569061279,
-      "learning_rate": 1.746282702804442e-05,
-      "loss": 2.3591,
+      "grad_norm": 10.300981521606445,
+      "learning_rate": 2.619424054206663e-05,
+      "loss": 1.8623,
       "step": 3370
     },
     {
       "epoch": 0.6361754187841144,
-      "grad_norm": 13.639214515686035,
-      "learning_rate": 1.7455298324863542e-05,
-      "loss": 2.3107,
+      "grad_norm": 17.23389434814453,
+      "learning_rate": 2.6182947487295312e-05,
+      "loss": 1.8112,
       "step": 3380
     },
     {
       "epoch": 0.6380575945793338,
-      "grad_norm": 4.316806793212891,
-      "learning_rate": 1.7447769621682665e-05,
-      "loss": 2.1789,
+      "grad_norm": 8.108388900756836,
+      "learning_rate": 2.6171654432523998e-05,
+      "loss": 1.6344,
       "step": 3390
     },
     {
       "epoch": 0.639939770374553,
-      "grad_norm": 4.804785251617432,
-      "learning_rate": 1.7440240918501788e-05,
-      "loss": 2.1812,
+      "grad_norm": 3.373798131942749,
+      "learning_rate": 2.6160361377752684e-05,
+      "loss": 1.5514,
       "step": 3400
     },
     {
       "epoch": 0.6418219461697723,
-      "grad_norm": 8.034902572631836,
-      "learning_rate": 1.743271221532091e-05,
-      "loss": 2.2914,
+      "grad_norm": 8.216525077819824,
+      "learning_rate": 2.6149068322981367e-05,
+      "loss": 1.727,
       "step": 3410
     },
     {
       "epoch": 0.6437041219649915,
-      "grad_norm": 19.507585525512695,
-      "learning_rate": 1.7425183512140037e-05,
-      "loss": 2.3718,
+      "grad_norm": 9.52584457397461,
+      "learning_rate": 2.6137775268210053e-05,
+      "loss": 1.8575,
       "step": 3420
     },
     {
       "epoch": 0.6455862977602108,
-      "grad_norm": 13.059826850891113,
-      "learning_rate": 1.741765480895916e-05,
-      "loss": 2.2965,
+      "grad_norm": 7.7391581535339355,
+      "learning_rate": 2.612648221343874e-05,
+      "loss": 1.7685,
       "step": 3430
     },
     {
       "epoch": 0.64746847355543,
-      "grad_norm": 5.1344499588012695,
-      "learning_rate": 1.7410126105778283e-05,
-      "loss": 2.2073,
+      "grad_norm": 8.019952774047852,
+      "learning_rate": 2.611518915866742e-05,
+      "loss": 1.6028,
       "step": 3440
     },
     {
       "epoch": 0.6493506493506493,
-      "grad_norm": 7.513482570648193,
-      "learning_rate": 1.7402597402597403e-05,
-      "loss": 2.4255,
+      "grad_norm": 11.445884704589844,
+      "learning_rate": 2.6103896103896104e-05,
+      "loss": 1.9301,
       "step": 3450
     },
     {
       "epoch": 0.6512328251458687,
-      "grad_norm": 4.743190288543701,
-      "learning_rate": 1.7395068699416526e-05,
-      "loss": 2.1949,
+      "grad_norm": 8.298888206481934,
+      "learning_rate": 2.6092603049124787e-05,
+      "loss": 1.5639,
       "step": 3460
     },
     {
       "epoch": 0.6531150009410879,
-      "grad_norm": 7.261030197143555,
-      "learning_rate": 1.738753999623565e-05,
-      "loss": 2.4901,
+      "grad_norm": 6.707033157348633,
+      "learning_rate": 2.6081309994353473e-05,
+      "loss": 1.9865,
       "step": 3470
     },
     {
       "epoch": 0.6549971767363072,
-      "grad_norm": 5.6540093421936035,
-      "learning_rate": 1.738001129305477e-05,
-      "loss": 2.3915,
+      "grad_norm": 13.604408264160156,
+      "learning_rate": 2.607001693958216e-05,
+      "loss": 1.8029,
       "step": 3480
     },
     {
       "epoch": 0.6568793525315264,
-      "grad_norm": 6.778534889221191,
-      "learning_rate": 1.7372482589873894e-05,
-      "loss": 2.2246,
+      "grad_norm": 13.318224906921387,
+      "learning_rate": 2.605872388481084e-05,
+      "loss": 1.6995,
       "step": 3490
     },
     {
       "epoch": 0.6587615283267457,
-      "grad_norm": 7.759307384490967,
-      "learning_rate": 1.7364953886693017e-05,
-      "loss": 2.0732,
+      "grad_norm": 10.210710525512695,
+      "learning_rate": 2.6047430830039528e-05,
+      "loss": 1.4445,
       "step": 3500
     },
     {
       "epoch": 0.6606437041219649,
-      "grad_norm": 16.40176010131836,
-      "learning_rate": 1.7357425183512144e-05,
-      "loss": 2.2766,
+      "grad_norm": 23.614696502685547,
+      "learning_rate": 2.603613777526821e-05,
+      "loss": 1.7893,
       "step": 3510
     },
     {
       "epoch": 0.6625258799171843,
-      "grad_norm": 7.301122188568115,
-      "learning_rate": 1.7349896480331267e-05,
-      "loss": 2.1197,
+      "grad_norm": 6.86085844039917,
+      "learning_rate": 2.6024844720496896e-05,
+      "loss": 1.5184,
       "step": 3520
     },
     {
       "epoch": 0.6644080557124036,
-      "grad_norm": 5.543787479400635,
-      "learning_rate": 1.7342367777150386e-05,
-      "loss": 2.2689,
+      "grad_norm": 7.119161128997803,
+      "learning_rate": 2.601355166572558e-05,
+      "loss": 1.6189,
       "step": 3530
     },
     {
       "epoch": 0.6662902315076228,
-      "grad_norm": 4.671676158905029,
-      "learning_rate": 1.733483907396951e-05,
-      "loss": 2.2419,
+      "grad_norm": 7.726982593536377,
+      "learning_rate": 2.6002258610954262e-05,
+      "loss": 1.7796,
       "step": 3540
     },
     {
       "epoch": 0.6681724073028421,
-      "grad_norm": 6.969106197357178,
-      "learning_rate": 1.7327310370788632e-05,
-      "loss": 2.127,
+      "grad_norm": 7.957070350646973,
+      "learning_rate": 2.5990965556182948e-05,
+      "loss": 1.5165,
       "step": 3550
     },
     {
       "epoch": 0.6700545830980613,
-      "grad_norm": 7.427960395812988,
-      "learning_rate": 1.7319781667607755e-05,
-      "loss": 2.3973,
+      "grad_norm": 11.317330360412598,
+      "learning_rate": 2.597967250141163e-05,
+      "loss": 1.835,
       "step": 3560
     },
     {
       "epoch": 0.6719367588932806,
-      "grad_norm": 4.417695999145508,
-      "learning_rate": 1.7312252964426878e-05,
-      "loss": 2.0037,
+      "grad_norm": 7.407431602478027,
+      "learning_rate": 2.5968379446640317e-05,
+      "loss": 1.3834,
       "step": 3570
     },
     {
       "epoch": 0.6738189346885,
-      "grad_norm": 6.873748779296875,
-      "learning_rate": 1.7304724261246e-05,
-      "loss": 2.0635,
+      "grad_norm": 9.898704528808594,
+      "learning_rate": 2.5957086391869003e-05,
+      "loss": 1.432,
       "step": 3580
     },
     {
       "epoch": 0.6757011104837192,
-      "grad_norm": 4.036855220794678,
-      "learning_rate": 1.7297195558065124e-05,
-      "loss": 2.2912,
+      "grad_norm": 5.612309455871582,
+      "learning_rate": 2.5945793337097685e-05,
+      "loss": 1.7256,
       "step": 3590
     },
     {
       "epoch": 0.6775832862789385,
-      "grad_norm": 6.932463645935059,
-      "learning_rate": 1.728966685488425e-05,
-      "loss": 2.1217,
+      "grad_norm": 5.5622663497924805,
+      "learning_rate": 2.593450028232637e-05,
+      "loss": 1.5599,
       "step": 3600
     },
     {
       "epoch": 0.6794654620741577,
-      "grad_norm": 15.238147735595703,
-      "learning_rate": 1.7282138151703373e-05,
-      "loss": 2.2328,
+      "grad_norm": 5.885371685028076,
+      "learning_rate": 2.5923207227555054e-05,
+      "loss": 1.7386,
       "step": 3610
     },
     {
       "epoch": 0.681347637869377,
-      "grad_norm": 4.133694648742676,
-      "learning_rate": 1.7274609448522492e-05,
-      "loss": 2.0326,
+      "grad_norm": 3.2068424224853516,
+      "learning_rate": 2.5911914172783737e-05,
+      "loss": 1.5411,
       "step": 3620
     },
     {
       "epoch": 0.6832298136645962,
-      "grad_norm": 5.9729766845703125,
-      "learning_rate": 1.7267080745341615e-05,
-      "loss": 2.2155,
+      "grad_norm": 3.0026233196258545,
+      "learning_rate": 2.5900621118012423e-05,
+      "loss": 1.7239,
       "step": 3630
     },
     {
       "epoch": 0.6851119894598156,
-      "grad_norm": 13.098869323730469,
-      "learning_rate": 1.7259552042160738e-05,
-      "loss": 2.267,
+      "grad_norm": 23.438610076904297,
+      "learning_rate": 2.5889328063241106e-05,
+      "loss": 1.7582,
       "step": 3640
     },
     {
       "epoch": 0.6869941652550349,
-      "grad_norm": 3.980104684829712,
-      "learning_rate": 1.725202333897986e-05,
-      "loss": 2.0865,
+      "grad_norm": 3.962045907974243,
+      "learning_rate": 2.5878035008469792e-05,
+      "loss": 1.4579,
       "step": 3650
     },
     {
       "epoch": 0.6888763410502541,
-      "grad_norm": 3.8054277896881104,
-      "learning_rate": 1.7244494635798984e-05,
-      "loss": 2.079,
+      "grad_norm": 3.3566129207611084,
+      "learning_rate": 2.5866741953698478e-05,
+      "loss": 1.4641,
       "step": 3660
     },
     {
       "epoch": 0.6907585168454734,
-      "grad_norm": 6.999955654144287,
-      "learning_rate": 1.7236965932618107e-05,
-      "loss": 2.1672,
+      "grad_norm": 8.942924499511719,
+      "learning_rate": 2.585544889892716e-05,
+      "loss": 1.5699,
       "step": 3670
     },
     {
       "epoch": 0.6926406926406926,
-      "grad_norm": 11.031990051269531,
-      "learning_rate": 1.722943722943723e-05,
-      "loss": 2.1624,
+      "grad_norm": 12.78019905090332,
+      "learning_rate": 2.5844155844155847e-05,
+      "loss": 1.6055,
       "step": 3680
     },
     {
       "epoch": 0.6945228684359119,
-      "grad_norm": 8.7633056640625,
-      "learning_rate": 1.7221908526256356e-05,
-      "loss": 2.227,
+      "grad_norm": 14.28182315826416,
+      "learning_rate": 2.583286278938453e-05,
+      "loss": 1.6354,
       "step": 3690
     },
     {
       "epoch": 0.6964050442311311,
-      "grad_norm": 7.400768280029297,
-      "learning_rate": 1.721437982307548e-05,
-      "loss": 2.1331,
+      "grad_norm": 10.26236629486084,
+      "learning_rate": 2.5821569734613215e-05,
+      "loss": 1.5518,
       "step": 3700
     },
     {
       "epoch": 0.6982872200263505,
-      "grad_norm": 4.83717155456543,
-      "learning_rate": 1.72068511198946e-05,
-      "loss": 2.049,
+      "grad_norm": 6.233335494995117,
+      "learning_rate": 2.5810276679841898e-05,
+      "loss": 1.4589,
       "step": 3710
     },
     {
       "epoch": 0.7001693958215698,
-      "grad_norm": 9.554513931274414,
-      "learning_rate": 1.719932241671372e-05,
-      "loss": 2.1901,
+      "grad_norm": 17.028034210205078,
+      "learning_rate": 2.579898362507058e-05,
+      "loss": 1.6329,
       "step": 3720
     },
     {
       "epoch": 0.702051571616789,
-      "grad_norm": 6.765610694885254,
-      "learning_rate": 1.7191793713532845e-05,
-      "loss": 2.0239,
+      "grad_norm": 9.508376121520996,
+      "learning_rate": 2.5787690570299267e-05,
+      "loss": 1.4574,
       "step": 3730
     },
     {
       "epoch": 0.7039337474120083,
-      "grad_norm": 9.109612464904785,
-      "learning_rate": 1.7184265010351967e-05,
-      "loss": 2.3195,
+      "grad_norm": 10.823552131652832,
+      "learning_rate": 2.577639751552795e-05,
+      "loss": 1.7267,
       "step": 3740
     },
     {
       "epoch": 0.7058159232072275,
-      "grad_norm": 14.373083114624023,
-      "learning_rate": 1.717673630717109e-05,
-      "loss": 2.2486,
+      "grad_norm": 19.552865982055664,
+      "learning_rate": 2.5765104460756636e-05,
+      "loss": 1.7066,
       "step": 3750
     },
     {
       "epoch": 0.7076980990024468,
-      "grad_norm": 8.647749900817871,
-      "learning_rate": 1.7169207603990213e-05,
-      "loss": 2.0991,
+      "grad_norm": 13.044384002685547,
+      "learning_rate": 2.575381140598532e-05,
+      "loss": 1.5771,
       "step": 3760
     },
     {
       "epoch": 0.709580274797666,
-      "grad_norm": 7.751832008361816,
-      "learning_rate": 1.7161678900809336e-05,
-      "loss": 2.0374,
+      "grad_norm": 7.333944797515869,
+      "learning_rate": 2.5742518351214004e-05,
+      "loss": 1.4535,
       "step": 3770
     },
     {
       "epoch": 0.7114624505928854,
-      "grad_norm": 21.32139015197754,
-      "learning_rate": 1.715415019762846e-05,
-      "loss": 2.0808,
+      "grad_norm": 12.022722244262695,
+      "learning_rate": 2.573122529644269e-05,
+      "loss": 1.6377,
       "step": 3780
     },
     {
       "epoch": 0.7133446263881047,
-      "grad_norm": 9.653078079223633,
-      "learning_rate": 1.7146621494447582e-05,
-      "loss": 2.0081,
+      "grad_norm": 6.42018461227417,
+      "learning_rate": 2.571993224167137e-05,
+      "loss": 1.4887,
       "step": 3790
     },
     {
       "epoch": 0.7152268021833239,
-      "grad_norm": 2.955333948135376,
-      "learning_rate": 1.7139092791266705e-05,
-      "loss": 2.2522,
+      "grad_norm": 2.690678596496582,
+      "learning_rate": 2.5708639186900056e-05,
+      "loss": 1.7047,
       "step": 3800
     },
     {
       "epoch": 0.7171089779785432,
-      "grad_norm": 19.1168212890625,
-      "learning_rate": 1.7131564088085828e-05,
-      "loss": 2.2064,
+      "grad_norm": 25.072540283203125,
+      "learning_rate": 2.5697346132128742e-05,
+      "loss": 1.6334,
       "step": 3810
     },
     {
       "epoch": 0.7189911537737624,
-      "grad_norm": 9.711030960083008,
-      "learning_rate": 1.712403538490495e-05,
-      "loss": 1.9549,
+      "grad_norm": 19.594491958618164,
+      "learning_rate": 2.5686053077357425e-05,
+      "loss": 1.3771,
       "step": 3820
     },
     {
       "epoch": 0.7208733295689818,
-      "grad_norm": 15.813447952270508,
-      "learning_rate": 1.7116506681724074e-05,
-      "loss": 2.1709,
+      "grad_norm": 32.11001205444336,
+      "learning_rate": 2.567476002258611e-05,
+      "loss": 1.5945,
       "step": 3830
     },
     {
       "epoch": 0.722755505364201,
-      "grad_norm": 11.953786849975586,
-      "learning_rate": 1.7108977978543197e-05,
-      "loss": 1.9825,
+      "grad_norm": 5.285523891448975,
+      "learning_rate": 2.5663466967814793e-05,
+      "loss": 1.4196,
       "step": 3840
     },
     {
       "epoch": 0.7246376811594203,
-      "grad_norm": 6.5752034187316895,
-      "learning_rate": 1.710144927536232e-05,
-      "loss": 2.1801,
+      "grad_norm": 9.592608451843262,
+      "learning_rate": 2.565217391304348e-05,
+      "loss": 1.6145,
       "step": 3850
     },
     {
       "epoch": 0.7265198569546396,
-      "grad_norm": 10.277215957641602,
-      "learning_rate": 1.7093920572181443e-05,
-      "loss": 2.0362,
+      "grad_norm": 11.879024505615234,
+      "learning_rate": 2.5640880858272165e-05,
+      "loss": 1.6065,
       "step": 3860
     },
     {
       "epoch": 0.7284020327498588,
-      "grad_norm": 12.632518768310547,
-      "learning_rate": 1.7086391869000565e-05,
-      "loss": 2.1342,
+      "grad_norm": 15.967653274536133,
+      "learning_rate": 2.5629587803500848e-05,
+      "loss": 1.716,
       "step": 3870
     },
     {
       "epoch": 0.7302842085450781,
-      "grad_norm": 8.751096725463867,
-      "learning_rate": 1.707886316581969e-05,
-      "loss": 1.9903,
+      "grad_norm": 20.394914627075195,
+      "learning_rate": 2.561829474872953e-05,
+      "loss": 1.4524,
       "step": 3880
     },
     {
       "epoch": 0.7321663843402973,
-      "grad_norm": 6.152645587921143,
-      "learning_rate": 1.707133446263881e-05,
-      "loss": 2.1692,
+      "grad_norm": 8.076727867126465,
+      "learning_rate": 2.5607001693958214e-05,
+      "loss": 1.6307,
       "step": 3890
     },
     {
       "epoch": 0.7340485601355167,
-      "grad_norm": 5.939425945281982,
-      "learning_rate": 1.7063805759457934e-05,
-      "loss": 1.8733,
+      "grad_norm": 15.703822135925293,
+      "learning_rate": 2.55957086391869e-05,
+      "loss": 1.2349,
       "step": 3900
     },
     {
       "epoch": 0.7359307359307359,
-      "grad_norm": 13.272156715393066,
-      "learning_rate": 1.7056277056277057e-05,
-      "loss": 2.1915,
+      "grad_norm": 13.136933326721191,
+      "learning_rate": 2.5584415584415586e-05,
+      "loss": 1.6267,
       "step": 3910
     },
     {
       "epoch": 0.7378129117259552,
-      "grad_norm": 4.635878086090088,
-      "learning_rate": 1.704874835309618e-05,
-      "loss": 2.0686,
+      "grad_norm": 10.711684226989746,
+      "learning_rate": 2.557312252964427e-05,
+      "loss": 1.5033,
       "step": 3920
     },
     {
       "epoch": 0.7396950875211745,
-      "grad_norm": 15.71611499786377,
-      "learning_rate": 1.7041219649915303e-05,
-      "loss": 1.9865,
+      "grad_norm": 12.624408721923828,
+      "learning_rate": 2.5561829474872954e-05,
+      "loss": 1.516,
       "step": 3930
     },
     {
       "epoch": 0.7415772633163937,
-      "grad_norm": 6.550897121429443,
-      "learning_rate": 1.7033690946734426e-05,
-      "loss": 1.9559,
+      "grad_norm": 5.2435479164123535,
+      "learning_rate": 2.555053642010164e-05,
+      "loss": 1.3178,
       "step": 3940
     },
     {
       "epoch": 0.743459439111613,
-      "grad_norm": 8.063326835632324,
-      "learning_rate": 1.702616224355355e-05,
-      "loss": 2.2405,
+      "grad_norm": 12.167356491088867,
+      "learning_rate": 2.5539243365330323e-05,
+      "loss": 1.7724,
       "step": 3950
     },
     {
       "epoch": 0.7453416149068323,
-      "grad_norm": 8.72902774810791,
-      "learning_rate": 1.7018633540372672e-05,
-      "loss": 1.9724,
+      "grad_norm": 11.038766860961914,
+      "learning_rate": 2.552795031055901e-05,
+      "loss": 1.397,
       "step": 3960
     },
     {
       "epoch": 0.7472237907020516,
-      "grad_norm": 10.127591133117676,
-      "learning_rate": 1.7011104837191795e-05,
-      "loss": 2.0511,
+      "grad_norm": 11.564054489135742,
+      "learning_rate": 2.551665725578769e-05,
+      "loss": 1.5357,
       "step": 3970
     },
     {
       "epoch": 0.7491059664972708,
-      "grad_norm": 16.748136520385742,
-      "learning_rate": 1.7003576134010918e-05,
-      "loss": 2.0714,
+      "grad_norm": 19.608352661132812,
+      "learning_rate": 2.5505364201016375e-05,
+      "loss": 1.5665,
       "step": 3980
     },
     {
       "epoch": 0.7509881422924901,
-      "grad_norm": 11.505964279174805,
-      "learning_rate": 1.699604743083004e-05,
-      "loss": 2.0682,
+      "grad_norm": 13.263577461242676,
+      "learning_rate": 2.549407114624506e-05,
+      "loss": 1.4692,
       "step": 3990
     },
     {
       "epoch": 0.7528703180877094,
-      "grad_norm": 12.179784774780273,
-      "learning_rate": 1.6988518727649163e-05,
-      "loss": 2.0715,
+      "grad_norm": 21.618656158447266,
+      "learning_rate": 2.5482778091473743e-05,
+      "loss": 1.5511,
       "step": 4000
     },
     {
       "epoch": 0.7547524938829286,
-      "grad_norm": 5.324981212615967,
-      "learning_rate": 1.6980990024468286e-05,
-      "loss": 2.0796,
+      "grad_norm": 4.379132270812988,
+      "learning_rate": 2.547148503670243e-05,
+      "loss": 1.6006,
       "step": 4010
     },
     {
       "epoch": 0.756634669678148,
-      "grad_norm": 6.018923759460449,
-      "learning_rate": 1.697346132128741e-05,
-      "loss": 1.9147,
+      "grad_norm": 5.841297626495361,
+      "learning_rate": 2.5460191981931112e-05,
+      "loss": 1.3704,
       "step": 4020
     },
     {
       "epoch": 0.7585168454733672,
-      "grad_norm": 5.128754138946533,
-      "learning_rate": 1.6965932618106532e-05,
-      "loss": 2.2234,
+      "grad_norm": 9.923529624938965,
+      "learning_rate": 2.5448898927159798e-05,
+      "loss": 1.6851,
       "step": 4030
     },
     {
       "epoch": 0.7603990212685865,
-      "grad_norm": 7.023144721984863,
-      "learning_rate": 1.6958403914925655e-05,
-      "loss": 1.9784,
+      "grad_norm": 6.657973766326904,
+      "learning_rate": 2.5437605872388484e-05,
+      "loss": 1.3996,
       "step": 4040
     },
     {
       "epoch": 0.7622811970638057,
-      "grad_norm": 5.513296127319336,
-      "learning_rate": 1.6950875211744778e-05,
-      "loss": 1.8725,
+      "grad_norm": 7.652090072631836,
+      "learning_rate": 2.5426312817617167e-05,
+      "loss": 1.3747,
       "step": 4050
     },
     {
       "epoch": 0.764163372859025,
-      "grad_norm": 5.376245498657227,
-      "learning_rate": 1.69433465085639e-05,
-      "loss": 1.9038,
+      "grad_norm": 6.833724498748779,
+      "learning_rate": 2.541501976284585e-05,
+      "loss": 1.3425,
       "step": 4060
     },
     {
       "epoch": 0.7660455486542443,
-      "grad_norm": 4.453749179840088,
-      "learning_rate": 1.6935817805383024e-05,
-      "loss": 2.0901,
+      "grad_norm": 3.7029807567596436,
+      "learning_rate": 2.5403726708074532e-05,
+      "loss": 1.5997,
       "step": 4070
     },
     {
       "epoch": 0.7679277244494636,
-      "grad_norm": 6.467559814453125,
-      "learning_rate": 1.6928289102202147e-05,
-      "loss": 1.9841,
+      "grad_norm": 9.09876537322998,
+      "learning_rate": 2.539243365330322e-05,
+      "loss": 1.3796,
       "step": 4080
     },
     {
       "epoch": 0.7698099002446829,
-      "grad_norm": 5.831012725830078,
-      "learning_rate": 1.692076039902127e-05,
-      "loss": 1.7553,
+      "grad_norm": 6.666367053985596,
+      "learning_rate": 2.5381140598531905e-05,
+      "loss": 1.2452,
       "step": 4090
     },
     {
       "epoch": 0.7716920760399021,
-      "grad_norm": 8.568511009216309,
-      "learning_rate": 1.6913231695840393e-05,
-      "loss": 2.0252,
+      "grad_norm": 53.3235969543457,
+      "learning_rate": 2.5369847543760587e-05,
+      "loss": 1.4592,
       "step": 4100
     },
     {
       "epoch": 0.7735742518351214,
-      "grad_norm": 6.630292892456055,
-      "learning_rate": 1.6905702992659516e-05,
-      "loss": 2.0024,
+      "grad_norm": 14.514080047607422,
+      "learning_rate": 2.5358554488989273e-05,
+      "loss": 1.4565,
       "step": 4110
     },
     {
       "epoch": 0.7754564276303407,
-      "grad_norm": 5.808889389038086,
-      "learning_rate": 1.689817428947864e-05,
-      "loss": 1.7828,
+      "grad_norm": 4.159520626068115,
+      "learning_rate": 2.5347261434217956e-05,
+      "loss": 1.2396,
       "step": 4120
     },
     {
       "epoch": 0.7773386034255599,
-      "grad_norm": 4.167357444763184,
-      "learning_rate": 1.689064558629776e-05,
-      "loss": 1.7878,
+      "grad_norm": 6.2468438148498535,
+      "learning_rate": 2.5335968379446642e-05,
+      "loss": 1.2646,
       "step": 4130
     },
     {
       "epoch": 0.7792207792207793,
-      "grad_norm": 16.51450538635254,
-      "learning_rate": 1.6883116883116884e-05,
-      "loss": 2.0933,
+      "grad_norm": 14.043899536132812,
+      "learning_rate": 2.5324675324675325e-05,
+      "loss": 1.5409,
       "step": 4140
     },
     {
       "epoch": 0.7811029550159985,
-      "grad_norm": 7.76092004776001,
-      "learning_rate": 1.6875588179936007e-05,
-      "loss": 1.8682,
+      "grad_norm": 10.741174697875977,
+      "learning_rate": 2.5313382269904007e-05,
+      "loss": 1.2822,
       "step": 4150
     },
     {
       "epoch": 0.7829851308112178,
-      "grad_norm": 3.58233904838562,
-      "learning_rate": 1.686805947675513e-05,
-      "loss": 2.0892,
+      "grad_norm": 5.443540573120117,
+      "learning_rate": 2.5302089215132694e-05,
+      "loss": 1.561,
       "step": 4160
     },
     {
       "epoch": 0.784867306606437,
-      "grad_norm": 4.204033851623535,
-      "learning_rate": 1.6860530773574253e-05,
-      "loss": 2.132,
+      "grad_norm": 4.099931716918945,
+      "learning_rate": 2.529079616036138e-05,
+      "loss": 1.6496,
       "step": 4170
     },
     {
       "epoch": 0.7867494824016563,
-      "grad_norm": 4.688413619995117,
-      "learning_rate": 1.6853002070393376e-05,
-      "loss": 1.8572,
+      "grad_norm": 4.217696666717529,
+      "learning_rate": 2.5279503105590062e-05,
+      "loss": 1.3227,
       "step": 4180
     },
     {
       "epoch": 0.7886316581968756,
-      "grad_norm": 10.84228515625,
-      "learning_rate": 1.68454733672125e-05,
-      "loss": 2.0358,
+      "grad_norm": 18.865585327148438,
+      "learning_rate": 2.526821005081875e-05,
+      "loss": 1.5359,
       "step": 4190
     },
     {
       "epoch": 0.7905138339920948,
-      "grad_norm": 6.188892841339111,
-      "learning_rate": 1.6837944664031622e-05,
-      "loss": 1.8994,
+      "grad_norm": 7.36782693862915,
+      "learning_rate": 2.525691699604743e-05,
+      "loss": 1.3648,
       "step": 4200
     },
     {
       "epoch": 0.7923960097873142,
-      "grad_norm": 14.771183013916016,
-      "learning_rate": 1.6830415960850745e-05,
-      "loss": 2.006,
+      "grad_norm": 11.05923843383789,
+      "learning_rate": 2.5245623941276117e-05,
+      "loss": 1.556,
       "step": 4210
     },
     {
       "epoch": 0.7942781855825334,
-      "grad_norm": 6.336653709411621,
-      "learning_rate": 1.6822887257669868e-05,
-      "loss": 1.9038,
+      "grad_norm": 9.203524589538574,
+      "learning_rate": 2.5234330886504803e-05,
+      "loss": 1.4646,
       "step": 4220
     },
     {
       "epoch": 0.7961603613777527,
-      "grad_norm": 7.478043079376221,
-      "learning_rate": 1.681535855448899e-05,
-      "loss": 1.9509,
+      "grad_norm": 9.965381622314453,
+      "learning_rate": 2.5223037831733483e-05,
+      "loss": 1.4233,
       "step": 4230
     },
     {
       "epoch": 0.7980425371729719,
-      "grad_norm": 12.106429100036621,
-      "learning_rate": 1.6807829851308114e-05,
-      "loss": 1.8751,
+      "grad_norm": 17.369232177734375,
+      "learning_rate": 2.521174477696217e-05,
+      "loss": 1.4067,
       "step": 4240
     },
     {
       "epoch": 0.7999247129681912,
-      "grad_norm": 8.335380554199219,
-      "learning_rate": 1.6800301148127236e-05,
-      "loss": 2.1869,
+      "grad_norm": 4.989372253417969,
+      "learning_rate": 2.520045172219085e-05,
+      "loss": 1.773,
       "step": 4250
     },
     {
       "epoch": 0.8018068887634106,
-      "grad_norm": 9.204015731811523,
-      "learning_rate": 1.679277244494636e-05,
-      "loss": 2.2222,
+      "grad_norm": 10.048711776733398,
+      "learning_rate": 2.5189158667419537e-05,
+      "loss": 1.8644,
       "step": 4260
     },
     {
       "epoch": 0.8036890645586298,
-      "grad_norm": 5.910819053649902,
-      "learning_rate": 1.6785243741765482e-05,
-      "loss": 1.7396,
+      "grad_norm": 4.219175338745117,
+      "learning_rate": 2.5177865612648223e-05,
+      "loss": 1.133,
       "step": 4270
     },
     {
       "epoch": 0.8055712403538491,
-      "grad_norm": 12.734546661376953,
-      "learning_rate": 1.6777715038584605e-05,
-      "loss": 2.0184,
+      "grad_norm": 20.187997817993164,
+      "learning_rate": 2.5166572557876906e-05,
+      "loss": 1.5066,
       "step": 4280
     },
     {
       "epoch": 0.8074534161490683,
-      "grad_norm": 8.117308616638184,
-      "learning_rate": 1.6770186335403728e-05,
-      "loss": 2.1315,
+      "grad_norm": 7.595444202423096,
+      "learning_rate": 2.5155279503105592e-05,
+      "loss": 1.6849,
       "step": 4290
     },
     {
       "epoch": 0.8093355919442876,
-      "grad_norm": 6.764369010925293,
-      "learning_rate": 1.676265763222285e-05,
-      "loss": 1.7082,
+      "grad_norm": 14.513413429260254,
+      "learning_rate": 2.5143986448334275e-05,
+      "loss": 1.3276,
       "step": 4300
     },
     {
       "epoch": 0.8112177677395068,
-      "grad_norm": 11.162191390991211,
-      "learning_rate": 1.6755128929041974e-05,
-      "loss": 1.8846,
+      "grad_norm": 15.310640335083008,
+      "learning_rate": 2.513269339356296e-05,
+      "loss": 1.3818,
       "step": 4310
     },
     {
       "epoch": 0.8130999435347261,
-      "grad_norm": 7.554187297821045,
-      "learning_rate": 1.6747600225861097e-05,
-      "loss": 1.7372,
+      "grad_norm": 10.264412879943848,
+      "learning_rate": 2.5121400338791644e-05,
+      "loss": 1.2301,
       "step": 4320
     },
     {
       "epoch": 0.8149821193299455,
-      "grad_norm": 6.000895977020264,
-      "learning_rate": 1.674007152268022e-05,
-      "loss": 2.1344,
+      "grad_norm": 6.609447002410889,
+      "learning_rate": 2.5110107284020326e-05,
+      "loss": 1.646,
       "step": 4330
     },
     {
       "epoch": 0.8168642951251647,
-      "grad_norm": 6.909875869750977,
-      "learning_rate": 1.6732542819499343e-05,
-      "loss": 1.7821,
+      "grad_norm": 5.502641677856445,
+      "learning_rate": 2.5098814229249012e-05,
+      "loss": 1.2485,
       "step": 4340
     },
     {
       "epoch": 0.818746470920384,
-      "grad_norm": 9.617115020751953,
-      "learning_rate": 1.6725014116318466e-05,
-      "loss": 1.8464,
+      "grad_norm": 10.554330825805664,
+      "learning_rate": 2.5087521174477695e-05,
+      "loss": 1.3115,
       "step": 4350
     },
     {
       "epoch": 0.8206286467156032,
-      "grad_norm": 3.998657703399658,
-      "learning_rate": 1.671748541313759e-05,
-      "loss": 1.6707,
+      "grad_norm": 4.31608247756958,
+      "learning_rate": 2.507622811970638e-05,
+      "loss": 1.1539,
       "step": 4360
     },
     {
       "epoch": 0.8225108225108225,
-      "grad_norm": 18.605613708496094,
-      "learning_rate": 1.670995670995671e-05,
-      "loss": 1.9964,
+      "grad_norm": 8.371481895446777,
+      "learning_rate": 2.5064935064935067e-05,
+      "loss": 1.4917,
       "step": 4370
     },
     {
       "epoch": 0.8243929983060417,
-      "grad_norm": 21.7368221282959,
-      "learning_rate": 1.6702428006775834e-05,
-      "loss": 1.7039,
+      "grad_norm": 19.33193016052246,
+      "learning_rate": 2.505364201016375e-05,
+      "loss": 1.2167,
       "step": 4380
     },
     {
       "epoch": 0.8262751741012611,
-      "grad_norm": 10.246116638183594,
-      "learning_rate": 1.6694899303594957e-05,
-      "loss": 1.8365,
+      "grad_norm": 14.059998512268066,
+      "learning_rate": 2.5042348955392436e-05,
+      "loss": 1.4155,
       "step": 4390
     },
     {
       "epoch": 0.8281573498964804,
-      "grad_norm": 12.197159767150879,
-      "learning_rate": 1.668737060041408e-05,
-      "loss": 2.0829,
+      "grad_norm": 19.17221450805664,
+      "learning_rate": 2.503105590062112e-05,
+      "loss": 1.6316,
       "step": 4400
     },
     {
       "epoch": 0.8300395256916996,
-      "grad_norm": 7.530276298522949,
-      "learning_rate": 1.6679841897233203e-05,
-      "loss": 1.8986,
+      "grad_norm": 3.8574047088623047,
+      "learning_rate": 2.50197628458498e-05,
+      "loss": 1.3991,
       "step": 4410
     },
     {
       "epoch": 0.8319217014869189,
-      "grad_norm": 9.89567756652832,
-      "learning_rate": 1.6672313194052326e-05,
-      "loss": 1.9985,
+      "grad_norm": 11.87222671508789,
+      "learning_rate": 2.5008469791078487e-05,
+      "loss": 1.5969,
       "step": 4420
     },
     {
       "epoch": 0.8338038772821381,
-      "grad_norm": 10.721818923950195,
-      "learning_rate": 1.666478449087145e-05,
-      "loss": 1.8921,
+      "grad_norm": 5.910922050476074,
+      "learning_rate": 2.499717673630717e-05,
+      "loss": 1.4359,
       "step": 4430
     },
     {
       "epoch": 0.8356860530773574,
-      "grad_norm": 12.71999740600586,
-      "learning_rate": 1.6657255787690572e-05,
-      "loss": 1.8283,
+      "grad_norm": 13.17109203338623,
+      "learning_rate": 2.4985883681535856e-05,
+      "loss": 1.3405,
       "step": 4440
     },
     {
       "epoch": 0.8375682288725766,
-      "grad_norm": 8.337474822998047,
-      "learning_rate": 1.6649727084509695e-05,
-      "loss": 1.892,
+      "grad_norm": 10.746623039245605,
+      "learning_rate": 2.4974590626764542e-05,
+      "loss": 1.4437,
       "step": 4450
     },
     {
       "epoch": 0.839450404667796,
-      "grad_norm": 12.982160568237305,
-      "learning_rate": 1.6642198381328818e-05,
-      "loss": 1.9852,
+      "grad_norm": 18.121070861816406,
+      "learning_rate": 2.4963297571993225e-05,
+      "loss": 1.4979,
       "step": 4460
     },
     {
       "epoch": 0.8413325804630153,
-      "grad_norm": 7.9721784591674805,
-      "learning_rate": 1.663466967814794e-05,
-      "loss": 2.096,
+      "grad_norm": 8.530396461486816,
+      "learning_rate": 2.495200451722191e-05,
+      "loss": 1.7246,
       "step": 4470
     },
     {
       "epoch": 0.8432147562582345,
-      "grad_norm": 13.718586921691895,
-      "learning_rate": 1.6627140974967064e-05,
-      "loss": 1.9495,
+      "grad_norm": 16.504993438720703,
+      "learning_rate": 2.4940711462450594e-05,
+      "loss": 1.4092,
       "step": 4480
     },
     {
       "epoch": 0.8450969320534538,
-      "grad_norm": 6.098572731018066,
-      "learning_rate": 1.6619612271786187e-05,
-      "loss": 1.5641,
+      "grad_norm": 7.536314010620117,
+      "learning_rate": 2.4929418407679276e-05,
+      "loss": 0.9995,
       "step": 4490
     },
     {
       "epoch": 0.846979107848673,
-      "grad_norm": 5.264159202575684,
-      "learning_rate": 1.661208356860531e-05,
-      "loss": 1.8201,
+      "grad_norm": 5.592616081237793,
+      "learning_rate": 2.4918125352907963e-05,
+      "loss": 1.3312,
       "step": 4500
     },
     {
       "epoch": 0.8488612836438924,
-      "grad_norm": 15.892148971557617,
-      "learning_rate": 1.6604554865424432e-05,
-      "loss": 1.8025,
+      "grad_norm": 18.722604751586914,
+      "learning_rate": 2.4906832298136645e-05,
+      "loss": 1.4819,
       "step": 4510
     },
     {
       "epoch": 0.8507434594391116,
-      "grad_norm": 19.255720138549805,
-      "learning_rate": 1.6597026162243555e-05,
-      "loss": 1.8699,
+      "grad_norm": 16.095701217651367,
+      "learning_rate": 2.489553924336533e-05,
+      "loss": 1.4257,
       "step": 4520
     },
     {
       "epoch": 0.8526256352343309,
-      "grad_norm": 12.752776145935059,
-      "learning_rate": 1.6589497459062678e-05,
-      "loss": 1.9059,
+      "grad_norm": 11.100549697875977,
+      "learning_rate": 2.4884246188594014e-05,
+      "loss": 1.4673,
       "step": 4530
     },
     {
       "epoch": 0.8545078110295502,
-      "grad_norm": 24.916339874267578,
-      "learning_rate": 1.65819687558818e-05,
-      "loss": 1.8327,
+      "grad_norm": 22.106538772583008,
+      "learning_rate": 2.48729531338227e-05,
+      "loss": 1.3014,
       "step": 4540
     },
     {
       "epoch": 0.8563899868247694,
-      "grad_norm": 16.435821533203125,
-      "learning_rate": 1.6574440052700924e-05,
-      "loss": 1.907,
+      "grad_norm": 9.377236366271973,
+      "learning_rate": 2.4861660079051386e-05,
+      "loss": 1.5026,
       "step": 4550
     },
     {
       "epoch": 0.8582721626199887,
-      "grad_norm": 8.208995819091797,
-      "learning_rate": 1.6566911349520047e-05,
-      "loss": 1.7911,
+      "grad_norm": 7.03391695022583,
+      "learning_rate": 2.485036702428007e-05,
+      "loss": 1.3629,
       "step": 4560
     },
     {
       "epoch": 0.8601543384152079,
-      "grad_norm": 19.37531089782715,
-      "learning_rate": 1.655938264633917e-05,
-      "loss": 2.0121,
+      "grad_norm": 19.828781127929688,
+      "learning_rate": 2.4839073969508755e-05,
+      "loss": 1.6024,
       "step": 4570
     },
     {
       "epoch": 0.8620365142104273,
-      "grad_norm": 9.35920524597168,
-      "learning_rate": 1.6551853943158293e-05,
-      "loss": 1.8435,
+      "grad_norm": 14.324966430664062,
+      "learning_rate": 2.4827780914737434e-05,
+      "loss": 1.3317,
       "step": 4580
     },
     {
       "epoch": 0.8639186900056465,
-      "grad_norm": 19.906599044799805,
-      "learning_rate": 1.6544325239977416e-05,
-      "loss": 2.0695,
+      "grad_norm": 17.969161987304688,
+      "learning_rate": 2.481648785996612e-05,
+      "loss": 1.5452,
       "step": 4590
     },
     {
       "epoch": 0.8658008658008658,
-      "grad_norm": 6.58824348449707,
-      "learning_rate": 1.653679653679654e-05,
-      "loss": 1.8859,
+      "grad_norm": 28.252727508544922,
+      "learning_rate": 2.4805194805194806e-05,
+      "loss": 1.3952,
       "step": 4600
     },
     {
       "epoch": 0.8676830415960851,
-      "grad_norm": 7.258162975311279,
-      "learning_rate": 1.652926783361566e-05,
-      "loss": 1.8014,
+      "grad_norm": 6.551124095916748,
+      "learning_rate": 2.479390175042349e-05,
+      "loss": 1.4017,
       "step": 4610
     },
     {
       "epoch": 0.8695652173913043,
-      "grad_norm": 10.79643726348877,
-      "learning_rate": 1.6521739130434785e-05,
-      "loss": 1.9955,
+      "grad_norm": 9.968870162963867,
+      "learning_rate": 2.4782608695652175e-05,
+      "loss": 1.4794,
       "step": 4620
     },
     {
       "epoch": 0.8714473931865236,
-      "grad_norm": 8.076590538024902,
-      "learning_rate": 1.6514210427253907e-05,
-      "loss": 1.9565,
+      "grad_norm": 10.416319847106934,
+      "learning_rate": 2.4771315640880858e-05,
+      "loss": 1.4633,
       "step": 4630
     },
     {
       "epoch": 0.8733295689817429,
-      "grad_norm": 34.28034591674805,
-      "learning_rate": 1.650668172407303e-05,
-      "loss": 1.712,
+      "grad_norm": 12.675484657287598,
+      "learning_rate": 2.4760022586109544e-05,
+      "loss": 1.1787,
       "step": 4640
     },
     {
       "epoch": 0.8752117447769622,
-      "grad_norm": 13.998528480529785,
-      "learning_rate": 1.6499153020892153e-05,
-      "loss": 1.9953,
+      "grad_norm": 4.599917411804199,
+      "learning_rate": 2.474872953133823e-05,
+      "loss": 1.4834,
       "step": 4650
     },
     {
       "epoch": 0.8770939205721815,
-      "grad_norm": 11.216374397277832,
-      "learning_rate": 1.6491624317711273e-05,
-      "loss": 1.7193,
+      "grad_norm": 23.191003799438477,
+      "learning_rate": 2.473743647656691e-05,
+      "loss": 1.2879,
       "step": 4660
     },
     {
       "epoch": 0.8789760963674007,
-      "grad_norm": 18.6248722076416,
-      "learning_rate": 1.64840956145304e-05,
-      "loss": 1.6458,
+      "grad_norm": 10.024175643920898,
+      "learning_rate": 2.4726143421795595e-05,
+      "loss": 1.1128,
       "step": 4670
     },
     {
       "epoch": 0.88085827216262,
-      "grad_norm": 4.484902858734131,
-      "learning_rate": 1.6476566911349522e-05,
-      "loss": 1.8185,
+      "grad_norm": 3.1574692726135254,
+      "learning_rate": 2.471485036702428e-05,
+      "loss": 1.2905,
       "step": 4680
     },
     {
       "epoch": 0.8827404479578392,
-      "grad_norm": 20.6184024810791,
-      "learning_rate": 1.6469038208168645e-05,
-      "loss": 1.8833,
+      "grad_norm": 5.446253776550293,
+      "learning_rate": 2.4703557312252964e-05,
+      "loss": 1.5343,
       "step": 4690
     },
     {
       "epoch": 0.8846226237530586,
-      "grad_norm": 10.890295028686523,
-      "learning_rate": 1.6461509504987768e-05,
-      "loss": 1.7697,
+      "grad_norm": 13.522638320922852,
+      "learning_rate": 2.469226425748165e-05,
+      "loss": 1.3689,
       "step": 4700
     },
     {
       "epoch": 0.8865047995482778,
-      "grad_norm": 10.369776725769043,
-      "learning_rate": 1.645398080180689e-05,
-      "loss": 1.8408,
+      "grad_norm": 20.1780948638916,
+      "learning_rate": 2.4680971202710333e-05,
+      "loss": 1.4354,
       "step": 4710
     },
     {
       "epoch": 0.8883869753434971,
-      "grad_norm": 6.13337516784668,
-      "learning_rate": 1.6446452098626014e-05,
-      "loss": 1.9334,
+      "grad_norm": 10.403576850891113,
+      "learning_rate": 2.466967814793902e-05,
+      "loss": 1.5017,
       "step": 4720
     },
     {
       "epoch": 0.8902691511387164,
-      "grad_norm": 7.776824951171875,
-      "learning_rate": 1.6438923395445137e-05,
-      "loss": 1.6361,
+      "grad_norm": 9.197463989257812,
+      "learning_rate": 2.4658385093167705e-05,
+      "loss": 1.2154,
       "step": 4730
     },
     {
       "epoch": 0.8921513269339356,
-      "grad_norm": 16.185203552246094,
-      "learning_rate": 1.643139469226426e-05,
-      "loss": 1.6398,
+      "grad_norm": 11.027040481567383,
+      "learning_rate": 2.4647092038396388e-05,
+      "loss": 1.1206,
       "step": 4740
     },
     {
       "epoch": 0.8940335027291549,
-      "grad_norm": 16.69852638244629,
-      "learning_rate": 1.642386598908338e-05,
-      "loss": 1.8242,
+      "grad_norm": 10.964394569396973,
+      "learning_rate": 2.463579898362507e-05,
+      "loss": 1.3562,
       "step": 4750
     },
     {
       "epoch": 0.8959156785243741,
-      "grad_norm": 8.41193675994873,
-      "learning_rate": 1.6416337285902505e-05,
-      "loss": 1.6251,
+      "grad_norm": 8.044187545776367,
+      "learning_rate": 2.4624505928853753e-05,
+      "loss": 1.1975,
       "step": 4760
     },
     {
       "epoch": 0.8977978543195935,
-      "grad_norm": 7.705557823181152,
-      "learning_rate": 1.640880858272163e-05,
-      "loss": 1.9138,
+      "grad_norm": 12.479483604431152,
+      "learning_rate": 2.461321287408244e-05,
+      "loss": 1.3726,
       "step": 4770
     },
     {
       "epoch": 0.8996800301148127,
-      "grad_norm": 6.7078537940979,
-      "learning_rate": 1.640127987954075e-05,
-      "loss": 1.8573,
+      "grad_norm": 10.56410026550293,
+      "learning_rate": 2.4601919819311125e-05,
+      "loss": 1.3986,
       "step": 4780
     },
     {
       "epoch": 0.901562205910032,
-      "grad_norm": 9.067506790161133,
-      "learning_rate": 1.6393751176359874e-05,
-      "loss": 1.7235,
+      "grad_norm": 13.470977783203125,
+      "learning_rate": 2.4590626764539808e-05,
+      "loss": 1.2694,
       "step": 4790
     },
     {
       "epoch": 0.9034443817052513,
-      "grad_norm": 6.557106971740723,
-      "learning_rate": 1.6386222473178997e-05,
-      "loss": 1.5589,
+      "grad_norm": 4.0369744300842285,
+      "learning_rate": 2.4579333709768494e-05,
+      "loss": 1.0665,
       "step": 4800
     },
     {
       "epoch": 0.9053265575004705,
-      "grad_norm": 18.50225067138672,
-      "learning_rate": 1.637869376999812e-05,
-      "loss": 1.7907,
+      "grad_norm": 15.835465431213379,
+      "learning_rate": 2.4568040654997177e-05,
+      "loss": 1.2974,
       "step": 4810
     },
     {
       "epoch": 0.9072087332956899,
-      "grad_norm": 13.424736022949219,
-      "learning_rate": 1.6371165066817243e-05,
-      "loss": 1.8507,
+      "grad_norm": 13.87853717803955,
+      "learning_rate": 2.4556747600225863e-05,
+      "loss": 1.3459,
       "step": 4820
     },
     {
       "epoch": 0.9090909090909091,
-      "grad_norm": 6.312253952026367,
-      "learning_rate": 1.6363636363636366e-05,
-      "loss": 1.8001,
+      "grad_norm": 27.265249252319336,
+      "learning_rate": 2.454545454545455e-05,
+      "loss": 1.3109,
       "step": 4830
     },
     {
       "epoch": 0.9109730848861284,
-      "grad_norm": 4.222917556762695,
-      "learning_rate": 1.6356107660455485e-05,
-      "loss": 1.9496,
+      "grad_norm": 5.671602725982666,
+      "learning_rate": 2.4534161490683228e-05,
+      "loss": 1.6534,
       "step": 4840
     },
     {
       "epoch": 0.9128552606813476,
-      "grad_norm": 14.68265438079834,
-      "learning_rate": 1.6348578957274612e-05,
-      "loss": 1.7783,
+      "grad_norm": 4.075274467468262,
+      "learning_rate": 2.4522868435911914e-05,
+      "loss": 1.3106,
       "step": 4850
     },
     {
       "epoch": 0.9147374364765669,
-      "grad_norm": 7.720460891723633,
-      "learning_rate": 1.6341050254093735e-05,
-      "loss": 1.6416,
+      "grad_norm": 17.051898956298828,
+      "learning_rate": 2.4511575381140597e-05,
+      "loss": 1.2559,
       "step": 4860
     },
     {
       "epoch": 0.9166196122717862,
-      "grad_norm": 7.769947052001953,
-      "learning_rate": 1.6333521550912858e-05,
-      "loss": 1.8749,
+      "grad_norm": 14.164190292358398,
+      "learning_rate": 2.4500282326369283e-05,
+      "loss": 1.4232,
       "step": 4870
     },
     {
       "epoch": 0.9185017880670054,
-      "grad_norm": 9.666947364807129,
-      "learning_rate": 1.632599284773198e-05,
-      "loss": 1.6327,
+      "grad_norm": 4.818867206573486,
+      "learning_rate": 2.448898927159797e-05,
+      "loss": 1.1879,
       "step": 4880
     },
     {
       "epoch": 0.9203839638622248,
-      "grad_norm": 12.226252555847168,
-      "learning_rate": 1.6318464144551103e-05,
-      "loss": 1.7772,
+      "grad_norm": 6.399837493896484,
+      "learning_rate": 2.4477696216826652e-05,
+      "loss": 1.3137,
       "step": 4890
     },
     {
       "epoch": 0.922266139657444,
-      "grad_norm": 21.034391403198242,
-      "learning_rate": 1.6310935441370226e-05,
-      "loss": 1.8364,
+      "grad_norm": 26.00821304321289,
+      "learning_rate": 2.4466403162055338e-05,
+      "loss": 1.4008,
       "step": 4900
     },
     {
       "epoch": 0.9241483154526633,
-      "grad_norm": 11.93802261352539,
-      "learning_rate": 1.630340673818935e-05,
-      "loss": 1.8696,
+      "grad_norm": 19.372074127197266,
+      "learning_rate": 2.445511010728402e-05,
+      "loss": 1.4288,
       "step": 4910
     },
     {
       "epoch": 0.9260304912478825,
-      "grad_norm": 16.049116134643555,
-      "learning_rate": 1.6295878035008472e-05,
-      "loss": 1.8541,
+      "grad_norm": 21.262939453125,
+      "learning_rate": 2.4443817052512707e-05,
+      "loss": 1.3958,
       "step": 4920
     },
     {
       "epoch": 0.9279126670431018,
-      "grad_norm": 12.634326934814453,
-      "learning_rate": 1.6288349331827592e-05,
-      "loss": 1.6714,
+      "grad_norm": 8.744524955749512,
+      "learning_rate": 2.443252399774139e-05,
+      "loss": 1.2075,
       "step": 4930
     },
     {
       "epoch": 0.9297948428383211,
-      "grad_norm": 17.020570755004883,
-      "learning_rate": 1.6280820628646715e-05,
-      "loss": 1.915,
+      "grad_norm": 19.05074691772461,
+      "learning_rate": 2.4421230942970072e-05,
+      "loss": 1.4352,
       "step": 4940
     },
     {
       "epoch": 0.9316770186335404,
-      "grad_norm": 3.1234688758850098,
-      "learning_rate": 1.627329192546584e-05,
-      "loss": 1.4168,
+      "grad_norm": 4.271505355834961,
+      "learning_rate": 2.4409937888198758e-05,
+      "loss": 0.8662,
       "step": 4950
     },
     {
       "epoch": 0.9335591944287597,
-      "grad_norm": 14.488174438476562,
-      "learning_rate": 1.6265763222284964e-05,
-      "loss": 1.4905,
+      "grad_norm": 11.079838752746582,
+      "learning_rate": 2.4398644833427444e-05,
+      "loss": 0.9989,
       "step": 4960
     },
     {
       "epoch": 0.9354413702239789,
-      "grad_norm": 9.10696029663086,
-      "learning_rate": 1.6258234519104087e-05,
-      "loss": 1.7308,
+      "grad_norm": 12.747160911560059,
+      "learning_rate": 2.4387351778656127e-05,
+      "loss": 1.3546,
       "step": 4970
     },
     {
       "epoch": 0.9373235460191982,
-      "grad_norm": 10.455379486083984,
-      "learning_rate": 1.625070581592321e-05,
-      "loss": 1.6388,
+      "grad_norm": 9.06821060180664,
+      "learning_rate": 2.4376058723884813e-05,
+      "loss": 1.237,
       "step": 4980
     },
     {
       "epoch": 0.9392057218144174,
-      "grad_norm": 12.714646339416504,
-      "learning_rate": 1.6243177112742333e-05,
-      "loss": 1.5879,
+      "grad_norm": 18.579593658447266,
+      "learning_rate": 2.4364765669113496e-05,
+      "loss": 1.0582,
       "step": 4990
     },
     {
       "epoch": 0.9410878976096367,
-      "grad_norm": 33.10146713256836,
-      "learning_rate": 1.6235648409561456e-05,
-      "loss": 1.4738,
+      "grad_norm": 25.581363677978516,
+      "learning_rate": 2.435347261434218e-05,
+      "loss": 1.1142,
       "step": 5000
     },
     {
       "epoch": 0.9429700734048561,
-      "grad_norm": 5.845310688018799,
-      "learning_rate": 1.6228119706380575e-05,
-      "loss": 1.6773,
+      "grad_norm": 3.7508864402770996,
+      "learning_rate": 2.4342179559570864e-05,
+      "loss": 1.1593,
       "step": 5010
     },
     {
       "epoch": 0.9448522492000753,
-      "grad_norm": 8.02888298034668,
-      "learning_rate": 1.6220591003199698e-05,
-      "loss": 1.7755,
+      "grad_norm": 12.199710845947266,
+      "learning_rate": 2.4330886504799547e-05,
+      "loss": 1.2923,
       "step": 5020
     },
     {
       "epoch": 0.9467344249952946,
-      "grad_norm": 17.18178367614746,
-      "learning_rate": 1.621306230001882e-05,
-      "loss": 1.669,
+      "grad_norm": 15.573914527893066,
+      "learning_rate": 2.4319593450028233e-05,
+      "loss": 1.2529,
       "step": 5030
     },
     {
       "epoch": 0.9486166007905138,
-      "grad_norm": 8.009747505187988,
-      "learning_rate": 1.6205533596837947e-05,
-      "loss": 1.6732,
+      "grad_norm": 10.434758186340332,
+      "learning_rate": 2.4308300395256916e-05,
+      "loss": 1.2285,
       "step": 5040
     },
     {
       "epoch": 0.9504987765857331,
-      "grad_norm": 7.900938034057617,
-      "learning_rate": 1.619800489365707e-05,
-      "loss": 1.6969,
+      "grad_norm": 14.420498847961426,
+      "learning_rate": 2.4297007340485602e-05,
+      "loss": 1.3006,
       "step": 5050
     },
     {
       "epoch": 0.9523809523809523,
-      "grad_norm": 12.25010871887207,
-      "learning_rate": 1.6190476190476193e-05,
-      "loss": 1.8035,
+      "grad_norm": 32.99720764160156,
+      "learning_rate": 2.4285714285714288e-05,
+      "loss": 1.4461,
       "step": 5060
     },
     {
       "epoch": 0.9542631281761716,
-      "grad_norm": 7.510868072509766,
-      "learning_rate": 1.6182947487295316e-05,
-      "loss": 1.5132,
+      "grad_norm": 4.45308780670166,
+      "learning_rate": 2.427442123094297e-05,
+      "loss": 1.1251,
       "step": 5070
     },
     {
       "epoch": 0.956145303971391,
-      "grad_norm": 6.639998435974121,
-      "learning_rate": 1.617541878411444e-05,
-      "loss": 1.5381,
+      "grad_norm": 23.680532455444336,
+      "learning_rate": 2.4263128176171657e-05,
+      "loss": 1.0521,
       "step": 5080
     },
     {
       "epoch": 0.9580274797666102,
-      "grad_norm": 11.637256622314453,
-      "learning_rate": 1.6167890080933562e-05,
-      "loss": 1.704,
+      "grad_norm": 7.266300201416016,
+      "learning_rate": 2.425183512140034e-05,
+      "loss": 1.2313,
       "step": 5090
     },
     {
       "epoch": 0.9599096555618295,
-      "grad_norm": 8.058938026428223,
-      "learning_rate": 1.616036137775268e-05,
-      "loss": 1.5851,
+      "grad_norm": 9.216297149658203,
+      "learning_rate": 2.4240542066629022e-05,
+      "loss": 1.1339,
       "step": 5100
     },
     {
       "epoch": 0.9617918313570487,
-      "grad_norm": 9.28378677368164,
-      "learning_rate": 1.6152832674571804e-05,
-      "loss": 1.6364,
+      "grad_norm": 11.177430152893066,
+      "learning_rate": 2.4229249011857708e-05,
+      "loss": 1.1926,
       "step": 5110
     },
     {
       "epoch": 0.963674007152268,
-      "grad_norm": 12.845236778259277,
-      "learning_rate": 1.6145303971390927e-05,
-      "loss": 1.845,
+      "grad_norm": 21.9661808013916,
+      "learning_rate": 2.421795595708639e-05,
+      "loss": 1.4161,
       "step": 5120
     },
     {
       "epoch": 0.9655561829474872,
-      "grad_norm": 10.16936206817627,
-      "learning_rate": 1.6137775268210054e-05,
-      "loss": 1.6366,
+      "grad_norm": 12.636842727661133,
+      "learning_rate": 2.4206662902315077e-05,
+      "loss": 1.1809,
       "step": 5130
     },
     {
       "epoch": 0.9674383587427066,
-      "grad_norm": 9.189414024353027,
-      "learning_rate": 1.6130246565029176e-05,
-      "loss": 1.5852,
+      "grad_norm": 12.683283805847168,
+      "learning_rate": 2.419536984754376e-05,
+      "loss": 1.2115,
       "step": 5140
     },
     {
       "epoch": 0.9693205345379259,
-      "grad_norm": 8.392788887023926,
-      "learning_rate": 1.61227178618483e-05,
-      "loss": 1.736,
+      "grad_norm": 14.036438941955566,
+      "learning_rate": 2.4184076792772446e-05,
+      "loss": 1.4084,
       "step": 5150
     },
     {
       "epoch": 0.9712027103331451,
-      "grad_norm": 7.551772594451904,
-      "learning_rate": 1.6115189158667422e-05,
-      "loss": 1.742,
+      "grad_norm": 8.703644752502441,
+      "learning_rate": 2.4172783738001132e-05,
+      "loss": 1.3985,
       "step": 5160
     },
     {
       "epoch": 0.9730848861283644,
-      "grad_norm": 11.65742015838623,
-      "learning_rate": 1.6107660455486545e-05,
-      "loss": 1.7394,
+      "grad_norm": 15.209477424621582,
+      "learning_rate": 2.4161490683229814e-05,
+      "loss": 1.3594,
       "step": 5170
     },
     {
       "epoch": 0.9749670619235836,
-      "grad_norm": 5.885232448577881,
-      "learning_rate": 1.6100131752305668e-05,
-      "loss": 1.6527,
+      "grad_norm": 8.397993087768555,
+      "learning_rate": 2.41501976284585e-05,
+      "loss": 1.2813,
       "step": 5180
     },
     {
       "epoch": 0.9768492377188029,
-      "grad_norm": 9.612196922302246,
-      "learning_rate": 1.6092603049124788e-05,
-      "loss": 1.5318,
+      "grad_norm": 11.615766525268555,
+      "learning_rate": 2.4138904573687183e-05,
+      "loss": 1.1447,
       "step": 5190
     },
     {
       "epoch": 0.9787314135140223,
-      "grad_norm": 16.069961547851562,
-      "learning_rate": 1.608507434594391e-05,
-      "loss": 1.5113,
+      "grad_norm": 3.9449048042297363,
+      "learning_rate": 2.4127611518915866e-05,
+      "loss": 1.0322,
       "step": 5200
     },
     {
       "epoch": 0.9806135893092415,
-      "grad_norm": 15.502707481384277,
-      "learning_rate": 1.6077545642763034e-05,
-      "loss": 1.5743,
+      "grad_norm": 23.256526947021484,
+      "learning_rate": 2.4116318464144552e-05,
+      "loss": 1.0706,
       "step": 5210
     },
     {
       "epoch": 0.9824957651044608,
-      "grad_norm": 24.306838989257812,
-      "learning_rate": 1.6070016939582156e-05,
-      "loss": 1.6745,
+      "grad_norm": 6.652702808380127,
+      "learning_rate": 2.4105025409373235e-05,
+      "loss": 1.2482,
       "step": 5220
     },
     {
       "epoch": 0.98437794089968,
-      "grad_norm": 4.061778545379639,
-      "learning_rate": 1.6062488236401283e-05,
-      "loss": 1.5577,
+      "grad_norm": 27.787336349487305,
+      "learning_rate": 2.409373235460192e-05,
+      "loss": 1.2317,
       "step": 5230
     },
     {
       "epoch": 0.9862601166948993,
-      "grad_norm": 5.753101825714111,
-      "learning_rate": 1.6054959533220406e-05,
-      "loss": 1.3725,
+      "grad_norm": 4.631372451782227,
+      "learning_rate": 2.4082439299830607e-05,
+      "loss": 0.9322,
       "step": 5240
     },
     {
       "epoch": 0.9881422924901185,
-      "grad_norm": 9.790191650390625,
-      "learning_rate": 1.604743083003953e-05,
-      "loss": 1.5817,
+      "grad_norm": 7.675413131713867,
+      "learning_rate": 2.407114624505929e-05,
+      "loss": 1.1535,
       "step": 5250
     },
     {
       "epoch": 0.9900244682853379,
-      "grad_norm": 4.075527191162109,
-      "learning_rate": 1.603990212685865e-05,
-      "loss": 1.4994,
+      "grad_norm": 11.056670188903809,
+      "learning_rate": 2.4059853190287976e-05,
+      "loss": 1.066,
       "step": 5260
     },
     {
       "epoch": 0.9919066440805572,
-      "grad_norm": 2.995372772216797,
-      "learning_rate": 1.603237342367777e-05,
-      "loss": 1.5081,
+      "grad_norm": 2.178229808807373,
+      "learning_rate": 2.4048560135516655e-05,
+      "loss": 1.0266,
       "step": 5270
     },
     {
       "epoch": 0.9937888198757764,
-      "grad_norm": 2.866976022720337,
-      "learning_rate": 1.6024844720496894e-05,
-      "loss": 1.7521,
+      "grad_norm": 5.711111068725586,
+      "learning_rate": 2.403726708074534e-05,
+      "loss": 1.2493,
       "step": 5280
     },
     {
       "epoch": 0.9956709956709957,
-      "grad_norm": 13.255521774291992,
-      "learning_rate": 1.6017316017316017e-05,
-      "loss": 1.4685,
+      "grad_norm": 27.90021514892578,
+      "learning_rate": 2.4025974025974027e-05,
+      "loss": 1.1559,
       "step": 5290
     },
     {
       "epoch": 0.9975531714662149,
-      "grad_norm": 5.982124328613281,
-      "learning_rate": 1.600978731413514e-05,
-      "loss": 1.5513,
+      "grad_norm": 8.907947540283203,
+      "learning_rate": 2.401468097120271e-05,
+      "loss": 1.0816,
       "step": 5300
     },
     {
       "epoch": 0.9994353472614342,
-      "grad_norm": 16.305452346801758,
-      "learning_rate": 1.6002258610954263e-05,
-      "loss": 1.4469,
+      "grad_norm": 21.335691452026367,
+      "learning_rate": 2.4003387916431396e-05,
+      "loss": 1.0634,
       "step": 5310
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.8716,
-      "eval_loss": 1.1871148347854614,
-      "eval_runtime": 143.0155,
-      "eval_samples_per_second": 52.442,
-      "eval_steps_per_second": 6.559,
+      "eval_accuracy": 0.8777333333333334,
+      "eval_loss": 0.7235307097434998,
+      "eval_runtime": 244.378,
+      "eval_samples_per_second": 30.69,
+      "eval_steps_per_second": 3.838,
       "step": 5313
     },
     {
       "epoch": 1.0013175230566536,
-      "grad_norm": 9.596165657043457,
-      "learning_rate": 1.599472990777339e-05,
-      "loss": 1.7778,
+      "grad_norm": 5.803168773651123,
+      "learning_rate": 2.399209486166008e-05,
+      "loss": 1.4338,
       "step": 5320
     },
     {
       "epoch": 1.0031996988518728,
-      "grad_norm": 9.728874206542969,
-      "learning_rate": 1.5987201204592512e-05,
-      "loss": 1.6959,
+      "grad_norm": 16.894943237304688,
+      "learning_rate": 2.3980801806888765e-05,
+      "loss": 1.3967,
       "step": 5330
     },
     {
       "epoch": 1.005081874647092,
-      "grad_norm": 11.146066665649414,
-      "learning_rate": 1.5979672501411635e-05,
-      "loss": 1.4466,
+      "grad_norm": 9.325886726379395,
+      "learning_rate": 2.396950875211745e-05,
+      "loss": 1.1418,
       "step": 5340
     },
     {
       "epoch": 1.0069640504423114,
-      "grad_norm": 11.417769432067871,
-      "learning_rate": 1.5972143798230758e-05,
-      "loss": 1.5719,
+      "grad_norm": 15.385432243347168,
+      "learning_rate": 2.3958215697346133e-05,
+      "loss": 1.1674,
       "step": 5350
     },
     {
       "epoch": 1.0088462262375306,
-      "grad_norm": 11.120503425598145,
-      "learning_rate": 1.5964615095049877e-05,
-      "loss": 1.6229,
+      "grad_norm": 16.328964233398438,
+      "learning_rate": 2.3946922642574816e-05,
+      "loss": 1.2397,
       "step": 5360
     },
     {
       "epoch": 1.0107284020327498,
-      "grad_norm": 24.465999603271484,
-      "learning_rate": 1.5957086391869e-05,
-      "loss": 1.6695,
+      "grad_norm": 16.835899353027344,
+      "learning_rate": 2.39356295878035e-05,
+      "loss": 1.1855,
       "step": 5370
     },
     {
       "epoch": 1.012610577827969,
-      "grad_norm": 6.379744529724121,
-      "learning_rate": 1.5949557688688123e-05,
-      "loss": 1.6745,
+      "grad_norm": 1.2168173789978027,
+      "learning_rate": 2.3924336533032185e-05,
+      "loss": 1.185,
       "step": 5380
     },
     {
       "epoch": 1.0144927536231885,
-      "grad_norm": 11.811795234680176,
-      "learning_rate": 1.5942028985507246e-05,
-      "loss": 1.3933,
+      "grad_norm": 17.501239776611328,
+      "learning_rate": 2.391304347826087e-05,
+      "loss": 1.0554,
       "step": 5390
     },
     {
       "epoch": 1.0163749294184077,
-      "grad_norm": 6.720681667327881,
-      "learning_rate": 1.593450028232637e-05,
-      "loss": 1.4849,
+      "grad_norm": 7.978825569152832,
+      "learning_rate": 2.3901750423489554e-05,
+      "loss": 1.0179,
       "step": 5400
     },
     {
       "epoch": 1.018257105213627,
-      "grad_norm": 5.741754531860352,
-      "learning_rate": 1.5926971579145495e-05,
-      "loss": 1.5725,
+      "grad_norm": 21.958223342895508,
+      "learning_rate": 2.389045736871824e-05,
+      "loss": 1.2771,
       "step": 5410
     },
     {
       "epoch": 1.0201392810088463,
-      "grad_norm": 15.412458419799805,
-      "learning_rate": 1.5919442875964618e-05,
-      "loss": 1.461,
+      "grad_norm": 10.229572296142578,
+      "learning_rate": 2.3879164313946926e-05,
+      "loss": 0.9884,
       "step": 5420
     },
     {
       "epoch": 1.0220214568040655,
-      "grad_norm": 19.5515079498291,
-      "learning_rate": 1.591191417278374e-05,
-      "loss": 1.7264,
+      "grad_norm": 10.10629653930664,
+      "learning_rate": 2.386787125917561e-05,
+      "loss": 1.2098,
       "step": 5430
     },
     {
       "epoch": 1.0239036325992847,
-      "grad_norm": 7.74648904800415,
-      "learning_rate": 1.5904385469602864e-05,
-      "loss": 1.4598,
+      "grad_norm": 8.695528030395508,
+      "learning_rate": 2.3856578204404294e-05,
+      "loss": 1.0261,
       "step": 5440
     },
     {
       "epoch": 1.025785808394504,
-      "grad_norm": 3.796482801437378,
-      "learning_rate": 1.5896856766421984e-05,
-      "loss": 1.501,
+      "grad_norm": 3.444990873336792,
+      "learning_rate": 2.3845285149632974e-05,
+      "loss": 1.0464,
       "step": 5450
     },
     {
       "epoch": 1.0276679841897234,
-      "grad_norm": 10.214385032653809,
-      "learning_rate": 1.5889328063241107e-05,
-      "loss": 1.5436,
+      "grad_norm": 6.334211826324463,
+      "learning_rate": 2.383399209486166e-05,
+      "loss": 1.1021,
       "step": 5460
     },
     {
       "epoch": 1.0295501599849426,
-      "grad_norm": 9.938682556152344,
-      "learning_rate": 1.588179936006023e-05,
-      "loss": 1.4925,
+      "grad_norm": 7.3270978927612305,
+      "learning_rate": 2.3822699040090346e-05,
+      "loss": 1.0433,
       "step": 5470
     },
     {
       "epoch": 1.0314323357801618,
-      "grad_norm": 9.411529541015625,
-      "learning_rate": 1.5874270656879352e-05,
-      "loss": 1.3498,
+      "grad_norm": 19.58551788330078,
+      "learning_rate": 2.381140598531903e-05,
+      "loss": 0.9511,
       "step": 5480
     },
     {
       "epoch": 1.0333145115753812,
-      "grad_norm": 5.797966003417969,
-      "learning_rate": 1.5866741953698475e-05,
-      "loss": 1.7169,
+      "grad_norm": 11.17794132232666,
+      "learning_rate": 2.3800112930547715e-05,
+      "loss": 1.3029,
       "step": 5490
     },
     {
       "epoch": 1.0351966873706004,
-      "grad_norm": 16.239484786987305,
-      "learning_rate": 1.58592132505176e-05,
-      "loss": 1.6468,
+      "grad_norm": 20.934467315673828,
+      "learning_rate": 2.3788819875776397e-05,
+      "loss": 1.2481,
       "step": 5500
     },
     {
       "epoch": 1.0370788631658197,
-      "grad_norm": 7.906765460968018,
-      "learning_rate": 1.5851684547336725e-05,
-      "loss": 1.4129,
+      "grad_norm": 5.4833083152771,
+      "learning_rate": 2.3777526821005083e-05,
+      "loss": 0.9677,
       "step": 5510
     },
     {
       "epoch": 1.0389610389610389,
-      "grad_norm": 23.242895126342773,
-      "learning_rate": 1.5844155844155847e-05,
-      "loss": 1.4746,
+      "grad_norm": 17.41068458557129,
+      "learning_rate": 2.376623376623377e-05,
+      "loss": 1.0862,
       "step": 5520
     },
     {
       "epoch": 1.0408432147562583,
-      "grad_norm": 11.383037567138672,
-      "learning_rate": 1.583662714097497e-05,
-      "loss": 1.4018,
+      "grad_norm": 10.463713645935059,
+      "learning_rate": 2.3754940711462452e-05,
+      "loss": 1.1165,
       "step": 5530
     },
     {
       "epoch": 1.0427253905514775,
-      "grad_norm": 5.408214092254639,
-      "learning_rate": 1.582909843779409e-05,
-      "loss": 1.2102,
+      "grad_norm": 4.4030232429504395,
+      "learning_rate": 2.3743647656691135e-05,
+      "loss": 0.8207,
       "step": 5540
     },
     {
       "epoch": 1.0446075663466967,
-      "grad_norm": 4.229701519012451,
-      "learning_rate": 1.5821569734613213e-05,
-      "loss": 1.2008,
+      "grad_norm": 8.17664909362793,
+      "learning_rate": 2.3732354601919818e-05,
+      "loss": 0.8548,
       "step": 5550
     },
     {
       "epoch": 1.0464897421419161,
-      "grad_norm": 10.835386276245117,
-      "learning_rate": 1.5814041031432336e-05,
-      "loss": 1.4252,
+      "grad_norm": 18.136306762695312,
+      "learning_rate": 2.3721061547148504e-05,
+      "loss": 1.0368,
       "step": 5560
     },
     {
       "epoch": 1.0483719179371354,
-      "grad_norm": 7.582424640655518,
-      "learning_rate": 1.580651232825146e-05,
-      "loss": 1.4494,
+      "grad_norm": 8.143134117126465,
+      "learning_rate": 2.370976849237719e-05,
+      "loss": 1.1101,
       "step": 5570
     },
     {
       "epoch": 1.0502540937323546,
-      "grad_norm": 22.497562408447266,
-      "learning_rate": 1.579898362507058e-05,
-      "loss": 1.6387,
+      "grad_norm": 17.662364959716797,
+      "learning_rate": 2.3698475437605872e-05,
+      "loss": 1.3467,
       "step": 5580
     },
     {
       "epoch": 1.0521362695275738,
-      "grad_norm": 3.5851354598999023,
-      "learning_rate": 1.5791454921889705e-05,
-      "loss": 1.2493,
+      "grad_norm": 18.27084732055664,
+      "learning_rate": 2.368718238283456e-05,
+      "loss": 0.7784,
       "step": 5590
     },
     {
       "epoch": 1.0540184453227932,
-      "grad_norm": 13.096941947937012,
-      "learning_rate": 1.578392621870883e-05,
-      "loss": 1.5457,
+      "grad_norm": 20.256084442138672,
+      "learning_rate": 2.367588932806324e-05,
+      "loss": 0.9596,
       "step": 5600
     },
     {
       "epoch": 1.0559006211180124,
-      "grad_norm": 4.804878234863281,
-      "learning_rate": 1.5776397515527954e-05,
-      "loss": 1.5518,
+      "grad_norm": 5.476850509643555,
+      "learning_rate": 2.3664596273291927e-05,
+      "loss": 1.2,
       "step": 5610
     },
     {
       "epoch": 1.0577827969132316,
-      "grad_norm": 7.501936912536621,
-      "learning_rate": 1.5768868812347073e-05,
-      "loss": 1.658,
+      "grad_norm": 13.74804973602295,
+      "learning_rate": 2.365330321852061e-05,
+      "loss": 1.3438,
       "step": 5620
     },
     {
       "epoch": 1.059664972708451,
-      "grad_norm": 8.500528335571289,
-      "learning_rate": 1.5761340109166196e-05,
-      "loss": 1.3127,
+      "grad_norm": 7.789999008178711,
+      "learning_rate": 2.3642010163749293e-05,
+      "loss": 0.8268,
       "step": 5630
     },
     {
       "epoch": 1.0615471485036703,
-      "grad_norm": 7.3146443367004395,
-      "learning_rate": 1.575381140598532e-05,
-      "loss": 1.257,
+      "grad_norm": 10.119312286376953,
+      "learning_rate": 2.363071710897798e-05,
+      "loss": 1.03,
       "step": 5640
     },
     {
       "epoch": 1.0634293242988895,
-      "grad_norm": 10.491667747497559,
-      "learning_rate": 1.5746282702804442e-05,
-      "loss": 1.4083,
+      "grad_norm": 17.798921585083008,
+      "learning_rate": 2.361942405420666e-05,
+      "loss": 0.9418,
       "step": 5650
     },
     {
       "epoch": 1.0653115000941087,
-      "grad_norm": 9.215764999389648,
-      "learning_rate": 1.5738753999623565e-05,
-      "loss": 1.3877,
+      "grad_norm": 16.53011703491211,
+      "learning_rate": 2.3608130999435347e-05,
+      "loss": 1.0546,
       "step": 5660
     },
     {
       "epoch": 1.0671936758893281,
-      "grad_norm": 8.09468936920166,
-      "learning_rate": 1.5731225296442688e-05,
-      "loss": 1.2621,
+      "grad_norm": 8.095433235168457,
+      "learning_rate": 2.3596837944664034e-05,
+      "loss": 0.9267,
       "step": 5670
     },
     {
       "epoch": 1.0690758516845473,
-      "grad_norm": 26.073856353759766,
-      "learning_rate": 1.572369659326181e-05,
-      "loss": 1.5949,
+      "grad_norm": 29.615936279296875,
+      "learning_rate": 2.3585544889892716e-05,
+      "loss": 1.2167,
       "step": 5680
     },
     {
       "epoch": 1.0709580274797665,
-      "grad_norm": 5.2231245040893555,
-      "learning_rate": 1.5716167890080937e-05,
-      "loss": 1.2968,
+      "grad_norm": 2.900878667831421,
+      "learning_rate": 2.3574251835121402e-05,
+      "loss": 0.9583,
       "step": 5690
     },
     {
       "epoch": 1.072840203274986,
-      "grad_norm": 3.045153856277466,
-      "learning_rate": 1.570863918690006e-05,
-      "loss": 1.2484,
+      "grad_norm": 2.4044950008392334,
+      "learning_rate": 2.356295878035009e-05,
+      "loss": 0.8682,
       "step": 5700
     },
     {
       "epoch": 1.0747223790702052,
-      "grad_norm": 2.576266288757324,
-      "learning_rate": 1.570111048371918e-05,
-      "loss": 1.4398,
+      "grad_norm": 3.0829226970672607,
+      "learning_rate": 2.3551665725578768e-05,
+      "loss": 0.9567,
       "step": 5710
     },
     {
       "epoch": 1.0766045548654244,
-      "grad_norm": 7.949709892272949,
-      "learning_rate": 1.5693581780538303e-05,
-      "loss": 1.3862,
+      "grad_norm": 8.497003555297852,
+      "learning_rate": 2.3540372670807454e-05,
+      "loss": 0.889,
       "step": 5720
     },
     {
       "epoch": 1.0784867306606438,
-      "grad_norm": 16.988584518432617,
-      "learning_rate": 1.5686053077357425e-05,
-      "loss": 1.5145,
+      "grad_norm": 5.439596652984619,
+      "learning_rate": 2.3529079616036136e-05,
+      "loss": 1.1877,
       "step": 5730
     },
     {
       "epoch": 1.080368906455863,
-      "grad_norm": 15.880838394165039,
-      "learning_rate": 1.567852437417655e-05,
-      "loss": 1.3245,
+      "grad_norm": 13.089673042297363,
+      "learning_rate": 2.3517786561264823e-05,
+      "loss": 0.9833,
       "step": 5740
     },
     {
       "epoch": 1.0822510822510822,
-      "grad_norm": 14.930453300476074,
-      "learning_rate": 1.567099567099567e-05,
-      "loss": 1.7517,
+      "grad_norm": 8.794281959533691,
+      "learning_rate": 2.350649350649351e-05,
+      "loss": 1.4184,
       "step": 5750
     },
     {
       "epoch": 1.0841332580463015,
-      "grad_norm": 9.984511375427246,
-      "learning_rate": 1.5663466967814794e-05,
-      "loss": 1.688,
+      "grad_norm": 11.838709831237793,
+      "learning_rate": 2.349520045172219e-05,
+      "loss": 1.2191,
       "step": 5760
     },
     {
       "epoch": 1.0860154338415209,
-      "grad_norm": 13.405725479125977,
-      "learning_rate": 1.5655938264633917e-05,
-      "loss": 1.4352,
+      "grad_norm": 12.768487930297852,
+      "learning_rate": 2.3483907396950877e-05,
+      "loss": 1.0469,
       "step": 5770
     },
     {
       "epoch": 1.08789760963674,
-      "grad_norm": 14.22274112701416,
-      "learning_rate": 1.5648409561453043e-05,
-      "loss": 1.592,
+      "grad_norm": 16.489809036254883,
+      "learning_rate": 2.347261434217956e-05,
+      "loss": 1.2401,
       "step": 5780
     },
     {
       "epoch": 1.0897797854319593,
-      "grad_norm": 15.278095245361328,
-      "learning_rate": 1.5640880858272166e-05,
-      "loss": 1.5002,
+      "grad_norm": 22.284759521484375,
+      "learning_rate": 2.3461321287408246e-05,
+      "loss": 1.1759,
       "step": 5790
     },
     {
       "epoch": 1.0916619612271785,
-      "grad_norm": 15.014613151550293,
-      "learning_rate": 1.5633352155091286e-05,
-      "loss": 1.5084,
+      "grad_norm": 22.32616424560547,
+      "learning_rate": 2.345002823263693e-05,
+      "loss": 1.0558,
       "step": 5800
     },
     {
       "epoch": 1.093544137022398,
-      "grad_norm": 10.296102523803711,
-      "learning_rate": 1.562582345191041e-05,
-      "loss": 1.4295,
+      "grad_norm": 6.583852767944336,
+      "learning_rate": 2.343873517786561e-05,
+      "loss": 1.1139,
       "step": 5810
     },
     {
       "epoch": 1.0954263128176172,
-      "grad_norm": 8.37751579284668,
-      "learning_rate": 1.5618294748729532e-05,
-      "loss": 1.3661,
+      "grad_norm": 10.06447982788086,
+      "learning_rate": 2.3427442123094298e-05,
+      "loss": 1.0589,
       "step": 5820
     },
     {
       "epoch": 1.0973084886128364,
-      "grad_norm": 5.940329551696777,
-      "learning_rate": 1.5610766045548655e-05,
-      "loss": 1.437,
+      "grad_norm": 9.685362815856934,
+      "learning_rate": 2.341614906832298e-05,
+      "loss": 1.0719,
       "step": 5830
     },
     {
       "epoch": 1.0991906644080558,
-      "grad_norm": 6.347326755523682,
-      "learning_rate": 1.5603237342367778e-05,
-      "loss": 1.2912,
+      "grad_norm": 10.641950607299805,
+      "learning_rate": 2.3404856013551666e-05,
+      "loss": 0.8788,
       "step": 5840
     },
     {
       "epoch": 1.101072840203275,
-      "grad_norm": 7.670511245727539,
-      "learning_rate": 1.55957086391869e-05,
-      "loss": 1.2584,
+      "grad_norm": 9.749815940856934,
+      "learning_rate": 2.3393562958780352e-05,
+      "loss": 0.9172,
       "step": 5850
     },
     {
       "epoch": 1.1029550159984942,
-      "grad_norm": 14.886139869689941,
-      "learning_rate": 1.5588179936006023e-05,
-      "loss": 1.5002,
+      "grad_norm": 15.738201141357422,
+      "learning_rate": 2.3382269904009035e-05,
+      "loss": 1.1344,
       "step": 5860
     },
     {
       "epoch": 1.1048371917937136,
-      "grad_norm": 9.98166275024414,
-      "learning_rate": 1.5580651232825146e-05,
-      "loss": 1.366,
+      "grad_norm": 5.296823024749756,
+      "learning_rate": 2.337097684923772e-05,
+      "loss": 0.9969,
       "step": 5870
     },
     {
       "epoch": 1.1067193675889329,
-      "grad_norm": 4.7431817054748535,
-      "learning_rate": 1.557312252964427e-05,
-      "loss": 1.5411,
+      "grad_norm": 9.7347993850708,
+      "learning_rate": 2.33596837944664e-05,
+      "loss": 1.2592,
       "step": 5880
     },
     {
       "epoch": 1.108601543384152,
-      "grad_norm": 12.848723411560059,
-      "learning_rate": 1.5565593826463392e-05,
-      "loss": 1.3226,
+      "grad_norm": 15.322550773620605,
+      "learning_rate": 2.3348390739695087e-05,
+      "loss": 0.8988,
       "step": 5890
     },
     {
       "epoch": 1.1104837191793713,
-      "grad_norm": 6.75590705871582,
-      "learning_rate": 1.5558065123282515e-05,
-      "loss": 1.4568,
+      "grad_norm": 8.814277648925781,
+      "learning_rate": 2.3337097684923773e-05,
+      "loss": 0.9639,
       "step": 5900
     },
     {
       "epoch": 1.1123658949745907,
-      "grad_norm": 8.309553146362305,
-      "learning_rate": 1.5550536420101638e-05,
-      "loss": 1.2237,
+      "grad_norm": 9.318452835083008,
+      "learning_rate": 2.3325804630152455e-05,
+      "loss": 0.9157,
       "step": 5910
     },
     {
       "epoch": 1.11424807076981,
-      "grad_norm": 46.17892837524414,
-      "learning_rate": 1.554300771692076e-05,
-      "loss": 1.3477,
+      "grad_norm": 9.616998672485352,
+      "learning_rate": 2.331451157538114e-05,
+      "loss": 0.9965,
       "step": 5920
     },
     {
       "epoch": 1.1161302465650291,
-      "grad_norm": 13.070061683654785,
-      "learning_rate": 1.5535479013739884e-05,
-      "loss": 1.3567,
+      "grad_norm": 20.399810791015625,
+      "learning_rate": 2.3303218520609827e-05,
+      "loss": 0.9574,
       "step": 5930
     },
     {
       "epoch": 1.1180124223602483,
-      "grad_norm": 19.65549659729004,
-      "learning_rate": 1.5527950310559007e-05,
-      "loss": 1.2769,
+      "grad_norm": 25.457443237304688,
+      "learning_rate": 2.329192546583851e-05,
+      "loss": 0.948,
       "step": 5940
     },
     {
       "epoch": 1.1198945981554678,
-      "grad_norm": 6.129930019378662,
-      "learning_rate": 1.552042160737813e-05,
-      "loss": 1.349,
+      "grad_norm": 10.880556106567383,
+      "learning_rate": 2.3280632411067196e-05,
+      "loss": 1.1425,
       "step": 5950
     },
     {
       "epoch": 1.121776773950687,
-      "grad_norm": 18.13608169555664,
-      "learning_rate": 1.5512892904197253e-05,
-      "loss": 1.6379,
+      "grad_norm": 26.9040584564209,
+      "learning_rate": 2.326933935629588e-05,
+      "loss": 1.2454,
       "step": 5960
     },
     {
       "epoch": 1.1236589497459062,
-      "grad_norm": 8.193785667419434,
-      "learning_rate": 1.5505364201016376e-05,
-      "loss": 1.3498,
+      "grad_norm": 12.784603118896484,
+      "learning_rate": 2.325804630152456e-05,
+      "loss": 1.0512,
       "step": 5970
     },
     {
       "epoch": 1.1255411255411256,
-      "grad_norm": 6.419458866119385,
-      "learning_rate": 1.54978354978355e-05,
-      "loss": 1.188,
+      "grad_norm": 6.434301853179932,
+      "learning_rate": 2.3246753246753248e-05,
+      "loss": 0.8505,
       "step": 5980
     },
     {
       "epoch": 1.1274233013363448,
-      "grad_norm": 8.998268127441406,
-      "learning_rate": 1.549030679465462e-05,
-      "loss": 1.42,
+      "grad_norm": 12.109904289245605,
+      "learning_rate": 2.323546019198193e-05,
+      "loss": 1.0467,
       "step": 5990
     },
     {
       "epoch": 1.129305477131564,
-      "grad_norm": 10.662075996398926,
-      "learning_rate": 1.5482778091473744e-05,
-      "loss": 1.0762,
+      "grad_norm": 11.11652660369873,
+      "learning_rate": 2.3224167137210616e-05,
+      "loss": 0.7246,
       "step": 6000
     },
     {
       "epoch": 1.1311876529267835,
-      "grad_norm": 12.713705062866211,
-      "learning_rate": 1.5475249388292867e-05,
-      "loss": 1.2355,
+      "grad_norm": 15.59671401977539,
+      "learning_rate": 2.32128740824393e-05,
+      "loss": 0.9292,
       "step": 6010
     },
     {
       "epoch": 1.1330698287220027,
-      "grad_norm": 7.3867268562316895,
-      "learning_rate": 1.546772068511199e-05,
-      "loss": 1.1358,
+      "grad_norm": 6.132266521453857,
+      "learning_rate": 2.3201581027667985e-05,
+      "loss": 0.721,
       "step": 6020
     },
     {
       "epoch": 1.134952004517222,
-      "grad_norm": 5.125856399536133,
-      "learning_rate": 1.5460191981931113e-05,
-      "loss": 1.3024,
+      "grad_norm": 3.5677640438079834,
+      "learning_rate": 2.319028797289667e-05,
+      "loss": 0.9246,
       "step": 6030
     },
     {
       "epoch": 1.136834180312441,
-      "grad_norm": 5.838713645935059,
-      "learning_rate": 1.5452663278750236e-05,
-      "loss": 1.2684,
+      "grad_norm": 15.319580078125,
+      "learning_rate": 2.3178994918125354e-05,
+      "loss": 0.9668,
       "step": 6040
     },
     {
       "epoch": 1.1387163561076605,
-      "grad_norm": 9.813702583312988,
-      "learning_rate": 1.544513457556936e-05,
-      "loss": 1.2061,
+      "grad_norm": 15.318434715270996,
+      "learning_rate": 2.316770186335404e-05,
+      "loss": 1.0329,
       "step": 6050
     },
     {
       "epoch": 1.1405985319028797,
-      "grad_norm": 5.236690998077393,
-      "learning_rate": 1.5437605872388482e-05,
-      "loss": 1.2684,
+      "grad_norm": 8.09421157836914,
+      "learning_rate": 2.315640880858272e-05,
+      "loss": 1.0839,
       "step": 6060
     },
     {
       "epoch": 1.142480707698099,
-      "grad_norm": 14.457210540771484,
-      "learning_rate": 1.5430077169207605e-05,
-      "loss": 1.3883,
+      "grad_norm": 33.64767837524414,
+      "learning_rate": 2.3145115753811405e-05,
+      "loss": 1.1178,
       "step": 6070
     },
     {
       "epoch": 1.1443628834933182,
-      "grad_norm": 29.406164169311523,
-      "learning_rate": 1.5422548466026728e-05,
-      "loss": 1.4574,
+      "grad_norm": 23.52187728881836,
+      "learning_rate": 2.313382269904009e-05,
+      "loss": 1.1034,
       "step": 6080
     },
     {
       "epoch": 1.1462450592885376,
-      "grad_norm": 17.454238891601562,
-      "learning_rate": 1.541501976284585e-05,
-      "loss": 1.2196,
+      "grad_norm": 13.548325538635254,
+      "learning_rate": 2.3122529644268774e-05,
+      "loss": 0.801,
       "step": 6090
     },
     {
       "epoch": 1.1481272350837568,
-      "grad_norm": 15.656752586364746,
-      "learning_rate": 1.5407491059664974e-05,
-      "loss": 1.204,
+      "grad_norm": 19.462448120117188,
+      "learning_rate": 2.311123658949746e-05,
+      "loss": 0.8789,
       "step": 6100
     },
     {
       "epoch": 1.150009410878976,
-      "grad_norm": 7.013121128082275,
-      "learning_rate": 1.5399962356484096e-05,
-      "loss": 1.4965,
+      "grad_norm": 32.65730285644531,
+      "learning_rate": 2.3099943534726143e-05,
+      "loss": 1.2276,
       "step": 6110
     },
     {
       "epoch": 1.1518915866741954,
-      "grad_norm": 2.7544403076171875,
-      "learning_rate": 1.539243365330322e-05,
-      "loss": 1.2982,
+      "grad_norm": 1.7889130115509033,
+      "learning_rate": 2.308865047995483e-05,
+      "loss": 0.9995,
       "step": 6120
     },
     {
       "epoch": 1.1537737624694147,
-      "grad_norm": 3.2940828800201416,
-      "learning_rate": 1.5384904950122342e-05,
-      "loss": 1.2389,
+      "grad_norm": 2.802178382873535,
+      "learning_rate": 2.3077357425183515e-05,
+      "loss": 0.9639,
       "step": 6130
     },
     {
       "epoch": 1.1556559382646339,
-      "grad_norm": 25.011899948120117,
-      "learning_rate": 1.5377376246941465e-05,
-      "loss": 1.2476,
+      "grad_norm": 30.473918914794922,
+      "learning_rate": 2.3066064370412198e-05,
+      "loss": 0.9189,
       "step": 6140
     },
     {
       "epoch": 1.1575381140598533,
-      "grad_norm": 8.474715232849121,
-      "learning_rate": 1.5369847543760588e-05,
-      "loss": 1.6291,
+      "grad_norm": 12.530424118041992,
+      "learning_rate": 2.305477131564088e-05,
+      "loss": 1.1765,
       "step": 6150
     },
     {
       "epoch": 1.1594202898550725,
-      "grad_norm": 18.783309936523438,
-      "learning_rate": 1.536231884057971e-05,
-      "loss": 1.364,
+      "grad_norm": 20.285593032836914,
+      "learning_rate": 2.3043478260869563e-05,
+      "loss": 1.0729,
       "step": 6160
     },
     {
       "epoch": 1.1613024656502917,
-      "grad_norm": 10.150487899780273,
-      "learning_rate": 1.5354790137398834e-05,
-      "loss": 1.3958,
+      "grad_norm": 17.06644058227539,
+      "learning_rate": 2.303218520609825e-05,
+      "loss": 1.1532,
       "step": 6170
     },
     {
       "epoch": 1.163184641445511,
-      "grad_norm": 7.357229232788086,
-      "learning_rate": 1.5347261434217957e-05,
-      "loss": 1.3219,
+      "grad_norm": 13.454843521118164,
+      "learning_rate": 2.3020892151326935e-05,
+      "loss": 1.0656,
       "step": 6180
     },
     {
       "epoch": 1.1650668172407304,
-      "grad_norm": 29.95680809020996,
-      "learning_rate": 1.533973273103708e-05,
-      "loss": 1.27,
+      "grad_norm": 14.98215103149414,
+      "learning_rate": 2.3009599096555618e-05,
+      "loss": 0.7957,
       "step": 6190
     },
     {
       "epoch": 1.1669489930359496,
-      "grad_norm": 14.412184715270996,
-      "learning_rate": 1.5332204027856203e-05,
-      "loss": 1.5809,
+      "grad_norm": 12.102943420410156,
+      "learning_rate": 2.2998306041784304e-05,
+      "loss": 1.3556,
       "step": 6200
     },
     {
       "epoch": 1.1688311688311688,
-      "grad_norm": 18.45595932006836,
-      "learning_rate": 1.5324675324675326e-05,
-      "loss": 1.3501,
+      "grad_norm": 10.94884204864502,
+      "learning_rate": 2.298701298701299e-05,
+      "loss": 1.0466,
       "step": 6210
     },
     {
       "epoch": 1.170713344626388,
-      "grad_norm": 6.867365837097168,
-      "learning_rate": 1.531714662149445e-05,
-      "loss": 1.3181,
+      "grad_norm": 5.2002177238464355,
+      "learning_rate": 2.2975719932241673e-05,
+      "loss": 1.1543,
       "step": 6220
     },
     {
       "epoch": 1.1725955204216074,
-      "grad_norm": 5.471830368041992,
-      "learning_rate": 1.530961791831357e-05,
-      "loss": 1.1032,
+      "grad_norm": 8.19458293914795,
+      "learning_rate": 2.2964426877470356e-05,
+      "loss": 0.7794,
       "step": 6230
     },
     {
       "epoch": 1.1744776962168266,
-      "grad_norm": 9.03870677947998,
-      "learning_rate": 1.5302089215132694e-05,
-      "loss": 1.1249,
+      "grad_norm": 7.3538970947265625,
+      "learning_rate": 2.2953133822699038e-05,
+      "loss": 0.7533,
       "step": 6240
     },
     {
       "epoch": 1.1763598720120458,
-      "grad_norm": 10.276759147644043,
-      "learning_rate": 1.5294560511951817e-05,
-      "loss": 1.4236,
+      "grad_norm": 13.170063018798828,
+      "learning_rate": 2.2941840767927724e-05,
+      "loss": 1.1663,
       "step": 6250
     },
     {
       "epoch": 1.1782420478072653,
-      "grad_norm": 5.121780872344971,
-      "learning_rate": 1.528703180877094e-05,
-      "loss": 1.3727,
+      "grad_norm": 22.807252883911133,
+      "learning_rate": 2.293054771315641e-05,
+      "loss": 0.9954,
       "step": 6260
     },
     {
       "epoch": 1.1801242236024845,
-      "grad_norm": 4.659458637237549,
-      "learning_rate": 1.5279503105590063e-05,
-      "loss": 1.2893,
+      "grad_norm": 4.545765399932861,
+      "learning_rate": 2.2919254658385093e-05,
+      "loss": 0.9198,
       "step": 6270
     },
     {
       "epoch": 1.1820063993977037,
-      "grad_norm": 33.238712310791016,
-      "learning_rate": 1.5271974402409186e-05,
-      "loss": 1.2643,
+      "grad_norm": 8.624770164489746,
+      "learning_rate": 2.290796160361378e-05,
+      "loss": 0.9429,
       "step": 6280
     },
     {
       "epoch": 1.1838885751929231,
-      "grad_norm": 13.528379440307617,
-      "learning_rate": 1.526444569922831e-05,
-      "loss": 1.4603,
+      "grad_norm": 12.10987377166748,
+      "learning_rate": 2.2896668548842462e-05,
+      "loss": 1.2487,
       "step": 6290
     },
     {
       "epoch": 1.1857707509881423,
-      "grad_norm": 5.262114524841309,
-      "learning_rate": 1.5256916996047434e-05,
-      "loss": 1.0799,
+      "grad_norm": 1.273734211921692,
+      "learning_rate": 2.2885375494071148e-05,
+      "loss": 0.9025,
       "step": 6300
     },
     {
       "epoch": 1.1876529267833615,
-      "grad_norm": 7.602123260498047,
-      "learning_rate": 1.5249388292866557e-05,
-      "loss": 1.3151,
+      "grad_norm": 6.115864276885986,
+      "learning_rate": 2.2874082439299834e-05,
+      "loss": 0.9436,
       "step": 6310
     },
     {
       "epoch": 1.1895351025785807,
-      "grad_norm": 6.674724578857422,
-      "learning_rate": 1.5241859589685676e-05,
-      "loss": 1.166,
+      "grad_norm": 6.128392696380615,
+      "learning_rate": 2.2862789384528513e-05,
+      "loss": 0.9464,
       "step": 6320
     },
     {
       "epoch": 1.1914172783738002,
-      "grad_norm": 8.049056053161621,
-      "learning_rate": 1.52343308865048e-05,
-      "loss": 1.217,
+      "grad_norm": 13.86414623260498,
+      "learning_rate": 2.28514963297572e-05,
+      "loss": 0.8451,
       "step": 6330
     },
     {
       "epoch": 1.1932994541690194,
-      "grad_norm": 12.107414245605469,
-      "learning_rate": 1.5226802183323924e-05,
-      "loss": 1.1538,
+      "grad_norm": 19.781795501708984,
+      "learning_rate": 2.2840203274985882e-05,
+      "loss": 0.7738,
       "step": 6340
     },
     {
       "epoch": 1.1951816299642386,
-      "grad_norm": 11.166797637939453,
-      "learning_rate": 1.5219273480143047e-05,
-      "loss": 1.3976,
+      "grad_norm": 30.369861602783203,
+      "learning_rate": 2.2828910220214568e-05,
+      "loss": 1.2183,
       "step": 6350
     },
     {
       "epoch": 1.1970638057594578,
-      "grad_norm": 6.7904953956604,
-      "learning_rate": 1.521174477696217e-05,
-      "loss": 1.5883,
+      "grad_norm": 12.43057918548584,
+      "learning_rate": 2.2817617165443254e-05,
+      "loss": 1.1899,
       "step": 6360
     },
     {
       "epoch": 1.1989459815546772,
-      "grad_norm": 9.514391899108887,
-      "learning_rate": 1.5204216073781292e-05,
-      "loss": 1.472,
+      "grad_norm": 9.697637557983398,
+      "learning_rate": 2.2806324110671937e-05,
+      "loss": 1.2109,
       "step": 6370
     },
     {
       "epoch": 1.2008281573498965,
-      "grad_norm": 29.685026168823242,
-      "learning_rate": 1.5196687370600415e-05,
-      "loss": 1.5798,
+      "grad_norm": 33.04664993286133,
+      "learning_rate": 2.2795031055900623e-05,
+      "loss": 1.2716,
       "step": 6380
     },
     {
       "epoch": 1.2027103331451157,
-      "grad_norm": 14.561470985412598,
-      "learning_rate": 1.5189158667419538e-05,
-      "loss": 1.4515,
+      "grad_norm": 16.485483169555664,
+      "learning_rate": 2.2783738001129306e-05,
+      "loss": 1.1587,
       "step": 6390
     },
     {
       "epoch": 1.204592508940335,
-      "grad_norm": 19.98004150390625,
-      "learning_rate": 1.5181629964238663e-05,
-      "loss": 1.3567,
+      "grad_norm": 8.360411643981934,
+      "learning_rate": 2.2772444946357992e-05,
+      "loss": 1.0419,
       "step": 6400
     },
     {
       "epoch": 1.2064746847355543,
-      "grad_norm": 10.195460319519043,
-      "learning_rate": 1.5174101261057782e-05,
-      "loss": 1.2189,
+      "grad_norm": 3.943709373474121,
+      "learning_rate": 2.2761151891586674e-05,
+      "loss": 1.0027,
       "step": 6410
     },
     {
       "epoch": 1.2083568605307735,
-      "grad_norm": 2.810354471206665,
-      "learning_rate": 1.5166572557876907e-05,
-      "loss": 1.4201,
+      "grad_norm": 17.139930725097656,
+      "learning_rate": 2.2749858836815357e-05,
+      "loss": 1.1939,
       "step": 6420
     },
     {
       "epoch": 1.210239036325993,
-      "grad_norm": 11.01855182647705,
-      "learning_rate": 1.515904385469603e-05,
-      "loss": 1.1615,
+      "grad_norm": 14.557121276855469,
+      "learning_rate": 2.2738565782044043e-05,
+      "loss": 0.8377,
       "step": 6430
     },
     {
       "epoch": 1.2121212121212122,
-      "grad_norm": 14.129963874816895,
-      "learning_rate": 1.5151515151515153e-05,
-      "loss": 1.2612,
+      "grad_norm": 17.73250961303711,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.9496,
       "step": 6440
     },
     {
       "epoch": 1.2140033879164314,
-      "grad_norm": 11.1683349609375,
-      "learning_rate": 1.5143986448334276e-05,
-      "loss": 1.3002,
+      "grad_norm": 15.301102638244629,
+      "learning_rate": 2.2715979672501412e-05,
+      "loss": 1.0668,
       "step": 6450
     },
     {
       "epoch": 1.2158855637116506,
-      "grad_norm": 5.463107585906982,
-      "learning_rate": 1.5136457745153399e-05,
-      "loss": 1.2832,
+      "grad_norm": 15.799238204956055,
+      "learning_rate": 2.2704686617730098e-05,
+      "loss": 1.0151,
       "step": 6460
     },
     {
       "epoch": 1.21776773950687,
-      "grad_norm": 11.25309944152832,
-      "learning_rate": 1.5128929041972522e-05,
-      "loss": 1.2142,
+      "grad_norm": 20.844924926757812,
+      "learning_rate": 2.269339356295878e-05,
+      "loss": 0.9301,
       "step": 6470
     },
     {
       "epoch": 1.2196499153020892,
-      "grad_norm": 8.699621200561523,
-      "learning_rate": 1.5121400338791645e-05,
-      "loss": 1.0461,
+      "grad_norm": 7.686192512512207,
+      "learning_rate": 2.2682100508187467e-05,
+      "loss": 0.797,
       "step": 6480
     },
     {
       "epoch": 1.2215320910973084,
-      "grad_norm": 16.165090560913086,
-      "learning_rate": 1.5113871635610766e-05,
-      "loss": 1.2469,
+      "grad_norm": 15.735981941223145,
+      "learning_rate": 2.267080745341615e-05,
+      "loss": 1.0817,
       "step": 6490
     },
     {
       "epoch": 1.2234142668925279,
-      "grad_norm": 25.876081466674805,
-      "learning_rate": 1.5106342932429889e-05,
-      "loss": 1.4,
+      "grad_norm": 10.269287109375,
+      "learning_rate": 2.2659514398644832e-05,
+      "loss": 1.1608,
       "step": 6500
     },
     {
       "epoch": 1.225296442687747,
-      "grad_norm": 10.362266540527344,
-      "learning_rate": 1.5098814229249013e-05,
-      "loss": 1.0884,
+      "grad_norm": 28.582277297973633,
+      "learning_rate": 2.2648221343873518e-05,
+      "loss": 0.8283,
       "step": 6510
     },
     {
       "epoch": 1.2271786184829663,
-      "grad_norm": 4.939754486083984,
-      "learning_rate": 1.5091285526068136e-05,
-      "loss": 1.5001,
+      "grad_norm": 8.514233589172363,
+      "learning_rate": 2.26369282891022e-05,
+      "loss": 1.2402,
       "step": 6520
     },
     {
       "epoch": 1.2290607942781855,
-      "grad_norm": 8.243157386779785,
-      "learning_rate": 1.5083756822887259e-05,
-      "loss": 1.3486,
+      "grad_norm": 9.531684875488281,
+      "learning_rate": 2.2625635234330887e-05,
+      "loss": 1.1151,
       "step": 6530
     },
     {
       "epoch": 1.230942970073405,
-      "grad_norm": 5.043724060058594,
-      "learning_rate": 1.5076228119706382e-05,
-      "loss": 1.1551,
+      "grad_norm": 10.337912559509277,
+      "learning_rate": 2.2614342179559573e-05,
+      "loss": 0.8757,
       "step": 6540
     },
     {
       "epoch": 1.2328251458686241,
-      "grad_norm": 5.159477233886719,
-      "learning_rate": 1.5068699416525505e-05,
-      "loss": 1.0274,
+      "grad_norm": 8.955284118652344,
+      "learning_rate": 2.2603049124788256e-05,
+      "loss": 0.76,
       "step": 6550
     },
     {
       "epoch": 1.2347073216638433,
-      "grad_norm": 5.960069179534912,
-      "learning_rate": 1.5061170713344628e-05,
-      "loss": 1.232,
+      "grad_norm": 14.72435474395752,
+      "learning_rate": 2.2591756070016942e-05,
+      "loss": 0.9185,
       "step": 6560
     },
     {
       "epoch": 1.2365894974590628,
-      "grad_norm": 5.939128875732422,
-      "learning_rate": 1.505364201016375e-05,
-      "loss": 1.6197,
+      "grad_norm": 16.09954071044922,
+      "learning_rate": 2.2580463015245625e-05,
+      "loss": 1.313,
       "step": 6570
     },
     {
       "epoch": 1.238471673254282,
-      "grad_norm": 13.313936233520508,
-      "learning_rate": 1.5046113306982872e-05,
-      "loss": 1.1685,
+      "grad_norm": 8.912398338317871,
+      "learning_rate": 2.2569169960474307e-05,
+      "loss": 0.8678,
       "step": 6580
     },
     {
       "epoch": 1.2403538490495012,
-      "grad_norm": 4.995264053344727,
-      "learning_rate": 1.5038584603801995e-05,
-      "loss": 1.2126,
+      "grad_norm": 8.538116455078125,
+      "learning_rate": 2.2557876905702993e-05,
+      "loss": 1.0072,
       "step": 6590
     },
     {
       "epoch": 1.2422360248447206,
-      "grad_norm": 19.458560943603516,
-      "learning_rate": 1.5031055900621118e-05,
-      "loss": 0.9409,
+      "grad_norm": 14.951505661010742,
+      "learning_rate": 2.2546583850931676e-05,
+      "loss": 0.6922,
       "step": 6600
     },
     {
       "epoch": 1.2441182006399398,
-      "grad_norm": 11.867140769958496,
-      "learning_rate": 1.5023527197440243e-05,
-      "loss": 1.1042,
+      "grad_norm": 8.427967071533203,
+      "learning_rate": 2.2535290796160362e-05,
+      "loss": 0.6708,
       "step": 6610
     },
     {
       "epoch": 1.246000376435159,
-      "grad_norm": 7.560145854949951,
-      "learning_rate": 1.5015998494259365e-05,
-      "loss": 1.399,
+      "grad_norm": 9.497509002685547,
+      "learning_rate": 2.2523997741389045e-05,
+      "loss": 1.0899,
       "step": 6620
     },
     {
       "epoch": 1.2478825522303783,
-      "grad_norm": 19.83662986755371,
-      "learning_rate": 1.5008469791078488e-05,
-      "loss": 1.2062,
+      "grad_norm": 30.798381805419922,
+      "learning_rate": 2.251270468661773e-05,
+      "loss": 0.9245,
       "step": 6630
     },
     {
       "epoch": 1.2497647280255977,
-      "grad_norm": 11.589738845825195,
-      "learning_rate": 1.5000941087897611e-05,
-      "loss": 1.2568,
+      "grad_norm": 12.764777183532715,
+      "learning_rate": 2.2501411631846417e-05,
+      "loss": 0.8884,
       "step": 6640
     },
     {
       "epoch": 1.251646903820817,
-      "grad_norm": 14.702427864074707,
-      "learning_rate": 1.4993412384716734e-05,
-      "loss": 1.1121,
+      "grad_norm": 6.89249324798584,
+      "learning_rate": 2.24901185770751e-05,
+      "loss": 0.9125,
       "step": 6650
     },
     {
       "epoch": 1.253529079616036,
-      "grad_norm": 10.332629203796387,
-      "learning_rate": 1.4985883681535857e-05,
-      "loss": 1.0371,
+      "grad_norm": 9.787991523742676,
+      "learning_rate": 2.2478825522303786e-05,
+      "loss": 0.8764,
       "step": 6660
     },
     {
       "epoch": 1.2554112554112553,
-      "grad_norm": 11.859658241271973,
-      "learning_rate": 1.4978354978354978e-05,
-      "loss": 1.3456,
+      "grad_norm": 20.8765811920166,
+      "learning_rate": 2.246753246753247e-05,
+      "loss": 1.1334,
       "step": 6670
     },
     {
       "epoch": 1.2572934312064747,
-      "grad_norm": 14.336831092834473,
-      "learning_rate": 1.4970826275174101e-05,
-      "loss": 1.5135,
+      "grad_norm": 19.523963928222656,
+      "learning_rate": 2.245623941276115e-05,
+      "loss": 1.2794,
       "step": 6680
     },
     {
       "epoch": 1.259175607001694,
-      "grad_norm": 21.42241096496582,
-      "learning_rate": 1.4963297571993224e-05,
-      "loss": 1.1505,
+      "grad_norm": 17.773313522338867,
+      "learning_rate": 2.2444946357989837e-05,
+      "loss": 0.8465,
       "step": 6690
     },
     {
       "epoch": 1.2610577827969132,
-      "grad_norm": 10.227978706359863,
-      "learning_rate": 1.4955768868812349e-05,
-      "loss": 1.4948,
+      "grad_norm": 11.96570110321045,
+      "learning_rate": 2.243365330321852e-05,
+      "loss": 1.1552,
       "step": 6700
     },
     {
       "epoch": 1.2629399585921326,
-      "grad_norm": 11.469473838806152,
-      "learning_rate": 1.4948240165631472e-05,
-      "loss": 1.2275,
+      "grad_norm": 8.867982864379883,
+      "learning_rate": 2.2422360248447206e-05,
+      "loss": 1.011,
       "step": 6710
     },
     {
       "epoch": 1.2648221343873518,
-      "grad_norm": 31.51825714111328,
-      "learning_rate": 1.4940711462450595e-05,
-      "loss": 1.1679,
+      "grad_norm": 29.349536895751953,
+      "learning_rate": 2.2411067193675892e-05,
+      "loss": 0.8635,
       "step": 6720
     },
     {
       "epoch": 1.266704310182571,
-      "grad_norm": 8.267791748046875,
-      "learning_rate": 1.4933182759269718e-05,
-      "loss": 1.1554,
+      "grad_norm": 12.906034469604492,
+      "learning_rate": 2.2399774138904575e-05,
+      "loss": 0.8193,
       "step": 6730
     },
     {
       "epoch": 1.2685864859777904,
-      "grad_norm": 12.076141357421875,
-      "learning_rate": 1.492565405608884e-05,
-      "loss": 1.441,
+      "grad_norm": 12.88680648803711,
+      "learning_rate": 2.238848108413326e-05,
+      "loss": 1.264,
       "step": 6740
     },
     {
       "epoch": 1.2704686617730097,
-      "grad_norm": 8.693479537963867,
-      "learning_rate": 1.4918125352907963e-05,
-      "loss": 1.0406,
+      "grad_norm": 5.127486228942871,
+      "learning_rate": 2.2377188029361943e-05,
+      "loss": 0.839,
       "step": 6750
     },
     {
       "epoch": 1.2723508375682289,
-      "grad_norm": 5.5710368156433105,
-      "learning_rate": 1.4910596649727085e-05,
-      "loss": 1.2251,
+      "grad_norm": 5.232704162597656,
+      "learning_rate": 2.2365894974590626e-05,
+      "loss": 0.9379,
       "step": 6760
     },
     {
       "epoch": 1.274233013363448,
-      "grad_norm": 19.289039611816406,
-      "learning_rate": 1.4903067946546208e-05,
-      "loss": 1.3377,
+      "grad_norm": 38.552825927734375,
+      "learning_rate": 2.2354601919819312e-05,
+      "loss": 0.9256,
       "step": 6770
     },
     {
       "epoch": 1.2761151891586673,
-      "grad_norm": 14.665457725524902,
-      "learning_rate": 1.489553924336533e-05,
-      "loss": 1.2305,
+      "grad_norm": 27.74937629699707,
+      "learning_rate": 2.2343308865047995e-05,
+      "loss": 0.9923,
       "step": 6780
     },
     {
       "epoch": 1.2779973649538867,
-      "grad_norm": 6.723300933837891,
-      "learning_rate": 1.4888010540184455e-05,
-      "loss": 1.3505,
+      "grad_norm": 26.892642974853516,
+      "learning_rate": 2.233201581027668e-05,
+      "loss": 1.0955,
       "step": 6790
     },
     {
       "epoch": 1.279879540749106,
-      "grad_norm": 165.9130859375,
-      "learning_rate": 1.4880481837003578e-05,
-      "loss": 1.1513,
+      "grad_norm": 11.914067268371582,
+      "learning_rate": 2.2320722755505364e-05,
+      "loss": 0.9401,
       "step": 6800
     },
     {
       "epoch": 1.2817617165443251,
-      "grad_norm": 11.876227378845215,
-      "learning_rate": 1.4872953133822701e-05,
-      "loss": 1.1086,
+      "grad_norm": 16.04292869567871,
+      "learning_rate": 2.230942970073405e-05,
+      "loss": 0.9384,
       "step": 6810
     },
     {
       "epoch": 1.2836438923395446,
-      "grad_norm": 5.059904098510742,
-      "learning_rate": 1.4865424430641824e-05,
-      "loss": 1.1162,
+      "grad_norm": 7.591270923614502,
+      "learning_rate": 2.2298136645962736e-05,
+      "loss": 0.8969,
       "step": 6820
     },
     {
       "epoch": 1.2855260681347638,
-      "grad_norm": 39.77632522583008,
-      "learning_rate": 1.4857895727460947e-05,
-      "loss": 1.1485,
+      "grad_norm": 23.051708221435547,
+      "learning_rate": 2.228684359119142e-05,
+      "loss": 0.7347,
       "step": 6830
     },
     {
       "epoch": 1.287408243929983,
-      "grad_norm": 17.867996215820312,
-      "learning_rate": 1.4850367024280068e-05,
-      "loss": 1.0449,
+      "grad_norm": 19.278038024902344,
+      "learning_rate": 2.22755505364201e-05,
+      "loss": 0.873,
       "step": 6840
     },
     {
       "epoch": 1.2892904197252024,
-      "grad_norm": 4.805659294128418,
-      "learning_rate": 1.4842838321099191e-05,
-      "loss": 1.1365,
+      "grad_norm": 6.090950012207031,
+      "learning_rate": 2.2264257481648784e-05,
+      "loss": 0.8029,
       "step": 6850
     },
     {
       "epoch": 1.2911725955204216,
-      "grad_norm": 4.266025543212891,
-      "learning_rate": 1.4835309617918314e-05,
-      "loss": 1.4389,
+      "grad_norm": 7.98006010055542,
+      "learning_rate": 2.225296442687747e-05,
+      "loss": 1.2243,
       "step": 6860
     },
     {
       "epoch": 1.2930547713156408,
-      "grad_norm": 6.8787455558776855,
-      "learning_rate": 1.4827780914737437e-05,
-      "loss": 1.2298,
+      "grad_norm": 10.448835372924805,
+      "learning_rate": 2.2241671372106156e-05,
+      "loss": 1.0374,
       "step": 6870
     },
     {
       "epoch": 1.2949369471108603,
-      "grad_norm": 8.046527862548828,
-      "learning_rate": 1.4820252211556561e-05,
-      "loss": 1.1355,
+      "grad_norm": 8.826786041259766,
+      "learning_rate": 2.223037831733484e-05,
+      "loss": 0.9484,
       "step": 6880
     },
     {
       "epoch": 1.2968191229060795,
-      "grad_norm": 13.009222984313965,
-      "learning_rate": 1.4812723508375684e-05,
-      "loss": 1.2157,
+      "grad_norm": 7.370842933654785,
+      "learning_rate": 2.2219085262563525e-05,
+      "loss": 1.0126,
       "step": 6890
     },
     {
       "epoch": 1.2987012987012987,
-      "grad_norm": 16.49818992614746,
-      "learning_rate": 1.4805194805194807e-05,
-      "loss": 1.4171,
+      "grad_norm": 18.180156707763672,
+      "learning_rate": 2.2207792207792207e-05,
+      "loss": 1.2361,
       "step": 6900
     },
     {
       "epoch": 1.300583474496518,
-      "grad_norm": 19.328983306884766,
-      "learning_rate": 1.479766610201393e-05,
-      "loss": 1.4212,
+      "grad_norm": 11.186609268188477,
+      "learning_rate": 2.2196499153020894e-05,
+      "loss": 1.0604,
       "step": 6910
     },
     {
       "epoch": 1.3024656502917373,
-      "grad_norm": 19.792497634887695,
-      "learning_rate": 1.4790137398833053e-05,
-      "loss": 1.0885,
+      "grad_norm": 20.87862777709961,
+      "learning_rate": 2.218520609824958e-05,
+      "loss": 0.8857,
       "step": 6920
     },
     {
       "epoch": 1.3043478260869565,
-      "grad_norm": 8.283934593200684,
-      "learning_rate": 1.4782608695652174e-05,
-      "loss": 1.0667,
+      "grad_norm": 12.313833236694336,
+      "learning_rate": 2.217391304347826e-05,
+      "loss": 0.9329,
       "step": 6930
     },
     {
       "epoch": 1.3062300018821758,
-      "grad_norm": 5.211995601654053,
-      "learning_rate": 1.4775079992471297e-05,
-      "loss": 1.1669,
+      "grad_norm": 9.278630256652832,
+      "learning_rate": 2.2162619988706945e-05,
+      "loss": 0.9219,
       "step": 6940
     },
     {
       "epoch": 1.308112177677395,
-      "grad_norm": 20.76984214782715,
-      "learning_rate": 1.476755128929042e-05,
-      "loss": 1.5103,
+      "grad_norm": 14.707159996032715,
+      "learning_rate": 2.215132693393563e-05,
+      "loss": 1.227,
       "step": 6950
     },
     {
       "epoch": 1.3099943534726144,
-      "grad_norm": 45.730899810791016,
-      "learning_rate": 1.4760022586109543e-05,
-      "loss": 1.0752,
+      "grad_norm": 31.15422248840332,
+      "learning_rate": 2.2140033879164314e-05,
+      "loss": 0.7853,
       "step": 6960
     },
     {
       "epoch": 1.3118765292678336,
-      "grad_norm": 8.765107154846191,
-      "learning_rate": 1.4752493882928666e-05,
-      "loss": 1.2052,
+      "grad_norm": 8.061625480651855,
+      "learning_rate": 2.2128740824393e-05,
+      "loss": 0.9793,
       "step": 6970
     },
     {
       "epoch": 1.3137587050630528,
-      "grad_norm": 46.306983947753906,
-      "learning_rate": 1.474496517974779e-05,
-      "loss": 1.4834,
+      "grad_norm": 9.95386791229248,
+      "learning_rate": 2.2117447769621683e-05,
+      "loss": 1.2524,
       "step": 6980
     },
     {
       "epoch": 1.3156408808582722,
-      "grad_norm": 9.06339168548584,
-      "learning_rate": 1.4737436476566914e-05,
-      "loss": 1.137,
+      "grad_norm": 6.791532516479492,
+      "learning_rate": 2.210615471485037e-05,
+      "loss": 0.8785,
       "step": 6990
     },
     {
       "epoch": 1.3175230566534915,
-      "grad_norm": 12.679614067077637,
-      "learning_rate": 1.4729907773386036e-05,
-      "loss": 1.1793,
+      "grad_norm": 7.024257659912109,
+      "learning_rate": 2.2094861660079055e-05,
+      "loss": 0.9782,
       "step": 7000
     },
     {
       "epoch": 1.3194052324487107,
-      "grad_norm": 7.356766223907471,
-      "learning_rate": 1.472237907020516e-05,
-      "loss": 1.1861,
+      "grad_norm": 7.559739589691162,
+      "learning_rate": 2.2083568605307737e-05,
+      "loss": 0.9149,
       "step": 7010
     },
     {
       "epoch": 1.32128740824393,
-      "grad_norm": 17.430709838867188,
-      "learning_rate": 1.471485036702428e-05,
-      "loss": 1.0291,
+      "grad_norm": 8.2528076171875,
+      "learning_rate": 2.207227555053642e-05,
+      "loss": 0.7117,
       "step": 7020
     },
     {
       "epoch": 1.3231695840391493,
-      "grad_norm": 11.125081062316895,
-      "learning_rate": 1.4707321663843404e-05,
-      "loss": 1.2338,
+      "grad_norm": 21.831268310546875,
+      "learning_rate": 2.2060982495765103e-05,
+      "loss": 0.9922,
       "step": 7030
     },
     {
       "epoch": 1.3250517598343685,
-      "grad_norm": 17.22634506225586,
-      "learning_rate": 1.4699792960662526e-05,
-      "loss": 1.2741,
+      "grad_norm": 16.495553970336914,
+      "learning_rate": 2.204968944099379e-05,
+      "loss": 1.0543,
       "step": 7040
     },
     {
       "epoch": 1.3269339356295877,
-      "grad_norm": 37.80469512939453,
-      "learning_rate": 1.469226425748165e-05,
-      "loss": 1.2436,
+      "grad_norm": 10.989031791687012,
+      "learning_rate": 2.2038396386222475e-05,
+      "loss": 0.9902,
       "step": 7050
     },
     {
       "epoch": 1.3288161114248072,
-      "grad_norm": 7.4153971672058105,
-      "learning_rate": 1.4684735554300772e-05,
-      "loss": 1.0263,
+      "grad_norm": 10.473587036132812,
+      "learning_rate": 2.2027103331451158e-05,
+      "loss": 0.8303,
       "step": 7060
     },
     {
       "epoch": 1.3306982872200264,
-      "grad_norm": 19.874597549438477,
-      "learning_rate": 1.4677206851119897e-05,
-      "loss": 1.142,
+      "grad_norm": 11.674604415893555,
+      "learning_rate": 2.2015810276679844e-05,
+      "loss": 0.79,
       "step": 7070
     },
     {
       "epoch": 1.3325804630152456,
-      "grad_norm": 18.514259338378906,
-      "learning_rate": 1.466967814793902e-05,
-      "loss": 1.3683,
+      "grad_norm": 8.529352188110352,
+      "learning_rate": 2.2004517221908526e-05,
+      "loss": 1.1502,
       "step": 7080
     },
     {
       "epoch": 1.3344626388104648,
-      "grad_norm": 25.53579330444336,
-      "learning_rate": 1.4662149444758143e-05,
-      "loss": 1.3415,
+      "grad_norm": 14.45956802368164,
+      "learning_rate": 2.1993224167137212e-05,
+      "loss": 1.0809,
       "step": 7090
     },
     {
       "epoch": 1.3363448146056842,
-      "grad_norm": 17.759546279907227,
-      "learning_rate": 1.4654620741577264e-05,
-      "loss": 1.6721,
+      "grad_norm": 7.735544681549072,
+      "learning_rate": 2.1981931112365895e-05,
+      "loss": 1.4952,
       "step": 7100
     },
     {
       "epoch": 1.3382269904009034,
-      "grad_norm": 7.8694915771484375,
-      "learning_rate": 1.4647092038396387e-05,
-      "loss": 1.0611,
+      "grad_norm": 4.9393134117126465,
+      "learning_rate": 2.1970638057594578e-05,
+      "loss": 0.8674,
       "step": 7110
     },
     {
       "epoch": 1.3401091661961226,
-      "grad_norm": 16.278396606445312,
-      "learning_rate": 1.463956333521551e-05,
-      "loss": 1.1083,
+      "grad_norm": 15.033855438232422,
+      "learning_rate": 2.1959345002823264e-05,
+      "loss": 0.8618,
       "step": 7120
     },
     {
       "epoch": 1.341991341991342,
-      "grad_norm": 7.207310676574707,
-      "learning_rate": 1.4632034632034633e-05,
-      "loss": 0.996,
+      "grad_norm": 6.540989398956299,
+      "learning_rate": 2.1948051948051947e-05,
+      "loss": 0.8127,
       "step": 7130
     },
     {
       "epoch": 1.3438735177865613,
-      "grad_norm": 10.18614673614502,
-      "learning_rate": 1.4624505928853756e-05,
-      "loss": 1.2492,
+      "grad_norm": 5.148128032684326,
+      "learning_rate": 2.1936758893280633e-05,
+      "loss": 0.8892,
       "step": 7140
     },
     {
       "epoch": 1.3457556935817805,
-      "grad_norm": 11.438788414001465,
-      "learning_rate": 1.4616977225672879e-05,
-      "loss": 1.2819,
+      "grad_norm": 54.40917205810547,
+      "learning_rate": 2.192546583850932e-05,
+      "loss": 1.0773,
       "step": 7150
     },
     {
       "epoch": 1.347637869377,
-      "grad_norm": 14.186439514160156,
-      "learning_rate": 1.4609448522492003e-05,
-      "loss": 1.2256,
+      "grad_norm": 20.81314468383789,
+      "learning_rate": 2.1914172783738e-05,
+      "loss": 1.0652,
       "step": 7160
     },
     {
       "epoch": 1.3495200451722191,
-      "grad_norm": 9.92445182800293,
-      "learning_rate": 1.4601919819311126e-05,
-      "loss": 1.1,
+      "grad_norm": 7.903687000274658,
+      "learning_rate": 2.1902879728966687e-05,
+      "loss": 0.8812,
       "step": 7170
     },
     {
       "epoch": 1.3514022209674383,
-      "grad_norm": 10.874330520629883,
-      "learning_rate": 1.4594391116130249e-05,
-      "loss": 1.3286,
+      "grad_norm": 13.962177276611328,
+      "learning_rate": 2.189158667419537e-05,
+      "loss": 1.0603,
       "step": 7180
     },
     {
       "epoch": 1.3532843967626578,
-      "grad_norm": 4.418787956237793,
-      "learning_rate": 1.458686241294937e-05,
-      "loss": 1.4054,
+      "grad_norm": 4.5833892822265625,
+      "learning_rate": 2.1880293619424053e-05,
+      "loss": 1.2734,
       "step": 7190
     },
     {
       "epoch": 1.355166572557877,
-      "grad_norm": 17.510101318359375,
-      "learning_rate": 1.4579333709768493e-05,
-      "loss": 1.1078,
+      "grad_norm": 13.545757293701172,
+      "learning_rate": 2.186900056465274e-05,
+      "loss": 0.9052,
       "step": 7200
     },
     {
       "epoch": 1.3570487483530962,
-      "grad_norm": 10.508594512939453,
-      "learning_rate": 1.4571805006587616e-05,
-      "loss": 1.1085,
+      "grad_norm": 31.69056510925293,
+      "learning_rate": 2.185770750988142e-05,
+      "loss": 0.8867,
       "step": 7210
     },
     {
       "epoch": 1.3589309241483154,
-      "grad_norm": 14.85457706451416,
-      "learning_rate": 1.4564276303406739e-05,
-      "loss": 1.1159,
+      "grad_norm": 13.554387092590332,
+      "learning_rate": 2.1846414455110108e-05,
+      "loss": 0.8744,
       "step": 7220
     },
     {
       "epoch": 1.3608130999435346,
-      "grad_norm": 5.535599231719971,
-      "learning_rate": 1.4556747600225862e-05,
-      "loss": 1.1347,
+      "grad_norm": 4.856804370880127,
+      "learning_rate": 2.1835121400338794e-05,
+      "loss": 0.8372,
       "step": 7230
     },
     {
       "epoch": 1.362695275738754,
-      "grad_norm": 9.540611267089844,
-      "learning_rate": 1.4549218897044985e-05,
-      "loss": 1.1789,
+      "grad_norm": 12.263862609863281,
+      "learning_rate": 2.1823828345567476e-05,
+      "loss": 1.0886,
       "step": 7240
     },
     {
       "epoch": 1.3645774515339733,
-      "grad_norm": 18.45024299621582,
-      "learning_rate": 1.4541690193864108e-05,
-      "loss": 1.0515,
+      "grad_norm": 18.752655029296875,
+      "learning_rate": 2.1812535290796163e-05,
+      "loss": 0.7373,
       "step": 7250
     },
     {
       "epoch": 1.3664596273291925,
-      "grad_norm": 79.7477035522461,
-      "learning_rate": 1.4534161490683232e-05,
-      "loss": 1.1787,
+      "grad_norm": 22.475677490234375,
+      "learning_rate": 2.1801242236024845e-05,
+      "loss": 1.0194,
       "step": 7260
     },
     {
       "epoch": 1.368341803124412,
-      "grad_norm": 3.590022087097168,
-      "learning_rate": 1.4526632787502355e-05,
-      "loss": 1.2198,
+      "grad_norm": 3.7865233421325684,
+      "learning_rate": 2.178994918125353e-05,
+      "loss": 0.9387,
       "step": 7270
     },
     {
       "epoch": 1.370223978919631,
-      "grad_norm": 4.568202018737793,
-      "learning_rate": 1.4519104084321477e-05,
-      "loss": 1.32,
+      "grad_norm": 23.85765266418457,
+      "learning_rate": 2.1778656126482214e-05,
+      "loss": 1.1156,
       "step": 7280
     },
     {
       "epoch": 1.3721061547148503,
-      "grad_norm": 12.585639953613281,
-      "learning_rate": 1.45115753811406e-05,
-      "loss": 1.1478,
+      "grad_norm": 13.753911972045898,
+      "learning_rate": 2.1767363071710897e-05,
+      "loss": 0.89,
       "step": 7290
     },
     {
       "epoch": 1.3739883305100697,
-      "grad_norm": 7.819540500640869,
-      "learning_rate": 1.4504046677959722e-05,
-      "loss": 1.3724,
+      "grad_norm": 13.338386535644531,
+      "learning_rate": 2.1756070016939583e-05,
+      "loss": 1.0544,
       "step": 7300
     },
     {
       "epoch": 1.375870506305289,
-      "grad_norm": 14.75216293334961,
-      "learning_rate": 1.4496517974778845e-05,
-      "loss": 1.2092,
+      "grad_norm": 7.914046287536621,
+      "learning_rate": 2.1744776962168265e-05,
+      "loss": 0.9427,
       "step": 7310
     },
     {
       "epoch": 1.3777526821005082,
-      "grad_norm": 25.505807876586914,
-      "learning_rate": 1.4488989271597968e-05,
-      "loss": 1.2517,
+      "grad_norm": 33.587493896484375,
+      "learning_rate": 2.173348390739695e-05,
+      "loss": 1.066,
       "step": 7320
     },
     {
       "epoch": 1.3796348578957276,
-      "grad_norm": 5.387889862060547,
-      "learning_rate": 1.4481460568417091e-05,
-      "loss": 0.861,
+      "grad_norm": 5.1027326583862305,
+      "learning_rate": 2.1722190852625638e-05,
+      "loss": 0.6101,
       "step": 7330
     },
     {
       "epoch": 1.3815170336909468,
-      "grad_norm": 2.2768774032592773,
-      "learning_rate": 1.4473931865236214e-05,
-      "loss": 1.1739,
+      "grad_norm": 0.7987704277038574,
+      "learning_rate": 2.171089779785432e-05,
+      "loss": 1.1217,
       "step": 7340
     },
     {
       "epoch": 1.383399209486166,
-      "grad_norm": 10.858841896057129,
-      "learning_rate": 1.4466403162055339e-05,
-      "loss": 1.217,
+      "grad_norm": 9.396214485168457,
+      "learning_rate": 2.1699604743083006e-05,
+      "loss": 1.0526,
       "step": 7350
     },
     {
       "epoch": 1.3852813852813852,
-      "grad_norm": 21.92953872680664,
-      "learning_rate": 1.4458874458874458e-05,
-      "loss": 1.0472,
+      "grad_norm": 21.794408798217773,
+      "learning_rate": 2.1688311688311686e-05,
+      "loss": 0.8303,
       "step": 7360
     },
     {
       "epoch": 1.3871635610766044,
-      "grad_norm": 10.877849578857422,
-      "learning_rate": 1.4451345755693583e-05,
-      "loss": 1.199,
+      "grad_norm": 22.232450485229492,
+      "learning_rate": 2.1677018633540372e-05,
+      "loss": 1.1084,
       "step": 7370
     },
     {
       "epoch": 1.3890457368718239,
-      "grad_norm": 7.3010149002075195,
-      "learning_rate": 1.4443817052512706e-05,
-      "loss": 1.3164,
+      "grad_norm": 7.189990520477295,
+      "learning_rate": 2.1665725578769058e-05,
+      "loss": 1.1183,
       "step": 7380
     },
     {
       "epoch": 1.390927912667043,
-      "grad_norm": 3.245511293411255,
-      "learning_rate": 1.4436288349331829e-05,
-      "loss": 0.9705,
+      "grad_norm": 4.046782970428467,
+      "learning_rate": 2.165443252399774e-05,
+      "loss": 0.6394,
       "step": 7390
     },
     {
       "epoch": 1.3928100884622623,
-      "grad_norm": 4.765195846557617,
-      "learning_rate": 1.4428759646150952e-05,
-      "loss": 1.0409,
+      "grad_norm": 6.194625377655029,
+      "learning_rate": 2.1643139469226427e-05,
+      "loss": 0.8828,
       "step": 7400
     },
     {
       "epoch": 1.3946922642574817,
-      "grad_norm": 9.110795021057129,
-      "learning_rate": 1.4421230942970075e-05,
-      "loss": 1.0267,
+      "grad_norm": 15.320564270019531,
+      "learning_rate": 2.163184641445511e-05,
+      "loss": 0.8554,
       "step": 7410
     },
     {
       "epoch": 1.396574440052701,
-      "grad_norm": 11.141672134399414,
-      "learning_rate": 1.4413702239789197e-05,
-      "loss": 1.0451,
+      "grad_norm": 9.800172805786133,
+      "learning_rate": 2.1620553359683795e-05,
+      "loss": 0.9175,
       "step": 7420
     },
     {
       "epoch": 1.3984566158479201,
-      "grad_norm": 3.296548843383789,
-      "learning_rate": 1.440617353660832e-05,
-      "loss": 1.0522,
+      "grad_norm": 4.258259296417236,
+      "learning_rate": 2.160926030491248e-05,
+      "loss": 0.872,
       "step": 7430
     },
     {
       "epoch": 1.4003387916431396,
-      "grad_norm": 11.544855117797852,
-      "learning_rate": 1.4398644833427445e-05,
-      "loss": 1.097,
+      "grad_norm": 16.88814353942871,
+      "learning_rate": 2.1597967250141164e-05,
+      "loss": 0.9596,
       "step": 7440
     },
     {
       "epoch": 1.4022209674383588,
-      "grad_norm": 9.588610649108887,
-      "learning_rate": 1.4391116130246565e-05,
-      "loss": 1.0146,
+      "grad_norm": 11.031537055969238,
+      "learning_rate": 2.1586674195369847e-05,
+      "loss": 0.8293,
       "step": 7450
     },
     {
       "epoch": 1.404103143233578,
-      "grad_norm": 17.363183975219727,
-      "learning_rate": 1.4383587427065689e-05,
-      "loss": 1.2154,
+      "grad_norm": 20.199750900268555,
+      "learning_rate": 2.1575381140598533e-05,
+      "loss": 0.9311,
       "step": 7460
     },
     {
       "epoch": 1.4059853190287974,
-      "grad_norm": 6.3251495361328125,
-      "learning_rate": 1.4376058723884812e-05,
-      "loss": 1.0566,
+      "grad_norm": 10.844975471496582,
+      "learning_rate": 2.1564088085827216e-05,
+      "loss": 0.8357,
       "step": 7470
     },
     {
       "epoch": 1.4078674948240166,
-      "grad_norm": 2.8543460369110107,
-      "learning_rate": 1.4368530020703935e-05,
-      "loss": 1.1234,
+      "grad_norm": 26.47841453552246,
+      "learning_rate": 2.15527950310559e-05,
+      "loss": 0.9514,
       "step": 7480
     },
     {
       "epoch": 1.4097496706192358,
-      "grad_norm": 41.35597229003906,
-      "learning_rate": 1.4361001317523058e-05,
-      "loss": 1.4789,
+      "grad_norm": 27.28598976135254,
+      "learning_rate": 2.1541501976284584e-05,
+      "loss": 1.2359,
       "step": 7490
     },
     {
       "epoch": 1.411631846414455,
-      "grad_norm": 8.920171737670898,
-      "learning_rate": 1.435347261434218e-05,
-      "loss": 1.0124,
+      "grad_norm": 22.267330169677734,
+      "learning_rate": 2.153020892151327e-05,
+      "loss": 0.7749,
       "step": 7500
     },
     {
       "epoch": 1.4135140222096743,
-      "grad_norm": 9.12064266204834,
-      "learning_rate": 1.4345943911161304e-05,
-      "loss": 0.9119,
+      "grad_norm": 10.783427238464355,
+      "learning_rate": 2.1518915866741956e-05,
+      "loss": 0.7608,
       "step": 7510
     },
     {
       "epoch": 1.4153961980048937,
-      "grad_norm": 51.954063415527344,
-      "learning_rate": 1.4338415207980427e-05,
-      "loss": 1.0014,
+      "grad_norm": 12.085504531860352,
+      "learning_rate": 2.150762281197064e-05,
+      "loss": 0.8435,
       "step": 7520
     },
     {
       "epoch": 1.417278373800113,
-      "grad_norm": 4.927883148193359,
-      "learning_rate": 1.4330886504799551e-05,
-      "loss": 1.3197,
+      "grad_norm": 4.772538185119629,
+      "learning_rate": 2.1496329757199325e-05,
+      "loss": 1.1515,
       "step": 7530
     },
     {
       "epoch": 1.419160549595332,
-      "grad_norm": 11.021320343017578,
-      "learning_rate": 1.432335780161867e-05,
-      "loss": 0.9996,
+      "grad_norm": 16.843612670898438,
+      "learning_rate": 2.1485036702428005e-05,
+      "loss": 0.6556,
       "step": 7540
     },
     {
       "epoch": 1.4210427253905515,
-      "grad_norm": 8.294415473937988,
-      "learning_rate": 1.4315829098437794e-05,
-      "loss": 1.1363,
+      "grad_norm": 16.49220085144043,
+      "learning_rate": 2.147374364765669e-05,
+      "loss": 0.9657,
       "step": 7550
     },
     {
       "epoch": 1.4229249011857708,
-      "grad_norm": 12.463217735290527,
-      "learning_rate": 1.4308300395256918e-05,
-      "loss": 1.1546,
+      "grad_norm": 9.660463333129883,
+      "learning_rate": 2.1462450592885377e-05,
+      "loss": 1.0375,
       "step": 7560
     },
     {
       "epoch": 1.42480707698099,
-      "grad_norm": 5.810076713562012,
-      "learning_rate": 1.4300771692076041e-05,
-      "loss": 1.3381,
+      "grad_norm": 14.508869171142578,
+      "learning_rate": 2.145115753811406e-05,
+      "loss": 1.191,
       "step": 7570
     },
     {
       "epoch": 1.4266892527762094,
-      "grad_norm": 7.27411413192749,
-      "learning_rate": 1.4293242988895164e-05,
-      "loss": 1.237,
+      "grad_norm": 9.467730522155762,
+      "learning_rate": 2.1439864483342745e-05,
+      "loss": 0.911,
       "step": 7580
     },
     {
       "epoch": 1.4285714285714286,
-      "grad_norm": 8.202913284301758,
-      "learning_rate": 1.4285714285714287e-05,
-      "loss": 0.8892,
+      "grad_norm": 9.366085052490234,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.6016,
       "step": 7590
     },
     {
       "epoch": 1.4304536043666478,
-      "grad_norm": 6.186138153076172,
-      "learning_rate": 1.427818558253341e-05,
-      "loss": 1.3076,
+      "grad_norm": 6.722076892852783,
+      "learning_rate": 2.1417278373800114e-05,
+      "loss": 1.1092,
       "step": 7600
     },
     {
       "epoch": 1.4323357801618672,
-      "grad_norm": 16.254478454589844,
-      "learning_rate": 1.4270656879352533e-05,
-      "loss": 1.0453,
+      "grad_norm": 7.0276336669921875,
+      "learning_rate": 2.14059853190288e-05,
+      "loss": 0.8131,
       "step": 7610
     },
     {
       "epoch": 1.4342179559570865,
-      "grad_norm": 5.887212753295898,
-      "learning_rate": 1.4263128176171656e-05,
-      "loss": 1.0952,
+      "grad_norm": 9.224479675292969,
+      "learning_rate": 2.1394692264257483e-05,
+      "loss": 0.8081,
       "step": 7620
     },
     {
       "epoch": 1.4361001317523057,
-      "grad_norm": 58.455604553222656,
-      "learning_rate": 1.4255599472990777e-05,
-      "loss": 1.1879,
+      "grad_norm": 18.966449737548828,
+      "learning_rate": 2.1383399209486166e-05,
+      "loss": 0.8604,
       "step": 7630
     },
     {
       "epoch": 1.4379823075475249,
-      "grad_norm": 6.455393314361572,
-      "learning_rate": 1.42480707698099e-05,
-      "loss": 0.881,
+      "grad_norm": 6.464701175689697,
+      "learning_rate": 2.137210615471485e-05,
+      "loss": 0.7825,
       "step": 7640
     },
     {
       "epoch": 1.439864483342744,
-      "grad_norm": 6.461804389953613,
-      "learning_rate": 1.4240542066629025e-05,
-      "loss": 0.9318,
+      "grad_norm": 31.871877670288086,
+      "learning_rate": 2.1360813099943534e-05,
+      "loss": 0.7266,
       "step": 7650
     },
     {
       "epoch": 1.4417466591379635,
-      "grad_norm": 1.1490024328231812,
-      "learning_rate": 1.4233013363448148e-05,
-      "loss": 1.0784,
+      "grad_norm": 0.4453635513782501,
+      "learning_rate": 2.134952004517222e-05,
+      "loss": 0.9575,
       "step": 7660
     },
     {
       "epoch": 1.4436288349331827,
-      "grad_norm": 22.495023727416992,
-      "learning_rate": 1.422548466026727e-05,
-      "loss": 1.086,
+      "grad_norm": 20.737749099731445,
+      "learning_rate": 2.1338226990400903e-05,
+      "loss": 0.8743,
       "step": 7670
     },
     {
       "epoch": 1.445511010728402,
-      "grad_norm": 9.839690208435059,
-      "learning_rate": 1.4217955957086393e-05,
-      "loss": 0.9114,
+      "grad_norm": 7.961155414581299,
+      "learning_rate": 2.132693393562959e-05,
+      "loss": 0.8414,
       "step": 7680
     },
     {
       "epoch": 1.4473931865236214,
-      "grad_norm": 9.664743423461914,
-      "learning_rate": 1.4210427253905516e-05,
-      "loss": 1.1183,
+      "grad_norm": 14.688976287841797,
+      "learning_rate": 2.1315640880858275e-05,
+      "loss": 0.8901,
       "step": 7690
     },
     {
       "epoch": 1.4492753623188406,
-      "grad_norm": 2.7624919414520264,
-      "learning_rate": 1.420289855072464e-05,
-      "loss": 1.0288,
+      "grad_norm": 1.112459421157837,
+      "learning_rate": 2.1304347826086958e-05,
+      "loss": 0.7949,
       "step": 7700
     },
     {
       "epoch": 1.4511575381140598,
-      "grad_norm": 2.4246561527252197,
-      "learning_rate": 1.419536984754376e-05,
-      "loss": 1.0188,
+      "grad_norm": 1.5727367401123047,
+      "learning_rate": 2.129305477131564e-05,
+      "loss": 0.834,
       "step": 7710
     },
     {
       "epoch": 1.4530397139092792,
-      "grad_norm": 7.706203937530518,
-      "learning_rate": 1.4187841144362883e-05,
-      "loss": 1.0697,
+      "grad_norm": 15.447263717651367,
+      "learning_rate": 2.1281761716544323e-05,
+      "loss": 0.8197,
       "step": 7720
     },
     {
       "epoch": 1.4549218897044984,
-      "grad_norm": 14.255431175231934,
-      "learning_rate": 1.4180312441182006e-05,
-      "loss": 1.0956,
+      "grad_norm": 16.137331008911133,
+      "learning_rate": 2.127046866177301e-05,
+      "loss": 0.7958,
       "step": 7730
     },
     {
       "epoch": 1.4568040654997176,
-      "grad_norm": 12.368314743041992,
-      "learning_rate": 1.4172783738001131e-05,
-      "loss": 1.0669,
+      "grad_norm": 10.978745460510254,
+      "learning_rate": 2.1259175607001696e-05,
+      "loss": 1.0626,
       "step": 7740
     },
     {
       "epoch": 1.458686241294937,
-      "grad_norm": 11.548198699951172,
-      "learning_rate": 1.4165255034820254e-05,
-      "loss": 1.0574,
+      "grad_norm": 13.138747215270996,
+      "learning_rate": 2.1247882552230378e-05,
+      "loss": 0.8053,
       "step": 7750
     },
     {
       "epoch": 1.4605684170901563,
-      "grad_norm": 8.626538276672363,
-      "learning_rate": 1.4157726331639377e-05,
-      "loss": 1.1298,
+      "grad_norm": 15.29321575164795,
+      "learning_rate": 2.1236589497459064e-05,
+      "loss": 1.0308,
       "step": 7760
     },
     {
       "epoch": 1.4624505928853755,
-      "grad_norm": 7.213776111602783,
-      "learning_rate": 1.41501976284585e-05,
-      "loss": 1.2538,
+      "grad_norm": 7.9823527336120605,
+      "learning_rate": 2.1225296442687747e-05,
+      "loss": 1.0307,
       "step": 7770
     },
     {
       "epoch": 1.4643327686805947,
-      "grad_norm": 28.54906463623047,
-      "learning_rate": 1.4142668925277623e-05,
-      "loss": 1.009,
+      "grad_norm": 28.051477432250977,
+      "learning_rate": 2.1214003387916433e-05,
+      "loss": 0.9783,
       "step": 7780
     },
     {
       "epoch": 1.466214944475814,
-      "grad_norm": 7.344525337219238,
-      "learning_rate": 1.4135140222096746e-05,
-      "loss": 0.8592,
+      "grad_norm": 2.81699800491333,
+      "learning_rate": 2.120271033314512e-05,
+      "loss": 0.6847,
       "step": 7790
     },
     {
       "epoch": 1.4680971202710333,
-      "grad_norm": 14.066884994506836,
-      "learning_rate": 1.4127611518915867e-05,
-      "loss": 0.9045,
+      "grad_norm": 20.243371963500977,
+      "learning_rate": 2.11914172783738e-05,
+      "loss": 0.8305,
       "step": 7800
     },
     {
       "epoch": 1.4699792960662525,
-      "grad_norm": 7.654214382171631,
-      "learning_rate": 1.412008281573499e-05,
-      "loss": 0.9722,
+      "grad_norm": 5.7048115730285645,
+      "learning_rate": 2.1180124223602485e-05,
+      "loss": 0.6723,
       "step": 7810
     },
     {
       "epoch": 1.4718614718614718,
-      "grad_norm": 4.771880149841309,
-      "learning_rate": 1.4112554112554113e-05,
-      "loss": 1.0675,
+      "grad_norm": 14.926733016967773,
+      "learning_rate": 2.1168831168831167e-05,
+      "loss": 0.9217,
       "step": 7820
     },
     {
       "epoch": 1.4737436476566912,
-      "grad_norm": 12.305994987487793,
-      "learning_rate": 1.4105025409373236e-05,
-      "loss": 1.1632,
+      "grad_norm": 22.269062042236328,
+      "learning_rate": 2.1157538114059853e-05,
+      "loss": 1.0226,
       "step": 7830
     },
     {
       "epoch": 1.4756258234519104,
-      "grad_norm": 4.854207992553711,
-      "learning_rate": 1.409749670619236e-05,
-      "loss": 0.9186,
+      "grad_norm": 5.699794292449951,
+      "learning_rate": 2.114624505928854e-05,
+      "loss": 0.749,
       "step": 7840
     },
     {
       "epoch": 1.4775079992471296,
-      "grad_norm": 11.940600395202637,
-      "learning_rate": 1.4089968003011483e-05,
-      "loss": 1.0449,
+      "grad_norm": 5.858451843261719,
+      "learning_rate": 2.1134952004517222e-05,
+      "loss": 0.7048,
       "step": 7850
     },
     {
       "epoch": 1.479390175042349,
-      "grad_norm": 9.252134323120117,
-      "learning_rate": 1.4082439299830606e-05,
-      "loss": 1.0849,
+      "grad_norm": 8.59745979309082,
+      "learning_rate": 2.1123658949745908e-05,
+      "loss": 0.9316,
       "step": 7860
     },
     {
       "epoch": 1.4812723508375683,
-      "grad_norm": 91.1706771850586,
-      "learning_rate": 1.4074910596649729e-05,
-      "loss": 1.1134,
+      "grad_norm": 12.093297958374023,
+      "learning_rate": 2.111236589497459e-05,
+      "loss": 1.0178,
       "step": 7870
     },
     {
       "epoch": 1.4831545266327875,
-      "grad_norm": 8.043750762939453,
-      "learning_rate": 1.4067381893468852e-05,
-      "loss": 1.0423,
+      "grad_norm": 12.393393516540527,
+      "learning_rate": 2.1101072840203277e-05,
+      "loss": 0.7238,
       "step": 7880
     },
     {
       "epoch": 1.485036702428007,
-      "grad_norm": 72.7489242553711,
-      "learning_rate": 1.4059853190287973e-05,
-      "loss": 0.9289,
+      "grad_norm": 46.78508758544922,
+      "learning_rate": 2.108977978543196e-05,
+      "loss": 0.8131,
       "step": 7890
     },
     {
       "epoch": 1.486918878223226,
-      "grad_norm": 2.386556625366211,
-      "learning_rate": 1.4052324487107096e-05,
-      "loss": 1.0292,
+      "grad_norm": 1.7145179510116577,
+      "learning_rate": 2.1078486730660642e-05,
+      "loss": 0.7799,
       "step": 7900
     },
     {
       "epoch": 1.4888010540184453,
-      "grad_norm": 25.438966751098633,
-      "learning_rate": 1.4044795783926219e-05,
-      "loss": 1.1286,
+      "grad_norm": 24.29248809814453,
+      "learning_rate": 2.106719367588933e-05,
+      "loss": 1.083,
       "step": 7910
     },
     {
       "epoch": 1.4906832298136645,
-      "grad_norm": 16.999435424804688,
-      "learning_rate": 1.4037267080745342e-05,
-      "loss": 1.0221,
+      "grad_norm": 10.044913291931152,
+      "learning_rate": 2.105590062111801e-05,
+      "loss": 0.7995,
       "step": 7920
     },
     {
       "epoch": 1.4925654056088837,
-      "grad_norm": 14.521089553833008,
-      "learning_rate": 1.4029738377564466e-05,
-      "loss": 1.3035,
+      "grad_norm": 17.686159133911133,
+      "learning_rate": 2.1044607566346697e-05,
+      "loss": 1.23,
       "step": 7930
     },
     {
       "epoch": 1.4944475814041032,
-      "grad_norm": 13.352957725524902,
-      "learning_rate": 1.402220967438359e-05,
-      "loss": 1.3361,
+      "grad_norm": 12.747879981994629,
+      "learning_rate": 2.1033314511575383e-05,
+      "loss": 1.0773,
       "step": 7940
     },
     {
       "epoch": 1.4963297571993224,
-      "grad_norm": 24.51966094970703,
-      "learning_rate": 1.4014680971202712e-05,
-      "loss": 1.0674,
+      "grad_norm": 14.406421661376953,
+      "learning_rate": 2.1022021456804066e-05,
+      "loss": 0.8235,
       "step": 7950
     },
     {
       "epoch": 1.4982119329945416,
-      "grad_norm": 4.013449668884277,
-      "learning_rate": 1.4007152268021835e-05,
-      "loss": 0.7399,
+      "grad_norm": 9.100472450256348,
+      "learning_rate": 2.1010728402032752e-05,
+      "loss": 0.5385,
       "step": 7960
     },
     {
       "epoch": 1.500094108789761,
-      "grad_norm": 8.218099594116211,
-      "learning_rate": 1.3999623564840956e-05,
-      "loss": 0.8721,
+      "grad_norm": 1.3265353441238403,
+      "learning_rate": 2.0999435347261435e-05,
+      "loss": 0.7218,
       "step": 7970
     },
     {
       "epoch": 1.5019762845849802,
-      "grad_norm": 15.218042373657227,
-      "learning_rate": 1.399209486166008e-05,
-      "loss": 1.145,
+      "grad_norm": 22.250818252563477,
+      "learning_rate": 2.0988142292490117e-05,
+      "loss": 0.9737,
       "step": 7980
     },
     {
       "epoch": 1.5038584603801994,
-      "grad_norm": 16.717836380004883,
-      "learning_rate": 1.3984566158479202e-05,
-      "loss": 0.9064,
+      "grad_norm": 6.612005710601807,
+      "learning_rate": 2.0976849237718803e-05,
+      "loss": 0.7305,
       "step": 7990
     },
     {
       "epoch": 1.5057406361754189,
-      "grad_norm": 61.43841552734375,
-      "learning_rate": 1.3977037455298325e-05,
-      "loss": 0.9818,
+      "grad_norm": 26.889253616333008,
+      "learning_rate": 2.0965556182947486e-05,
+      "loss": 0.82,
       "step": 8000
     },
     {
       "epoch": 1.507622811970638,
-      "grad_norm": 20.666854858398438,
-      "learning_rate": 1.3969508752117448e-05,
-      "loss": 0.9895,
+      "grad_norm": 15.2909574508667,
+      "learning_rate": 2.0954263128176172e-05,
+      "loss": 0.8488,
       "step": 8010
     },
     {
       "epoch": 1.5095049877658573,
-      "grad_norm": 36.14076614379883,
-      "learning_rate": 1.3961980048936573e-05,
-      "loss": 1.021,
+      "grad_norm": 14.072726249694824,
+      "learning_rate": 2.0942970073404858e-05,
+      "loss": 0.7917,
       "step": 8020
     },
     {
       "epoch": 1.5113871635610767,
-      "grad_norm": 18.77960777282715,
-      "learning_rate": 1.3954451345755696e-05,
-      "loss": 0.8724,
+      "grad_norm": 19.601593017578125,
+      "learning_rate": 2.093167701863354e-05,
+      "loss": 0.8319,
       "step": 8030
     },
     {
       "epoch": 1.513269339356296,
-      "grad_norm": 6.997078895568848,
-      "learning_rate": 1.3946922642574819e-05,
-      "loss": 1.0927,
+      "grad_norm": 11.248756408691406,
+      "learning_rate": 2.0920383963862227e-05,
+      "loss": 0.8455,
       "step": 8040
     },
     {
       "epoch": 1.5151515151515151,
-      "grad_norm": 12.119707107543945,
-      "learning_rate": 1.3939393939393942e-05,
-      "loss": 0.9838,
+      "grad_norm": 14.29757022857666,
+      "learning_rate": 2.090909090909091e-05,
+      "loss": 0.836,
       "step": 8050
     },
     {
       "epoch": 1.5170336909467346,
-      "grad_norm": 28.478229522705078,
-      "learning_rate": 1.3931865236213063e-05,
-      "loss": 1.0145,
+      "grad_norm": 18.773252487182617,
+      "learning_rate": 2.0897797854319592e-05,
+      "loss": 0.7697,
       "step": 8060
     },
     {
       "epoch": 1.5189158667419536,
-      "grad_norm": 23.44718360900879,
-      "learning_rate": 1.3924336533032186e-05,
-      "loss": 0.9658,
+      "grad_norm": 13.430651664733887,
+      "learning_rate": 2.088650479954828e-05,
+      "loss": 0.7545,
       "step": 8070
     },
     {
       "epoch": 1.520798042537173,
-      "grad_norm": 11.981337547302246,
-      "learning_rate": 1.3916807829851309e-05,
-      "loss": 1.0584,
+      "grad_norm": 18.43965721130371,
+      "learning_rate": 2.087521174477696e-05,
+      "loss": 0.9061,
       "step": 8080
     },
     {
       "epoch": 1.5226802183323922,
-      "grad_norm": 9.942193984985352,
-      "learning_rate": 1.3909279126670432e-05,
-      "loss": 0.985,
+      "grad_norm": 4.812375068664551,
+      "learning_rate": 2.0863918690005647e-05,
+      "loss": 0.7921,
       "step": 8090
     },
     {
       "epoch": 1.5245623941276114,
-      "grad_norm": 6.533130645751953,
-      "learning_rate": 1.3901750423489554e-05,
-      "loss": 1.1664,
+      "grad_norm": 13.704950332641602,
+      "learning_rate": 2.085262563523433e-05,
+      "loss": 1.0409,
       "step": 8100
     },
     {
       "epoch": 1.5264445699228308,
-      "grad_norm": 4.729006290435791,
-      "learning_rate": 1.3894221720308679e-05,
-      "loss": 1.0228,
+      "grad_norm": 7.7468414306640625,
+      "learning_rate": 2.0841332580463016e-05,
+      "loss": 0.8707,
       "step": 8110
     },
     {
       "epoch": 1.52832674571805,
-      "grad_norm": 15.206283569335938,
-      "learning_rate": 1.3886693017127802e-05,
-      "loss": 1.2411,
+      "grad_norm": 8.92183780670166,
+      "learning_rate": 2.0830039525691702e-05,
+      "loss": 1.0549,
       "step": 8120
     },
     {
       "epoch": 1.5302089215132693,
-      "grad_norm": 16.714759826660156,
-      "learning_rate": 1.3879164313946925e-05,
-      "loss": 1.2209,
+      "grad_norm": 12.238114356994629,
+      "learning_rate": 2.0818746470920385e-05,
+      "loss": 1.0467,
       "step": 8130
     },
     {
       "epoch": 1.5320910973084887,
-      "grad_norm": 20.130807876586914,
-      "learning_rate": 1.3871635610766048e-05,
-      "loss": 1.0768,
+      "grad_norm": 15.757767677307129,
+      "learning_rate": 2.080745341614907e-05,
+      "loss": 0.8356,
       "step": 8140
     },
     {
       "epoch": 1.533973273103708,
-      "grad_norm": 4.701301097869873,
-      "learning_rate": 1.3864106907585169e-05,
-      "loss": 0.9803,
+      "grad_norm": 15.579939842224121,
+      "learning_rate": 2.079616036137775e-05,
+      "loss": 0.7802,
       "step": 8150
     },
     {
       "epoch": 1.5358554488989271,
-      "grad_norm": 23.33013153076172,
-      "learning_rate": 1.3856578204404292e-05,
-      "loss": 0.9926,
+      "grad_norm": 7.932995796203613,
+      "learning_rate": 2.0784867306606436e-05,
+      "loss": 0.782,
       "step": 8160
     },
     {
       "epoch": 1.5377376246941465,
-      "grad_norm": 17.344905853271484,
-      "learning_rate": 1.3849049501223415e-05,
-      "loss": 1.2403,
+      "grad_norm": 34.61065673828125,
+      "learning_rate": 2.0773574251835122e-05,
+      "loss": 1.0706,
       "step": 8170
     },
     {
       "epoch": 1.5396198004893658,
-      "grad_norm": 4.836775302886963,
-      "learning_rate": 1.3841520798042538e-05,
-      "loss": 1.0559,
+      "grad_norm": 6.347330093383789,
+      "learning_rate": 2.0762281197063805e-05,
+      "loss": 0.862,
       "step": 8180
     },
     {
       "epoch": 1.541501976284585,
-      "grad_norm": 13.495597839355469,
-      "learning_rate": 1.383399209486166e-05,
-      "loss": 0.995,
+      "grad_norm": 12.102984428405762,
+      "learning_rate": 2.075098814229249e-05,
+      "loss": 0.976,
       "step": 8190
     },
     {
       "epoch": 1.5433841520798044,
-      "grad_norm": 57.71432113647461,
-      "learning_rate": 1.3826463391680784e-05,
-      "loss": 1.0178,
+      "grad_norm": 34.155494689941406,
+      "learning_rate": 2.0739695087521177e-05,
+      "loss": 0.9479,
       "step": 8200
     },
     {
       "epoch": 1.5452663278750234,
-      "grad_norm": 16.587297439575195,
-      "learning_rate": 1.3818934688499908e-05,
-      "loss": 0.8892,
+      "grad_norm": 4.237630844116211,
+      "learning_rate": 2.072840203274986e-05,
+      "loss": 0.6754,
       "step": 8210
     },
     {
       "epoch": 1.5471485036702428,
-      "grad_norm": 10.470789909362793,
-      "learning_rate": 1.3811405985319031e-05,
-      "loss": 0.8105,
+      "grad_norm": 9.868550300598145,
+      "learning_rate": 2.0717108977978546e-05,
+      "loss": 0.6415,
       "step": 8220
     },
     {
       "epoch": 1.549030679465462,
-      "grad_norm": 8.34557056427002,
-      "learning_rate": 1.3803877282138154e-05,
-      "loss": 1.214,
+      "grad_norm": 9.84210205078125,
+      "learning_rate": 2.070581592320723e-05,
+      "loss": 1.0436,
       "step": 8230
     },
     {
       "epoch": 1.5509128552606812,
-      "grad_norm": 15.27405071258545,
-      "learning_rate": 1.3796348578957275e-05,
-      "loss": 0.8583,
+      "grad_norm": 6.024266242980957,
+      "learning_rate": 2.069452286843591e-05,
+      "loss": 0.749,
       "step": 8240
     },
     {
       "epoch": 1.5527950310559007,
-      "grad_norm": 10.931471824645996,
-      "learning_rate": 1.3788819875776398e-05,
-      "loss": 1.2385,
+      "grad_norm": 24.58380889892578,
+      "learning_rate": 2.0683229813664597e-05,
+      "loss": 1.102,
       "step": 8250
     },
     {
       "epoch": 1.5546772068511199,
-      "grad_norm": 14.83597469329834,
-      "learning_rate": 1.3781291172595521e-05,
-      "loss": 0.8043,
+      "grad_norm": 15.549361228942871,
+      "learning_rate": 2.067193675889328e-05,
+      "loss": 0.619,
       "step": 8260
     },
     {
       "epoch": 1.556559382646339,
-      "grad_norm": 13.999913215637207,
-      "learning_rate": 1.3773762469414644e-05,
-      "loss": 0.8869,
+      "grad_norm": 11.536001205444336,
+      "learning_rate": 2.0660643704121966e-05,
+      "loss": 0.7282,
       "step": 8270
     },
     {
       "epoch": 1.5584415584415585,
-      "grad_norm": 2.012145519256592,
-      "learning_rate": 1.3766233766233767e-05,
-      "loss": 0.975,
+      "grad_norm": 1.458553671836853,
+      "learning_rate": 2.064935064935065e-05,
+      "loss": 0.8282,
       "step": 8280
     },
     {
       "epoch": 1.5603237342367777,
-      "grad_norm": 21.390483856201172,
-      "learning_rate": 1.375870506305289e-05,
-      "loss": 1.3669,
+      "grad_norm": 22.097503662109375,
+      "learning_rate": 2.0638057594579335e-05,
+      "loss": 1.32,
       "step": 8290
     },
     {
       "epoch": 1.562205910031997,
-      "grad_norm": 2.5964221954345703,
-      "learning_rate": 1.3751176359872015e-05,
-      "loss": 0.8333,
+      "grad_norm": 1.0465900897979736,
+      "learning_rate": 2.062676453980802e-05,
+      "loss": 0.6825,
       "step": 8300
     },
     {
       "epoch": 1.5640880858272164,
-      "grad_norm": 7.925995349884033,
-      "learning_rate": 1.3743647656691137e-05,
-      "loss": 1.3074,
+      "grad_norm": 4.5169782638549805,
+      "learning_rate": 2.0615471485036704e-05,
+      "loss": 1.0293,
       "step": 8310
     },
     {
       "epoch": 1.5659702616224356,
-      "grad_norm": 11.784333229064941,
-      "learning_rate": 1.3736118953510259e-05,
-      "loss": 0.6514,
+      "grad_norm": 24.006298065185547,
+      "learning_rate": 2.0604178430265386e-05,
+      "loss": 0.4358,
       "step": 8320
     },
     {
       "epoch": 1.5678524374176548,
-      "grad_norm": 13.306680679321289,
-      "learning_rate": 1.3728590250329382e-05,
-      "loss": 0.9919,
+      "grad_norm": 34.115413665771484,
+      "learning_rate": 2.059288537549407e-05,
+      "loss": 0.7429,
       "step": 8330
     },
     {
       "epoch": 1.5697346132128742,
-      "grad_norm": 4.451740264892578,
-      "learning_rate": 1.3721061547148505e-05,
-      "loss": 0.8925,
+      "grad_norm": 0.8864557147026062,
+      "learning_rate": 2.0581592320722755e-05,
+      "loss": 0.6736,
       "step": 8340
     },
     {
       "epoch": 1.5716167890080932,
-      "grad_norm": 7.311500072479248,
-      "learning_rate": 1.3713532843967627e-05,
-      "loss": 0.8031,
+      "grad_norm": 3.0202507972717285,
+      "learning_rate": 2.057029926595144e-05,
+      "loss": 0.6662,
       "step": 8350
     },
     {
       "epoch": 1.5734989648033126,
-      "grad_norm": 20.71784210205078,
-      "learning_rate": 1.370600414078675e-05,
-      "loss": 1.1857,
+      "grad_norm": 16.685216903686523,
+      "learning_rate": 2.0559006211180124e-05,
+      "loss": 1.1072,
       "step": 8360
     },
     {
       "epoch": 1.5753811405985318,
-      "grad_norm": 10.758478164672852,
-      "learning_rate": 1.3698475437605873e-05,
-      "loss": 0.8431,
+      "grad_norm": 14.748647689819336,
+      "learning_rate": 2.054771315640881e-05,
+      "loss": 0.6912,
       "step": 8370
     },
     {
       "epoch": 1.577263316393751,
-      "grad_norm": 7.647435665130615,
-      "learning_rate": 1.3690946734424996e-05,
-      "loss": 1.2386,
+      "grad_norm": 19.778024673461914,
+      "learning_rate": 2.0536420101637493e-05,
+      "loss": 1.0492,
       "step": 8380
     },
     {
       "epoch": 1.5791454921889705,
-      "grad_norm": 1.5989456176757812,
-      "learning_rate": 1.368341803124412e-05,
-      "loss": 1.154,
+      "grad_norm": 1.4614105224609375,
+      "learning_rate": 2.052512704686618e-05,
+      "loss": 1.0389,
       "step": 8390
     },
     {
       "epoch": 1.5810276679841897,
-      "grad_norm": 2.973799467086792,
-      "learning_rate": 1.3675889328063244e-05,
-      "loss": 0.8838,
+      "grad_norm": 2.2111265659332275,
+      "learning_rate": 2.0513833992094865e-05,
+      "loss": 0.7137,
       "step": 8400
     },
     {
       "epoch": 1.582909843779409,
-      "grad_norm": 8.04106616973877,
-      "learning_rate": 1.3668360624882365e-05,
-      "loss": 1.0488,
+      "grad_norm": 25.663915634155273,
+      "learning_rate": 2.0502540937323544e-05,
+      "loss": 0.8827,
       "step": 8410
     },
     {
       "epoch": 1.5847920195746283,
-      "grad_norm": 23.760622024536133,
-      "learning_rate": 1.3660831921701488e-05,
-      "loss": 1.4356,
+      "grad_norm": 32.34187316894531,
+      "learning_rate": 2.049124788255223e-05,
+      "loss": 1.359,
       "step": 8420
     },
     {
       "epoch": 1.5866741953698476,
-      "grad_norm": 9.534244537353516,
-      "learning_rate": 1.365330321852061e-05,
-      "loss": 0.9192,
+      "grad_norm": 9.49844741821289,
+      "learning_rate": 2.0479954827780913e-05,
+      "loss": 0.7594,
       "step": 8430
     },
     {
       "epoch": 1.5885563711650668,
-      "grad_norm": 8.14865493774414,
-      "learning_rate": 1.3645774515339734e-05,
-      "loss": 1.0321,
+      "grad_norm": 8.139872550964355,
+      "learning_rate": 2.04686617730096e-05,
+      "loss": 0.9236,
       "step": 8440
     },
     {
       "epoch": 1.5904385469602862,
-      "grad_norm": 5.678312301635742,
-      "learning_rate": 1.3638245812158857e-05,
-      "loss": 1.1119,
+      "grad_norm": 4.609650135040283,
+      "learning_rate": 2.0457368718238285e-05,
+      "loss": 1.0342,
       "step": 8450
     },
     {
       "epoch": 1.5923207227555054,
-      "grad_norm": 19.022024154663086,
-      "learning_rate": 1.363071710897798e-05,
-      "loss": 1.0776,
+      "grad_norm": 13.851377487182617,
+      "learning_rate": 2.0446075663466968e-05,
+      "loss": 0.985,
       "step": 8460
     },
     {
       "epoch": 1.5942028985507246,
-      "grad_norm": 23.45431137084961,
-      "learning_rate": 1.3623188405797103e-05,
-      "loss": 0.9972,
+      "grad_norm": 17.376338958740234,
+      "learning_rate": 2.0434782608695654e-05,
+      "loss": 0.8613,
       "step": 8470
     },
     {
       "epoch": 1.596085074345944,
-      "grad_norm": 26.307212829589844,
-      "learning_rate": 1.3615659702616225e-05,
-      "loss": 0.9018,
+      "grad_norm": 16.559776306152344,
+      "learning_rate": 2.042348955392434e-05,
+      "loss": 0.6948,
       "step": 8480
     },
     {
       "epoch": 1.597967250141163,
-      "grad_norm": 38.280635833740234,
-      "learning_rate": 1.360813099943535e-05,
-      "loss": 0.9647,
+      "grad_norm": 28.88027000427246,
+      "learning_rate": 2.0412196499153023e-05,
+      "loss": 0.7705,
       "step": 8490
     },
     {
       "epoch": 1.5998494259363825,
-      "grad_norm": 13.799212455749512,
-      "learning_rate": 1.360060229625447e-05,
-      "loss": 0.9421,
+      "grad_norm": 25.906251907348633,
+      "learning_rate": 2.0400903444381705e-05,
+      "loss": 0.8089,
       "step": 8500
     },
     {
       "epoch": 1.601731601731602,
-      "grad_norm": 3.608424425125122,
-      "learning_rate": 1.3593073593073594e-05,
-      "loss": 0.7064,
+      "grad_norm": 3.415189266204834,
+      "learning_rate": 2.0389610389610388e-05,
+      "loss": 0.4868,
       "step": 8510
     },
     {
       "epoch": 1.6036137775268209,
-      "grad_norm": 6.385775566101074,
-      "learning_rate": 1.3585544889892717e-05,
-      "loss": 0.8039,
+      "grad_norm": 7.368528366088867,
+      "learning_rate": 2.0378317334839074e-05,
+      "loss": 0.6249,
       "step": 8520
     },
     {
       "epoch": 1.6054959533220403,
-      "grad_norm": 13.767354965209961,
-      "learning_rate": 1.357801618671184e-05,
-      "loss": 0.9991,
+      "grad_norm": 12.94930362701416,
+      "learning_rate": 2.036702428006776e-05,
+      "loss": 0.8234,
       "step": 8530
     },
     {
       "epoch": 1.6073781291172595,
-      "grad_norm": 4.132275581359863,
-      "learning_rate": 1.3570487483530963e-05,
-      "loss": 0.9002,
+      "grad_norm": 1.4030075073242188,
+      "learning_rate": 2.0355731225296443e-05,
+      "loss": 0.7259,
       "step": 8540
     },
     {
       "epoch": 1.6092603049124787,
-      "grad_norm": 20.56551170349121,
-      "learning_rate": 1.3562958780350086e-05,
-      "loss": 1.114,
+      "grad_norm": 21.617116928100586,
+      "learning_rate": 2.034443817052513e-05,
+      "loss": 1.0274,
       "step": 8550
     },
     {
       "epoch": 1.6111424807076982,
-      "grad_norm": 12.178905487060547,
-      "learning_rate": 1.3555430077169209e-05,
-      "loss": 0.9161,
+      "grad_norm": 14.721446990966797,
+      "learning_rate": 2.033314511575381e-05,
+      "loss": 0.7038,
       "step": 8560
     },
     {
       "epoch": 1.6130246565029174,
-      "grad_norm": 9.772829055786133,
-      "learning_rate": 1.3547901373988332e-05,
-      "loss": 1.0685,
+      "grad_norm": 37.175865173339844,
+      "learning_rate": 2.0321852060982498e-05,
+      "loss": 0.8129,
       "step": 8570
     },
     {
       "epoch": 1.6149068322981366,
-      "grad_norm": 12.571735382080078,
-      "learning_rate": 1.3540372670807453e-05,
-      "loss": 0.8599,
+      "grad_norm": 18.872203826904297,
+      "learning_rate": 2.031055900621118e-05,
+      "loss": 0.7054,
       "step": 8580
     },
     {
       "epoch": 1.616789008093356,
-      "grad_norm": 39.86054611206055,
-      "learning_rate": 1.3532843967626576e-05,
-      "loss": 0.8537,
+      "grad_norm": 18.754467010498047,
+      "learning_rate": 2.0299265951439863e-05,
+      "loss": 0.7435,
       "step": 8590
     },
     {
       "epoch": 1.6186711838885752,
-      "grad_norm": 40.22355651855469,
-      "learning_rate": 1.35253152644457e-05,
-      "loss": 1.1449,
+      "grad_norm": 17.79302406311035,
+      "learning_rate": 2.028797289666855e-05,
+      "loss": 0.8313,
       "step": 8600
     },
     {
       "epoch": 1.6205533596837944,
-      "grad_norm": 13.236491203308105,
-      "learning_rate": 1.3517786561264823e-05,
-      "loss": 0.8977,
+      "grad_norm": 17.659814834594727,
+      "learning_rate": 2.0276679841897232e-05,
+      "loss": 0.7874,
       "step": 8610
     },
     {
       "epoch": 1.6224355354790139,
-      "grad_norm": 11.93606948852539,
-      "learning_rate": 1.3510257858083946e-05,
-      "loss": 0.8524,
+      "grad_norm": 14.71423625946045,
+      "learning_rate": 2.0265386787125918e-05,
+      "loss": 0.83,
       "step": 8620
     },
     {
       "epoch": 1.6243177112742329,
-      "grad_norm": 11.476794242858887,
-      "learning_rate": 1.350272915490307e-05,
-      "loss": 1.1183,
+      "grad_norm": 17.527603149414062,
+      "learning_rate": 2.0254093732354604e-05,
+      "loss": 1.0773,
       "step": 8630
     },
     {
       "epoch": 1.6261998870694523,
-      "grad_norm": 8.892841339111328,
-      "learning_rate": 1.3495200451722192e-05,
-      "loss": 1.1046,
+      "grad_norm": 12.642634391784668,
+      "learning_rate": 2.0242800677583287e-05,
+      "loss": 0.9208,
       "step": 8640
     },
     {
       "epoch": 1.6280820628646717,
-      "grad_norm": 3.0642027854919434,
-      "learning_rate": 1.3487671748541315e-05,
-      "loss": 1.0486,
+      "grad_norm": 1.2495397329330444,
+      "learning_rate": 2.0231507622811973e-05,
+      "loss": 0.9029,
       "step": 8650
     },
     {
       "epoch": 1.6299642386598907,
-      "grad_norm": 1.5371010303497314,
-      "learning_rate": 1.3480143045360438e-05,
-      "loss": 0.8492,
+      "grad_norm": 0.4178260564804077,
+      "learning_rate": 2.0220214568040655e-05,
+      "loss": 0.6525,
       "step": 8660
     },
     {
       "epoch": 1.6318464144551101,
-      "grad_norm": 5.408362865447998,
-      "learning_rate": 1.347261434217956e-05,
-      "loss": 0.8306,
+      "grad_norm": 12.84122085571289,
+      "learning_rate": 2.0208921513269338e-05,
+      "loss": 0.6563,
       "step": 8670
     },
     {
       "epoch": 1.6337285902503293,
-      "grad_norm": 18.592185974121094,
-      "learning_rate": 1.3465085638998682e-05,
-      "loss": 1.0851,
+      "grad_norm": 14.375418663024902,
+      "learning_rate": 2.0197628458498024e-05,
+      "loss": 0.9404,
       "step": 8680
     },
     {
       "epoch": 1.6356107660455486,
-      "grad_norm": 5.693188667297363,
-      "learning_rate": 1.3457556935817807e-05,
-      "loss": 0.9415,
+      "grad_norm": 5.309718132019043,
+      "learning_rate": 2.0186335403726707e-05,
+      "loss": 0.8255,
       "step": 8690
     },
     {
       "epoch": 1.637492941840768,
-      "grad_norm": 17.856067657470703,
-      "learning_rate": 1.345002823263693e-05,
-      "loss": 1.0766,
+      "grad_norm": 19.447336196899414,
+      "learning_rate": 2.0175042348955393e-05,
+      "loss": 0.9448,
       "step": 8700
     },
     {
       "epoch": 1.6393751176359872,
-      "grad_norm": 22.922794342041016,
-      "learning_rate": 1.3442499529456053e-05,
-      "loss": 0.9564,
+      "grad_norm": 5.150761127471924,
+      "learning_rate": 2.016374929418408e-05,
+      "loss": 0.7773,
       "step": 8710
     },
     {
       "epoch": 1.6412572934312064,
-      "grad_norm": 10.05300235748291,
-      "learning_rate": 1.3434970826275176e-05,
-      "loss": 0.7716,
+      "grad_norm": 10.67483139038086,
+      "learning_rate": 2.015245623941276e-05,
+      "loss": 0.6024,
       "step": 8720
     },
     {
       "epoch": 1.6431394692264258,
-      "grad_norm": 20.4290771484375,
-      "learning_rate": 1.3427442123094298e-05,
-      "loss": 0.7658,
+      "grad_norm": 18.900775909423828,
+      "learning_rate": 2.0141163184641448e-05,
+      "loss": 0.6435,
       "step": 8730
     },
     {
       "epoch": 1.645021645021645,
-      "grad_norm": 5.071466445922852,
-      "learning_rate": 1.3419913419913421e-05,
-      "loss": 1.0068,
+      "grad_norm": 7.011523246765137,
+      "learning_rate": 2.012987012987013e-05,
+      "loss": 0.9581,
       "step": 8740
     },
     {
       "epoch": 1.6469038208168643,
-      "grad_norm": 12.650185585021973,
-      "learning_rate": 1.3412384716732544e-05,
-      "loss": 1.2319,
+      "grad_norm": 20.141672134399414,
+      "learning_rate": 2.0118577075098816e-05,
+      "loss": 0.8604,
       "step": 8750
     },
     {
       "epoch": 1.6487859966120837,
-      "grad_norm": 27.368560791015625,
-      "learning_rate": 1.3404856013551666e-05,
-      "loss": 0.8958,
+      "grad_norm": 19.226642608642578,
+      "learning_rate": 2.01072840203275e-05,
+      "loss": 0.8389,
       "step": 8760
     },
     {
       "epoch": 1.6506681724073027,
-      "grad_norm": 10.492888450622559,
-      "learning_rate": 1.3397327310370788e-05,
-      "loss": 1.0703,
+      "grad_norm": 12.022441864013672,
+      "learning_rate": 2.0095990965556182e-05,
+      "loss": 0.8879,
       "step": 8770
     },
     {
       "epoch": 1.6525503482025221,
-      "grad_norm": 15.578258514404297,
-      "learning_rate": 1.3389798607189911e-05,
-      "loss": 1.0944,
+      "grad_norm": 16.575210571289062,
+      "learning_rate": 2.0084697910784868e-05,
+      "loss": 0.9649,
       "step": 8780
     },
     {
       "epoch": 1.6544325239977415,
-      "grad_norm": 7.515652179718018,
-      "learning_rate": 1.3382269904009036e-05,
-      "loss": 1.3246,
+      "grad_norm": 13.532410621643066,
+      "learning_rate": 2.007340485601355e-05,
+      "loss": 1.1161,
       "step": 8790
     },
     {
       "epoch": 1.6563146997929605,
-      "grad_norm": 9.480490684509277,
-      "learning_rate": 1.3374741200828159e-05,
-      "loss": 0.89,
+      "grad_norm": 13.928875923156738,
+      "learning_rate": 2.0062111801242237e-05,
+      "loss": 0.57,
       "step": 8800
     },
     {
       "epoch": 1.65819687558818,
-      "grad_norm": 19.28694725036621,
-      "learning_rate": 1.3367212497647282e-05,
-      "loss": 0.6789,
+      "grad_norm": 9.75232982635498,
+      "learning_rate": 2.0050818746470923e-05,
+      "loss": 0.5475,
       "step": 8810
     },
     {
       "epoch": 1.6600790513833992,
-      "grad_norm": 7.083702564239502,
-      "learning_rate": 1.3359683794466405e-05,
-      "loss": 0.9053,
+      "grad_norm": 19.574495315551758,
+      "learning_rate": 2.0039525691699605e-05,
+      "loss": 0.8012,
       "step": 8820
     },
     {
       "epoch": 1.6619612271786184,
-      "grad_norm": 49.1333122253418,
-      "learning_rate": 1.3352155091285528e-05,
-      "loss": 1.0765,
+      "grad_norm": 8.311776161193848,
+      "learning_rate": 2.002823263692829e-05,
+      "loss": 0.8989,
       "step": 8830
     },
     {
       "epoch": 1.6638434029738378,
-      "grad_norm": 3.065033197402954,
-      "learning_rate": 1.334462638810465e-05,
-      "loss": 0.9586,
+      "grad_norm": 3.7794203758239746,
+      "learning_rate": 2.0016939582156974e-05,
+      "loss": 0.7828,
       "step": 8840
     },
     {
       "epoch": 1.665725578769057,
-      "grad_norm": 21.612653732299805,
-      "learning_rate": 1.3337097684923772e-05,
-      "loss": 1.0886,
+      "grad_norm": 22.881847381591797,
+      "learning_rate": 2.0005646527385657e-05,
+      "loss": 1.0597,
       "step": 8850
     },
     {
       "epoch": 1.6676077545642762,
-      "grad_norm": 24.604717254638672,
-      "learning_rate": 1.3329568981742895e-05,
-      "loss": 0.9681,
+      "grad_norm": 23.629314422607422,
+      "learning_rate": 1.9994353472614343e-05,
+      "loss": 0.905,
       "step": 8860
     },
     {
       "epoch": 1.6694899303594957,
-      "grad_norm": 15.733131408691406,
-      "learning_rate": 1.3322040278562018e-05,
-      "loss": 1.0073,
+      "grad_norm": 19.81004524230957,
+      "learning_rate": 1.9983060417843026e-05,
+      "loss": 1.0952,
       "step": 8870
     },
     {
       "epoch": 1.6713721061547149,
-      "grad_norm": 30.626779556274414,
-      "learning_rate": 1.3314511575381142e-05,
-      "loss": 0.8748,
+      "grad_norm": 8.309386253356934,
+      "learning_rate": 1.9971767363071712e-05,
+      "loss": 0.7427,
       "step": 8880
     },
     {
       "epoch": 1.673254281949934,
-      "grad_norm": 7.506359100341797,
-      "learning_rate": 1.3306982872200265e-05,
-      "loss": 0.9653,
+      "grad_norm": 6.19067907333374,
+      "learning_rate": 1.9960474308300394e-05,
+      "loss": 0.7889,
       "step": 8890
     },
     {
       "epoch": 1.6751364577451535,
-      "grad_norm": 5.364382266998291,
-      "learning_rate": 1.3299454169019388e-05,
-      "loss": 1.0913,
+      "grad_norm": 2.2768592834472656,
+      "learning_rate": 1.994918125352908e-05,
+      "loss": 0.8985,
       "step": 8900
     },
     {
       "epoch": 1.6770186335403725,
-      "grad_norm": 7.690507411956787,
-      "learning_rate": 1.3291925465838511e-05,
-      "loss": 0.8708,
+      "grad_norm": 7.979760646820068,
+      "learning_rate": 1.9937888198757767e-05,
+      "loss": 0.6913,
       "step": 8910
     },
     {
       "epoch": 1.678900809335592,
-      "grad_norm": 5.078372478485107,
-      "learning_rate": 1.3284396762657634e-05,
-      "loss": 1.0961,
+      "grad_norm": 10.671245574951172,
+      "learning_rate": 1.992659514398645e-05,
+      "loss": 0.995,
       "step": 8920
     },
     {
       "epoch": 1.6807829851308114,
-      "grad_norm": 8.640311241149902,
-      "learning_rate": 1.3276868059476755e-05,
-      "loss": 1.1776,
+      "grad_norm": 8.021512031555176,
+      "learning_rate": 1.9915302089215132e-05,
+      "loss": 1.1053,
       "step": 8930
     },
     {
       "epoch": 1.6826651609260304,
-      "grad_norm": 8.303173065185547,
-      "learning_rate": 1.3269339356295878e-05,
-      "loss": 0.8504,
+      "grad_norm": 10.151590347290039,
+      "learning_rate": 1.9904009034443818e-05,
+      "loss": 0.7479,
       "step": 8940
     },
     {
       "epoch": 1.6845473367212498,
-      "grad_norm": 13.893898010253906,
-      "learning_rate": 1.3261810653115001e-05,
-      "loss": 0.9924,
+      "grad_norm": 15.058424949645996,
+      "learning_rate": 1.98927159796725e-05,
+      "loss": 0.9105,
       "step": 8950
     },
     {
       "epoch": 1.686429512516469,
-      "grad_norm": 1.1722054481506348,
-      "learning_rate": 1.3254281949934124e-05,
-      "loss": 0.8902,
+      "grad_norm": 0.7414102554321289,
+      "learning_rate": 1.9881422924901187e-05,
+      "loss": 0.7906,
       "step": 8960
     },
     {
       "epoch": 1.6883116883116882,
-      "grad_norm": 14.810260772705078,
-      "learning_rate": 1.3246753246753249e-05,
-      "loss": 0.9838,
+      "grad_norm": 16.738800048828125,
+      "learning_rate": 1.987012987012987e-05,
+      "loss": 0.891,
       "step": 8970
     },
     {
       "epoch": 1.6901938641069076,
-      "grad_norm": 6.508338451385498,
-      "learning_rate": 1.3239224543572372e-05,
-      "loss": 0.9525,
+      "grad_norm": 11.300554275512695,
+      "learning_rate": 1.9858836815358556e-05,
+      "loss": 0.849,
       "step": 8980
     },
     {
       "epoch": 1.6920760399021268,
-      "grad_norm": 18.174537658691406,
-      "learning_rate": 1.3231695840391494e-05,
-      "loss": 1.2055,
+      "grad_norm": 23.922866821289062,
+      "learning_rate": 1.984754376058724e-05,
+      "loss": 1.182,
       "step": 8990
     },
     {
       "epoch": 1.693958215697346,
-      "grad_norm": 14.68772029876709,
-      "learning_rate": 1.3224167137210617e-05,
-      "loss": 0.7929,
+      "grad_norm": 15.910576820373535,
+      "learning_rate": 1.9836250705815924e-05,
+      "loss": 0.692,
       "step": 9000
     },
     {
       "epoch": 1.6958403914925655,
-      "grad_norm": 8.183045387268066,
-      "learning_rate": 1.321663843402974e-05,
-      "loss": 1.0087,
+      "grad_norm": 12.029934883117676,
+      "learning_rate": 1.982495765104461e-05,
+      "loss": 0.8861,
       "step": 9010
     },
     {
       "epoch": 1.6977225672877847,
-      "grad_norm": 29.93727684020996,
-      "learning_rate": 1.3209109730848861e-05,
-      "loss": 0.973,
+      "grad_norm": 11.433530807495117,
+      "learning_rate": 1.981366459627329e-05,
+      "loss": 0.8564,
       "step": 9020
     },
     {
       "epoch": 1.699604743083004,
-      "grad_norm": 10.451583862304688,
-      "learning_rate": 1.3201581027667984e-05,
-      "loss": 0.868,
+      "grad_norm": 16.016300201416016,
+      "learning_rate": 1.9802371541501976e-05,
+      "loss": 0.7998,
       "step": 9030
     },
     {
       "epoch": 1.7014869188782233,
-      "grad_norm": 9.627779006958008,
-      "learning_rate": 1.3194052324487107e-05,
-      "loss": 0.9869,
+      "grad_norm": 13.163691520690918,
+      "learning_rate": 1.9791078486730662e-05,
+      "loss": 0.8364,
       "step": 9040
     },
     {
       "epoch": 1.7033690946734426,
-      "grad_norm": 15.182428359985352,
-      "learning_rate": 1.318652362130623e-05,
-      "loss": 0.9524,
+      "grad_norm": 15.584578514099121,
+      "learning_rate": 1.9779785431959345e-05,
+      "loss": 0.8937,
       "step": 9050
     },
     {
       "epoch": 1.7052512704686618,
-      "grad_norm": 18.331754684448242,
-      "learning_rate": 1.3178994918125355e-05,
-      "loss": 0.6643,
+      "grad_norm": 34.573814392089844,
+      "learning_rate": 1.976849237718803e-05,
+      "loss": 0.4642,
       "step": 9060
     },
     {
       "epoch": 1.7071334462638812,
-      "grad_norm": 17.56291961669922,
-      "learning_rate": 1.3171466214944478e-05,
-      "loss": 1.0118,
+      "grad_norm": 3.945563316345215,
+      "learning_rate": 1.9757199322416713e-05,
+      "loss": 0.9052,
       "step": 9070
     },
     {
       "epoch": 1.7090156220591002,
-      "grad_norm": 31.774778366088867,
-      "learning_rate": 1.31639375117636e-05,
-      "loss": 0.8729,
+      "grad_norm": 14.851612091064453,
+      "learning_rate": 1.97459062676454e-05,
+      "loss": 0.766,
       "step": 9080
     },
     {
       "epoch": 1.7108977978543196,
-      "grad_norm": 31.31987953186035,
-      "learning_rate": 1.3156408808582724e-05,
-      "loss": 0.8254,
+      "grad_norm": 20.84731101989746,
+      "learning_rate": 1.9734613212874085e-05,
+      "loss": 0.7886,
       "step": 9090
     },
     {
       "epoch": 1.7127799736495388,
-      "grad_norm": 11.088515281677246,
-      "learning_rate": 1.3148880105401847e-05,
-      "loss": 0.9627,
+      "grad_norm": 14.792206764221191,
+      "learning_rate": 1.9723320158102768e-05,
+      "loss": 0.8115,
       "step": 9100
     },
     {
       "epoch": 1.714662149444758,
-      "grad_norm": 14.68917179107666,
-      "learning_rate": 1.3141351402220968e-05,
-      "loss": 0.8448,
+      "grad_norm": 8.901041030883789,
+      "learning_rate": 1.971202710333145e-05,
+      "loss": 0.9397,
       "step": 9110
     },
     {
       "epoch": 1.7165443252399775,
-      "grad_norm": 8.825004577636719,
-      "learning_rate": 1.313382269904009e-05,
-      "loss": 1.0459,
+      "grad_norm": 8.225708961486816,
+      "learning_rate": 1.9700734048560134e-05,
+      "loss": 1.1384,
       "step": 9120
     },
     {
       "epoch": 1.7184265010351967,
-      "grad_norm": 5.718069076538086,
-      "learning_rate": 1.3126293995859214e-05,
-      "loss": 0.8609,
+      "grad_norm": 5.585742473602295,
+      "learning_rate": 1.968944099378882e-05,
+      "loss": 0.794,
       "step": 9130
     },
     {
       "epoch": 1.7203086768304159,
-      "grad_norm": 4.432228088378906,
-      "learning_rate": 1.3118765292678337e-05,
-      "loss": 0.7144,
+      "grad_norm": 4.999485969543457,
+      "learning_rate": 1.9678147939017506e-05,
+      "loss": 0.6178,
       "step": 9140
     },
     {
       "epoch": 1.7221908526256353,
-      "grad_norm": 17.18754768371582,
-      "learning_rate": 1.311123658949746e-05,
-      "loss": 1.2041,
+      "grad_norm": 15.160903930664062,
+      "learning_rate": 1.966685488424619e-05,
+      "loss": 1.1031,
       "step": 9150
     },
     {
       "epoch": 1.7240730284208545,
-      "grad_norm": 31.874040603637695,
-      "learning_rate": 1.3103707886316584e-05,
-      "loss": 1.1127,
+      "grad_norm": 27.700048446655273,
+      "learning_rate": 1.9655561829474874e-05,
+      "loss": 0.9744,
       "step": 9160
     },
     {
       "epoch": 1.7259552042160737,
-      "grad_norm": 20.094186782836914,
-      "learning_rate": 1.3096179183135707e-05,
-      "loss": 1.0748,
+      "grad_norm": 15.794424057006836,
+      "learning_rate": 1.9644268774703557e-05,
+      "loss": 0.9782,
       "step": 9170
     },
     {
       "epoch": 1.7278373800112932,
-      "grad_norm": 11.178129196166992,
-      "learning_rate": 1.308865047995483e-05,
-      "loss": 1.1735,
+      "grad_norm": 13.252302169799805,
+      "learning_rate": 1.9632975719932243e-05,
+      "loss": 0.9227,
       "step": 9180
     },
     {
       "epoch": 1.7297195558065124,
-      "grad_norm": 15.87864875793457,
-      "learning_rate": 1.3081121776773951e-05,
-      "loss": 0.8207,
+      "grad_norm": 19.900714874267578,
+      "learning_rate": 1.9621682665160926e-05,
+      "loss": 0.7203,
       "step": 9190
     },
     {
       "epoch": 1.7316017316017316,
-      "grad_norm": 13.691402435302734,
-      "learning_rate": 1.3073593073593074e-05,
-      "loss": 1.0621,
+      "grad_norm": 8.167475700378418,
+      "learning_rate": 1.961038961038961e-05,
+      "loss": 0.9278,
       "step": 9200
     },
     {
       "epoch": 1.733483907396951,
-      "grad_norm": 10.819106101989746,
-      "learning_rate": 1.3066064370412197e-05,
-      "loss": 1.1812,
+      "grad_norm": 18.376384735107422,
+      "learning_rate": 1.9599096555618295e-05,
+      "loss": 1.005,
       "step": 9210
     },
     {
       "epoch": 1.73536608319217,
-      "grad_norm": 13.477234840393066,
-      "learning_rate": 1.305853566723132e-05,
-      "loss": 0.8893,
+      "grad_norm": 12.970340728759766,
+      "learning_rate": 1.958780350084698e-05,
+      "loss": 0.7519,
       "step": 9220
     },
     {
       "epoch": 1.7372482589873894,
-      "grad_norm": 14.02323055267334,
-      "learning_rate": 1.3051006964050443e-05,
-      "loss": 0.9262,
+      "grad_norm": 10.988688468933105,
+      "learning_rate": 1.9576510446075663e-05,
+      "loss": 0.8544,
       "step": 9230
     },
     {
       "epoch": 1.7391304347826086,
-      "grad_norm": 18.27639389038086,
-      "learning_rate": 1.3043478260869566e-05,
-      "loss": 1.1554,
+      "grad_norm": 7.907069206237793,
+      "learning_rate": 1.956521739130435e-05,
+      "loss": 1.2205,
       "step": 9240
     },
     {
       "epoch": 1.7410126105778279,
-      "grad_norm": 23.120223999023438,
-      "learning_rate": 1.303594955768869e-05,
-      "loss": 0.9298,
+      "grad_norm": 24.321718215942383,
+      "learning_rate": 1.9553924336533032e-05,
+      "loss": 0.7836,
       "step": 9250
     },
     {
       "epoch": 1.7428947863730473,
-      "grad_norm": 9.989583015441895,
-      "learning_rate": 1.3028420854507813e-05,
-      "loss": 0.9628,
+      "grad_norm": 11.95626449584961,
+      "learning_rate": 1.9542631281761718e-05,
+      "loss": 0.7509,
       "step": 9260
     },
     {
       "epoch": 1.7447769621682665,
-      "grad_norm": 1.417387843132019,
-      "learning_rate": 1.3020892151326936e-05,
-      "loss": 0.7619,
+      "grad_norm": 1.1062142848968506,
+      "learning_rate": 1.9531338226990404e-05,
+      "loss": 0.5899,
       "step": 9270
     },
     {
       "epoch": 1.7466591379634857,
-      "grad_norm": 17.73150062561035,
-      "learning_rate": 1.3013363448146057e-05,
-      "loss": 0.8362,
+      "grad_norm": 12.256535530090332,
+      "learning_rate": 1.9520045172219084e-05,
+      "loss": 0.7338,
       "step": 9280
     },
     {
       "epoch": 1.7485413137587051,
-      "grad_norm": 22.77169418334961,
-      "learning_rate": 1.300583474496518e-05,
-      "loss": 1.1307,
+      "grad_norm": 28.20414924621582,
+      "learning_rate": 1.950875211744777e-05,
+      "loss": 0.848,
       "step": 9290
     },
     {
       "epoch": 1.7504234895539243,
-      "grad_norm": 14.768538475036621,
-      "learning_rate": 1.2998306041784303e-05,
-      "loss": 1.1565,
+      "grad_norm": 10.83683967590332,
+      "learning_rate": 1.9497459062676452e-05,
+      "loss": 1.0118,
       "step": 9300
     },
     {
       "epoch": 1.7523056653491436,
-      "grad_norm": 1.281310796737671,
-      "learning_rate": 1.2990777338603426e-05,
-      "loss": 0.7435,
+      "grad_norm": 12.281524658203125,
+      "learning_rate": 1.948616600790514e-05,
+      "loss": 0.7269,
       "step": 9310
     },
     {
       "epoch": 1.754187841144363,
-      "grad_norm": 5.074652671813965,
-      "learning_rate": 1.2983248635422549e-05,
-      "loss": 0.7562,
+      "grad_norm": 3.447277545928955,
+      "learning_rate": 1.9474872953133825e-05,
+      "loss": 0.6222,
       "step": 9320
     },
     {
       "epoch": 1.7560700169395822,
-      "grad_norm": 24.476675033569336,
-      "learning_rate": 1.2975719932241672e-05,
-      "loss": 0.9345,
+      "grad_norm": 21.582672119140625,
+      "learning_rate": 1.9463579898362507e-05,
+      "loss": 0.9156,
       "step": 9330
     },
     {
       "epoch": 1.7579521927348014,
-      "grad_norm": 5.09979772567749,
-      "learning_rate": 1.2968191229060797e-05,
-      "loss": 0.8928,
+      "grad_norm": 8.327972412109375,
+      "learning_rate": 1.9452286843591193e-05,
+      "loss": 0.7264,
       "step": 9340
     },
     {
       "epoch": 1.7598343685300208,
-      "grad_norm": 18.816225051879883,
-      "learning_rate": 1.296066252587992e-05,
-      "loss": 0.6661,
+      "grad_norm": 9.982987403869629,
+      "learning_rate": 1.9440993788819876e-05,
+      "loss": 0.5234,
       "step": 9350
     },
     {
       "epoch": 1.7617165443252398,
-      "grad_norm": 14.58993911743164,
-      "learning_rate": 1.2953133822699043e-05,
-      "loss": 0.7788,
+      "grad_norm": 18.895675659179688,
+      "learning_rate": 1.9429700734048562e-05,
+      "loss": 0.6,
       "step": 9360
     },
     {
       "epoch": 1.7635987201204593,
-      "grad_norm": 8.356761932373047,
-      "learning_rate": 1.2945605119518164e-05,
-      "loss": 0.9851,
+      "grad_norm": 3.999253749847412,
+      "learning_rate": 1.9418407679277245e-05,
+      "loss": 0.8148,
       "step": 9370
     },
     {
       "epoch": 1.7654808959156785,
-      "grad_norm": 19.27338218688965,
-      "learning_rate": 1.2938076416337287e-05,
-      "loss": 0.8405,
+      "grad_norm": 7.097409725189209,
+      "learning_rate": 1.9407114624505927e-05,
+      "loss": 0.6838,
       "step": 9380
     },
     {
       "epoch": 1.7673630717108977,
-      "grad_norm": 9.037829399108887,
-      "learning_rate": 1.293054771315641e-05,
-      "loss": 0.9395,
+      "grad_norm": 10.425472259521484,
+      "learning_rate": 1.9395821569734614e-05,
+      "loss": 0.7291,
       "step": 9390
     },
     {
       "epoch": 1.7692452475061171,
-      "grad_norm": 26.172557830810547,
-      "learning_rate": 1.2923019009975533e-05,
-      "loss": 1.0054,
+      "grad_norm": 15.1722993850708,
+      "learning_rate": 1.9384528514963296e-05,
+      "loss": 0.7761,
       "step": 9400
     },
     {
       "epoch": 1.7711274233013363,
-      "grad_norm": 17.3187313079834,
-      "learning_rate": 1.2915490306794655e-05,
-      "loss": 1.0615,
+      "grad_norm": 11.412665367126465,
+      "learning_rate": 1.9373235460191982e-05,
+      "loss": 0.904,
       "step": 9410
     },
     {
       "epoch": 1.7730095990965555,
-      "grad_norm": 15.355307579040527,
-      "learning_rate": 1.2907961603613778e-05,
-      "loss": 0.8999,
+      "grad_norm": 19.17182159423828,
+      "learning_rate": 1.936194240542067e-05,
+      "loss": 0.9217,
       "step": 9420
     },
     {
       "epoch": 1.774891774891775,
-      "grad_norm": 5.734640598297119,
-      "learning_rate": 1.2900432900432901e-05,
-      "loss": 0.6632,
+      "grad_norm": 4.5345845222473145,
+      "learning_rate": 1.935064935064935e-05,
+      "loss": 0.6485,
       "step": 9430
     },
     {
       "epoch": 1.7767739506869942,
-      "grad_norm": 15.65912914276123,
-      "learning_rate": 1.2892904197252026e-05,
-      "loss": 1.1037,
+      "grad_norm": 24.256860733032227,
+      "learning_rate": 1.9339356295878037e-05,
+      "loss": 1.0271,
       "step": 9440
     },
     {
       "epoch": 1.7786561264822134,
-      "grad_norm": 13.188482284545898,
-      "learning_rate": 1.2885375494071149e-05,
-      "loss": 1.1157,
+      "grad_norm": 13.506421089172363,
+      "learning_rate": 1.932806324110672e-05,
+      "loss": 0.9675,
       "step": 9450
     },
     {
       "epoch": 1.7805383022774328,
-      "grad_norm": 5.424443244934082,
-      "learning_rate": 1.287784679089027e-05,
-      "loss": 0.8627,
+      "grad_norm": 20.032943725585938,
+      "learning_rate": 1.9316770186335403e-05,
+      "loss": 0.8068,
       "step": 9460
     },
     {
       "epoch": 1.782420478072652,
-      "grad_norm": 12.912461280822754,
-      "learning_rate": 1.2870318087709393e-05,
-      "loss": 0.6204,
+      "grad_norm": 30.704687118530273,
+      "learning_rate": 1.930547713156409e-05,
+      "loss": 0.4575,
       "step": 9470
     },
     {
       "epoch": 1.7843026538678712,
-      "grad_norm": 7.88910436630249,
-      "learning_rate": 1.2862789384528516e-05,
-      "loss": 1.0365,
+      "grad_norm": 29.728713989257812,
+      "learning_rate": 1.929418407679277e-05,
+      "loss": 0.9168,
       "step": 9480
     },
     {
       "epoch": 1.7861848296630907,
-      "grad_norm": 15.304657936096191,
-      "learning_rate": 1.2855260681347639e-05,
-      "loss": 1.1819,
+      "grad_norm": 21.02536964416504,
+      "learning_rate": 1.9282891022021457e-05,
+      "loss": 1.0197,
       "step": 9490
     },
     {
       "epoch": 1.7880670054583097,
-      "grad_norm": 1.5771760940551758,
-      "learning_rate": 1.2847731978166762e-05,
-      "loss": 0.7879,
+      "grad_norm": 0.558627724647522,
+      "learning_rate": 1.9271597967250143e-05,
+      "loss": 0.7217,
       "step": 9500
     },
     {
       "epoch": 1.789949181253529,
-      "grad_norm": 24.63703155517578,
-      "learning_rate": 1.2840203274985885e-05,
-      "loss": 0.8802,
+      "grad_norm": 18.84478759765625,
+      "learning_rate": 1.9260304912478826e-05,
+      "loss": 0.7922,
       "step": 9510
     },
     {
       "epoch": 1.7918313570487485,
-      "grad_norm": 16.861093521118164,
-      "learning_rate": 1.2832674571805008e-05,
-      "loss": 1.1376,
+      "grad_norm": 17.696800231933594,
+      "learning_rate": 1.9249011857707512e-05,
+      "loss": 1.0852,
       "step": 9520
     },
     {
       "epoch": 1.7937135328439675,
-      "grad_norm": 8.824302673339844,
-      "learning_rate": 1.2825145868624132e-05,
-      "loss": 0.9262,
+      "grad_norm": 20.650711059570312,
+      "learning_rate": 1.9237718802936195e-05,
+      "loss": 0.7487,
       "step": 9530
     },
     {
       "epoch": 1.795595708639187,
-      "grad_norm": 5.604579448699951,
-      "learning_rate": 1.2817617165443252e-05,
-      "loss": 1.1257,
+      "grad_norm": 11.20978832244873,
+      "learning_rate": 1.9226425748164878e-05,
+      "loss": 1.1077,
       "step": 9540
     },
     {
       "epoch": 1.7974778844344061,
-      "grad_norm": 15.983642578125,
-      "learning_rate": 1.2810088462262376e-05,
-      "loss": 0.94,
+      "grad_norm": 9.016942977905273,
+      "learning_rate": 1.9215132693393564e-05,
+      "loss": 0.7859,
       "step": 9550
     },
     {
       "epoch": 1.7993600602296254,
-      "grad_norm": 25.5296573638916,
-      "learning_rate": 1.28025597590815e-05,
-      "loss": 0.7451,
+      "grad_norm": 18.516727447509766,
+      "learning_rate": 1.9203839638622246e-05,
+      "loss": 0.6019,
       "step": 9560
     },
     {
       "epoch": 1.8012422360248448,
-      "grad_norm": 8.427245140075684,
-      "learning_rate": 1.2795031055900622e-05,
-      "loss": 0.8745,
+      "grad_norm": 14.529791831970215,
+      "learning_rate": 1.9192546583850932e-05,
+      "loss": 0.8048,
       "step": 9570
     },
     {
       "epoch": 1.803124411820064,
-      "grad_norm": 12.769954681396484,
-      "learning_rate": 1.2787502352719745e-05,
-      "loss": 1.1612,
+      "grad_norm": 13.547689437866211,
+      "learning_rate": 1.9181253529079615e-05,
+      "loss": 1.0294,
       "step": 9580
     },
     {
       "epoch": 1.8050065876152832,
-      "grad_norm": 12.274664878845215,
-      "learning_rate": 1.2779973649538868e-05,
-      "loss": 0.8479,
+      "grad_norm": 11.232539176940918,
+      "learning_rate": 1.91699604743083e-05,
+      "loss": 0.798,
       "step": 9590
     },
     {
       "epoch": 1.8068887634105026,
-      "grad_norm": 13.23458194732666,
-      "learning_rate": 1.2772444946357991e-05,
-      "loss": 0.7665,
+      "grad_norm": 13.629487037658691,
+      "learning_rate": 1.9158667419536987e-05,
+      "loss": 0.6611,
       "step": 9600
     },
     {
       "epoch": 1.8087709392057219,
-      "grad_norm": 61.233150482177734,
-      "learning_rate": 1.2764916243177114e-05,
-      "loss": 0.5963,
+      "grad_norm": 9.937800407409668,
+      "learning_rate": 1.914737436476567e-05,
+      "loss": 0.5109,
       "step": 9610
     },
     {
       "epoch": 1.810653115000941,
-      "grad_norm": 20.53067970275879,
-      "learning_rate": 1.2757387539996238e-05,
-      "loss": 1.1002,
+      "grad_norm": 14.390426635742188,
+      "learning_rate": 1.9136081309994356e-05,
+      "loss": 0.8687,
       "step": 9620
     },
     {
       "epoch": 1.8125352907961605,
-      "grad_norm": 16.09151268005371,
-      "learning_rate": 1.2749858836815358e-05,
-      "loss": 1.1078,
+      "grad_norm": 21.730375289916992,
+      "learning_rate": 1.9124788255223035e-05,
+      "loss": 1.0062,
       "step": 9630
     },
     {
       "epoch": 1.8144174665913795,
-      "grad_norm": 4.28153133392334,
-      "learning_rate": 1.2742330133634483e-05,
-      "loss": 1.008,
+      "grad_norm": 17.650148391723633,
+      "learning_rate": 1.911349520045172e-05,
+      "loss": 0.7727,
       "step": 9640
     },
     {
       "epoch": 1.816299642386599,
-      "grad_norm": 11.185966491699219,
-      "learning_rate": 1.2734801430453606e-05,
-      "loss": 1.1953,
+      "grad_norm": 13.232230186462402,
+      "learning_rate": 1.9102202145680407e-05,
+      "loss": 0.987,
       "step": 9650
     },
     {
       "epoch": 1.8181818181818183,
-      "grad_norm": 4.550100326538086,
-      "learning_rate": 1.2727272727272728e-05,
-      "loss": 1.1886,
+      "grad_norm": 21.150760650634766,
+      "learning_rate": 1.909090909090909e-05,
+      "loss": 1.2057,
       "step": 9660
     },
     {
       "epoch": 1.8200639939770373,
-      "grad_norm": 10.768220901489258,
-      "learning_rate": 1.2719744024091851e-05,
-      "loss": 1.1359,
+      "grad_norm": 15.0765962600708,
+      "learning_rate": 1.9079616036137776e-05,
+      "loss": 0.9163,
       "step": 9670
     },
     {
       "epoch": 1.8219461697722568,
-      "grad_norm": 10.578123092651367,
-      "learning_rate": 1.2712215320910974e-05,
-      "loss": 0.8207,
+      "grad_norm": 16.338342666625977,
+      "learning_rate": 1.906832298136646e-05,
+      "loss": 0.6045,
       "step": 9680
     },
     {
       "epoch": 1.823828345567476,
-      "grad_norm": 25.884841918945312,
-      "learning_rate": 1.2704686617730097e-05,
-      "loss": 0.9762,
+      "grad_norm": 28.87519073486328,
+      "learning_rate": 1.9057029926595145e-05,
+      "loss": 1.0076,
       "step": 9690
     },
     {
       "epoch": 1.8257105213626952,
-      "grad_norm": 19.521249771118164,
-      "learning_rate": 1.269715791454922e-05,
-      "loss": 1.1357,
+      "grad_norm": 20.42050552368164,
+      "learning_rate": 1.904573687182383e-05,
+      "loss": 1.0798,
       "step": 9700
     },
     {
       "epoch": 1.8275926971579146,
-      "grad_norm": 26.879844665527344,
-      "learning_rate": 1.2689629211368343e-05,
-      "loss": 1.2003,
+      "grad_norm": 14.828946113586426,
+      "learning_rate": 1.9034443817052514e-05,
+      "loss": 1.1065,
       "step": 9710
     },
     {
       "epoch": 1.8294748729531338,
-      "grad_norm": 11.771387100219727,
-      "learning_rate": 1.2682100508187464e-05,
-      "loss": 0.8366,
+      "grad_norm": 17.2311954498291,
+      "learning_rate": 1.9023150762281196e-05,
+      "loss": 0.8212,
       "step": 9720
     },
     {
       "epoch": 1.831357048748353,
-      "grad_norm": 3.633502960205078,
-      "learning_rate": 1.2674571805006587e-05,
-      "loss": 1.1486,
+      "grad_norm": 17.343181610107422,
+      "learning_rate": 1.9011857707509883e-05,
+      "loss": 1.0861,
       "step": 9730
     },
     {
       "epoch": 1.8332392245435725,
-      "grad_norm": 9.438892364501953,
-      "learning_rate": 1.2667043101825712e-05,
-      "loss": 0.9775,
+      "grad_norm": 15.251105308532715,
+      "learning_rate": 1.9000564652738565e-05,
+      "loss": 0.8787,
       "step": 9740
     },
     {
       "epoch": 1.8351214003387917,
-      "grad_norm": 9.89268684387207,
-      "learning_rate": 1.2659514398644835e-05,
-      "loss": 1.4666,
+      "grad_norm": 7.921949863433838,
+      "learning_rate": 1.898927159796725e-05,
+      "loss": 1.5269,
       "step": 9750
     },
     {
       "epoch": 1.8370035761340109,
-      "grad_norm": 8.895881652832031,
-      "learning_rate": 1.2651985695463958e-05,
-      "loss": 0.8229,
+      "grad_norm": 11.668200492858887,
+      "learning_rate": 1.8977978543195934e-05,
+      "loss": 0.7354,
       "step": 9760
     },
     {
       "epoch": 1.8388857519292303,
-      "grad_norm": 18.56089210510254,
-      "learning_rate": 1.264445699228308e-05,
-      "loss": 0.8295,
+      "grad_norm": 27.10658073425293,
+      "learning_rate": 1.896668548842462e-05,
+      "loss": 0.8172,
       "step": 9770
     },
     {
       "epoch": 1.8407679277244493,
-      "grad_norm": 11.154679298400879,
-      "learning_rate": 1.2636928289102204e-05,
-      "loss": 0.8866,
+      "grad_norm": 16.760330200195312,
+      "learning_rate": 1.8955392433653306e-05,
+      "loss": 0.8198,
       "step": 9780
     },
     {
       "epoch": 1.8426501035196687,
-      "grad_norm": 23.42184829711914,
-      "learning_rate": 1.2629399585921326e-05,
-      "loss": 1.0182,
+      "grad_norm": 16.54085350036621,
+      "learning_rate": 1.894409937888199e-05,
+      "loss": 0.8366,
       "step": 9790
     },
     {
       "epoch": 1.8445322793148882,
-      "grad_norm": 11.54366683959961,
-      "learning_rate": 1.2621870882740448e-05,
-      "loss": 0.858,
+      "grad_norm": 12.325793266296387,
+      "learning_rate": 1.893280632411067e-05,
+      "loss": 0.7522,
       "step": 9800
     },
     {
       "epoch": 1.8464144551101072,
-      "grad_norm": 32.32461929321289,
-      "learning_rate": 1.261434217955957e-05,
-      "loss": 0.8094,
+      "grad_norm": 21.898651123046875,
+      "learning_rate": 1.8921513269339354e-05,
+      "loss": 0.7455,
       "step": 9810
     },
     {
       "epoch": 1.8482966309053266,
-      "grad_norm": 17.969310760498047,
-      "learning_rate": 1.2606813476378694e-05,
-      "loss": 0.8606,
+      "grad_norm": 8.566146850585938,
+      "learning_rate": 1.891022021456804e-05,
+      "loss": 0.7789,
       "step": 9820
     },
     {
       "epoch": 1.8501788067005458,
-      "grad_norm": 12.051008224487305,
-      "learning_rate": 1.2599284773197818e-05,
-      "loss": 0.8854,
+      "grad_norm": 13.785645484924316,
+      "learning_rate": 1.8898927159796726e-05,
+      "loss": 0.7233,
       "step": 9830
     },
     {
       "epoch": 1.852060982495765,
-      "grad_norm": 17.03795623779297,
-      "learning_rate": 1.2591756070016941e-05,
-      "loss": 1.0436,
+      "grad_norm": 15.743664741516113,
+      "learning_rate": 1.888763410502541e-05,
+      "loss": 0.9657,
       "step": 9840
     },
     {
       "epoch": 1.8539431582909844,
-      "grad_norm": 21.712039947509766,
-      "learning_rate": 1.2584227366836064e-05,
-      "loss": 0.8406,
+      "grad_norm": 32.00886535644531,
+      "learning_rate": 1.8876341050254095e-05,
+      "loss": 0.7379,
       "step": 9850
     },
     {
       "epoch": 1.8558253340862036,
-      "grad_norm": 6.554458141326904,
-      "learning_rate": 1.2576698663655187e-05,
-      "loss": 0.9822,
+      "grad_norm": 7.174389839172363,
+      "learning_rate": 1.8865047995482778e-05,
+      "loss": 0.8011,
       "step": 9860
     },
     {
       "epoch": 1.8577075098814229,
-      "grad_norm": 17.28873634338379,
-      "learning_rate": 1.256916996047431e-05,
-      "loss": 0.8904,
+      "grad_norm": 20.806581497192383,
+      "learning_rate": 1.8853754940711464e-05,
+      "loss": 0.8431,
       "step": 9870
     },
     {
       "epoch": 1.8595896856766423,
-      "grad_norm": 6.655860424041748,
-      "learning_rate": 1.2561641257293433e-05,
-      "loss": 1.0043,
+      "grad_norm": 21.621334075927734,
+      "learning_rate": 1.884246188594015e-05,
+      "loss": 0.9357,
       "step": 9880
     },
     {
       "epoch": 1.8614718614718615,
-      "grad_norm": 6.995543956756592,
-      "learning_rate": 1.2554112554112554e-05,
-      "loss": 1.1564,
+      "grad_norm": 4.900033473968506,
+      "learning_rate": 1.883116883116883e-05,
+      "loss": 1.0361,
       "step": 9890
     },
     {
       "epoch": 1.8633540372670807,
-      "grad_norm": 18.659198760986328,
-      "learning_rate": 1.2546583850931677e-05,
-      "loss": 1.0026,
+      "grad_norm": 19.732091903686523,
+      "learning_rate": 1.8819875776397515e-05,
+      "loss": 0.6999,
       "step": 9900
     },
     {
       "epoch": 1.8652362130623001,
-      "grad_norm": 5.457601070404053,
-      "learning_rate": 1.25390551477508e-05,
-      "loss": 1.1253,
+      "grad_norm": 41.449031829833984,
+      "learning_rate": 1.8808582721626198e-05,
+      "loss": 0.9272,
       "step": 9910
     },
     {
       "epoch": 1.8671183888575191,
-      "grad_norm": 21.024433135986328,
-      "learning_rate": 1.2531526444569924e-05,
-      "loss": 1.077,
+      "grad_norm": 21.400299072265625,
+      "learning_rate": 1.8797289666854884e-05,
+      "loss": 0.9431,
       "step": 9920
     },
     {
       "epoch": 1.8690005646527386,
-      "grad_norm": 14.646206855773926,
-      "learning_rate": 1.2523997741389047e-05,
-      "loss": 0.6792,
+      "grad_norm": 17.52521514892578,
+      "learning_rate": 1.878599661208357e-05,
+      "loss": 0.6384,
       "step": 9930
     },
     {
       "epoch": 1.870882740447958,
-      "grad_norm": 14.446063995361328,
-      "learning_rate": 1.251646903820817e-05,
-      "loss": 1.0064,
+      "grad_norm": 12.966445922851562,
+      "learning_rate": 1.8774703557312253e-05,
+      "loss": 0.899,
       "step": 9940
     },
     {
       "epoch": 1.872764916243177,
-      "grad_norm": 19.97396469116211,
-      "learning_rate": 1.2508940335027293e-05,
-      "loss": 0.9491,
+      "grad_norm": 23.314476013183594,
+      "learning_rate": 1.876341050254094e-05,
+      "loss": 0.7705,
       "step": 9950
     },
     {
       "epoch": 1.8746470920383964,
-      "grad_norm": 51.85739517211914,
-      "learning_rate": 1.2501411631846416e-05,
-      "loss": 0.9612,
+      "grad_norm": 14.737226486206055,
+      "learning_rate": 1.8752117447769625e-05,
+      "loss": 0.7806,
       "step": 9960
     },
     {
       "epoch": 1.8765292678336156,
-      "grad_norm": 4.516988277435303,
-      "learning_rate": 1.2493882928665539e-05,
-      "loss": 0.9424,
+      "grad_norm": 9.335084915161133,
+      "learning_rate": 1.8740824392998308e-05,
+      "loss": 0.8484,
       "step": 9970
     },
     {
       "epoch": 1.8784114436288348,
-      "grad_norm": 9.412343978881836,
-      "learning_rate": 1.248635422548466e-05,
-      "loss": 1.0825,
+      "grad_norm": 5.914954662322998,
+      "learning_rate": 1.872953133822699e-05,
+      "loss": 0.9675,
       "step": 9980
     },
     {
       "epoch": 1.8802936194240543,
-      "grad_norm": 8.06126880645752,
-      "learning_rate": 1.2478825522303783e-05,
-      "loss": 0.7844,
+      "grad_norm": 13.312799453735352,
+      "learning_rate": 1.8718238283455673e-05,
+      "loss": 0.7423,
       "step": 9990
     },
     {
       "epoch": 1.8821757952192735,
-      "grad_norm": 5.933414459228516,
-      "learning_rate": 1.2471296819122906e-05,
+      "grad_norm": 10.06735610961914,
+      "learning_rate": 1.870694522868436e-05,
       "loss": 1.0082,
       "step": 10000
     },
     {
       "epoch": 1.8840579710144927,
-      "grad_norm": 17.543575286865234,
-      "learning_rate": 1.2463768115942029e-05,
-      "loss": 0.8493,
+      "grad_norm": 13.502887725830078,
+      "learning_rate": 1.8695652173913045e-05,
+      "loss": 0.7395,
       "step": 10010
     },
     {
       "epoch": 1.8859401468097121,
-      "grad_norm": 10.805532455444336,
-      "learning_rate": 1.2456239412761154e-05,
-      "loss": 0.9843,
+      "grad_norm": 19.229324340820312,
+      "learning_rate": 1.8684359119141728e-05,
+      "loss": 1.0014,
       "step": 10020
     },
     {
       "epoch": 1.8878223226049313,
-      "grad_norm": 7.1772260665893555,
-      "learning_rate": 1.2448710709580277e-05,
-      "loss": 1.028,
+      "grad_norm": 15.201828002929688,
+      "learning_rate": 1.8673066064370414e-05,
+      "loss": 0.8707,
       "step": 10030
     },
     {
       "epoch": 1.8897044984001505,
-      "grad_norm": 20.371374130249023,
-      "learning_rate": 1.24411820063994e-05,
-      "loss": 0.8241,
+      "grad_norm": 28.766010284423828,
+      "learning_rate": 1.8661773009599097e-05,
+      "loss": 0.7835,
       "step": 10040
     },
     {
       "epoch": 1.89158667419537,
-      "grad_norm": 9.451379776000977,
-      "learning_rate": 1.2433653303218522e-05,
-      "loss": 1.057,
+      "grad_norm": 2.099621295928955,
+      "learning_rate": 1.8650479954827783e-05,
+      "loss": 0.8257,
       "step": 10050
     },
     {
       "epoch": 1.8934688499905892,
-      "grad_norm": 12.519644737243652,
-      "learning_rate": 1.2426124600037644e-05,
-      "loss": 0.8994,
+      "grad_norm": 11.689505577087402,
+      "learning_rate": 1.8639186900056465e-05,
+      "loss": 0.8311,
       "step": 10060
     },
     {
       "epoch": 1.8953510257858084,
-      "grad_norm": 2.6986730098724365,
-      "learning_rate": 1.2418595896856767e-05,
-      "loss": 0.7619,
+      "grad_norm": 1.5185855627059937,
+      "learning_rate": 1.8627893845285148e-05,
+      "loss": 0.7473,
       "step": 10070
     },
     {
       "epoch": 1.8972332015810278,
-      "grad_norm": 21.105154037475586,
-      "learning_rate": 1.241106719367589e-05,
-      "loss": 1.0232,
+      "grad_norm": 16.93794822692871,
+      "learning_rate": 1.8616600790513834e-05,
+      "loss": 0.8576,
       "step": 10080
     },
     {
       "epoch": 1.8991153773762468,
-      "grad_norm": 12.050444602966309,
-      "learning_rate": 1.2403538490495012e-05,
-      "loss": 0.7324,
+      "grad_norm": 12.918668746948242,
+      "learning_rate": 1.8605307735742517e-05,
+      "loss": 0.6406,
       "step": 10090
     },
     {
       "epoch": 1.9009975531714662,
-      "grad_norm": 19.549890518188477,
-      "learning_rate": 1.2396009787314135e-05,
-      "loss": 1.2509,
+      "grad_norm": 38.772605895996094,
+      "learning_rate": 1.8594014680971203e-05,
+      "loss": 1.0629,
       "step": 10100
     },
     {
       "epoch": 1.9028797289666854,
-      "grad_norm": 12.393698692321777,
-      "learning_rate": 1.238848108413326e-05,
-      "loss": 0.982,
+      "grad_norm": 8.743882179260254,
+      "learning_rate": 1.858272162619989e-05,
+      "loss": 0.8832,
       "step": 10110
     },
     {
       "epoch": 1.9047619047619047,
-      "grad_norm": 2.7038516998291016,
-      "learning_rate": 1.2380952380952383e-05,
-      "loss": 0.7445,
+      "grad_norm": 18.133054733276367,
+      "learning_rate": 1.8571428571428572e-05,
+      "loss": 0.5021,
       "step": 10120
     },
     {
       "epoch": 1.906644080557124,
-      "grad_norm": 14.256768226623535,
-      "learning_rate": 1.2373423677771506e-05,
-      "loss": 0.7698,
+      "grad_norm": 18.592235565185547,
+      "learning_rate": 1.8560135516657258e-05,
+      "loss": 0.6759,
       "step": 10130
     },
     {
       "epoch": 1.9085262563523433,
-      "grad_norm": 7.038768291473389,
-      "learning_rate": 1.2365894974590629e-05,
-      "loss": 0.6723,
+      "grad_norm": 18.375886917114258,
+      "learning_rate": 1.854884246188594e-05,
+      "loss": 0.5934,
       "step": 10140
     },
     {
       "epoch": 1.9104084321475625,
-      "grad_norm": 11.244647026062012,
-      "learning_rate": 1.235836627140975e-05,
-      "loss": 1.0177,
+      "grad_norm": 9.453605651855469,
+      "learning_rate": 1.8537549407114623e-05,
+      "loss": 0.8925,
       "step": 10150
     },
     {
       "epoch": 1.912290607942782,
-      "grad_norm": 10.397238731384277,
-      "learning_rate": 1.2350837568228873e-05,
-      "loss": 1.1339,
+      "grad_norm": 20.740285873413086,
+      "learning_rate": 1.852625635234331e-05,
+      "loss": 0.931,
       "step": 10160
     },
     {
       "epoch": 1.9141727837380011,
-      "grad_norm": 15.638545036315918,
-      "learning_rate": 1.2343308865047996e-05,
-      "loss": 1.1344,
+      "grad_norm": 21.86232566833496,
+      "learning_rate": 1.8514963297571992e-05,
+      "loss": 1.0436,
       "step": 10170
     },
     {
       "epoch": 1.9160549595332204,
-      "grad_norm": 42.80206298828125,
-      "learning_rate": 1.2335780161867119e-05,
-      "loss": 0.9122,
+      "grad_norm": 31.541776657104492,
+      "learning_rate": 1.8503670242800678e-05,
+      "loss": 0.8187,
       "step": 10180
     },
     {
       "epoch": 1.9179371353284398,
-      "grad_norm": 13.100602149963379,
-      "learning_rate": 1.2328251458686242e-05,
-      "loss": 0.7872,
+      "grad_norm": 3.2206337451934814,
+      "learning_rate": 1.849237718802936e-05,
+      "loss": 0.6882,
       "step": 10190
     },
     {
       "epoch": 1.919819311123659,
-      "grad_norm": 4.41287088394165,
-      "learning_rate": 1.2320722755505366e-05,
-      "loss": 0.6886,
+      "grad_norm": 1.3856621980667114,
+      "learning_rate": 1.8481084133258047e-05,
+      "loss": 0.6564,
       "step": 10200
     },
     {
       "epoch": 1.9217014869188782,
-      "grad_norm": 17.352832794189453,
-      "learning_rate": 1.2313194052324489e-05,
-      "loss": 0.7823,
+      "grad_norm": 45.16568374633789,
+      "learning_rate": 1.8469791078486733e-05,
+      "loss": 0.7004,
       "step": 10210
     },
     {
       "epoch": 1.9235836627140976,
-      "grad_norm": 8.47523021697998,
-      "learning_rate": 1.2305665349143612e-05,
-      "loss": 1.1246,
+      "grad_norm": 12.040491104125977,
+      "learning_rate": 1.8458498023715416e-05,
+      "loss": 0.8935,
       "step": 10220
     },
     {
       "epoch": 1.9254658385093166,
-      "grad_norm": 4.320629596710205,
-      "learning_rate": 1.2298136645962735e-05,
-      "loss": 1.0178,
+      "grad_norm": 1.0651041269302368,
+      "learning_rate": 1.84472049689441e-05,
+      "loss": 0.9316,
       "step": 10230
     },
     {
       "epoch": 1.927348014304536,
-      "grad_norm": 17.435626983642578,
-      "learning_rate": 1.2290607942781856e-05,
-      "loss": 1.2824,
+      "grad_norm": 24.65813636779785,
+      "learning_rate": 1.8435911914172784e-05,
+      "loss": 1.1372,
       "step": 10240
     },
     {
       "epoch": 1.9292301900997553,
-      "grad_norm": 18.203798294067383,
-      "learning_rate": 1.2283079239600979e-05,
-      "loss": 1.0708,
+      "grad_norm": 11.479782104492188,
+      "learning_rate": 1.8424618859401467e-05,
+      "loss": 1.0353,
       "step": 10250
     },
     {
       "epoch": 1.9311123658949745,
-      "grad_norm": 2.8385987281799316,
-      "learning_rate": 1.2275550536420102e-05,
-      "loss": 0.8138,
+      "grad_norm": 12.16601848602295,
+      "learning_rate": 1.8413325804630153e-05,
+      "loss": 0.7897,
       "step": 10260
     },
     {
       "epoch": 1.932994541690194,
-      "grad_norm": 17.48655128479004,
-      "learning_rate": 1.2268021833239225e-05,
-      "loss": 0.935,
+      "grad_norm": 17.336091995239258,
+      "learning_rate": 1.8402032749858836e-05,
+      "loss": 0.8251,
       "step": 10270
     },
     {
       "epoch": 1.9348767174854131,
-      "grad_norm": 7.490472793579102,
-      "learning_rate": 1.2260493130058348e-05,
-      "loss": 0.8238,
+      "grad_norm": 9.198282241821289,
+      "learning_rate": 1.8390739695087522e-05,
+      "loss": 0.7357,
       "step": 10280
     },
     {
       "epoch": 1.9367588932806323,
-      "grad_norm": 11.629676818847656,
-      "learning_rate": 1.2252964426877473e-05,
-      "loss": 0.816,
+      "grad_norm": 12.248185157775879,
+      "learning_rate": 1.8379446640316208e-05,
+      "loss": 0.7766,
       "step": 10290
     },
     {
       "epoch": 1.9386410690758518,
-      "grad_norm": 1.6730352640151978,
-      "learning_rate": 1.2245435723696595e-05,
-      "loss": 0.718,
+      "grad_norm": 0.7270705103874207,
+      "learning_rate": 1.836815358554489e-05,
+      "loss": 0.5565,
       "step": 10300
     },
     {
       "epoch": 1.940523244871071,
-      "grad_norm": 29.646865844726562,
-      "learning_rate": 1.2237907020515718e-05,
-      "loss": 0.948,
+      "grad_norm": 18.369525909423828,
+      "learning_rate": 1.8356860530773577e-05,
+      "loss": 1.054,
       "step": 10310
     },
     {
       "epoch": 1.9424054206662902,
-      "grad_norm": 3.246751070022583,
-      "learning_rate": 1.2230378317334841e-05,
-      "loss": 1.0785,
+      "grad_norm": 3.5889933109283447,
+      "learning_rate": 1.834556747600226e-05,
+      "loss": 1.0666,
       "step": 10320
     },
     {
       "epoch": 1.9442875964615096,
-      "grad_norm": 14.672032356262207,
-      "learning_rate": 1.2222849614153963e-05,
-      "loss": 0.6108,
+      "grad_norm": 16.404603958129883,
+      "learning_rate": 1.8334274421230942e-05,
+      "loss": 0.5469,
       "step": 10330
     },
     {
       "epoch": 1.9461697722567288,
-      "grad_norm": 8.470264434814453,
-      "learning_rate": 1.2215320910973085e-05,
-      "loss": 0.9098,
+      "grad_norm": 5.033247470855713,
+      "learning_rate": 1.8322981366459628e-05,
+      "loss": 0.7387,
       "step": 10340
     },
     {
       "epoch": 1.948051948051948,
-      "grad_norm": 6.679137706756592,
-      "learning_rate": 1.2207792207792208e-05,
-      "loss": 0.6887,
+      "grad_norm": 20.548690795898438,
+      "learning_rate": 1.831168831168831e-05,
+      "loss": 0.6346,
       "step": 10350
     },
     {
       "epoch": 1.9499341238471675,
-      "grad_norm": 17.44939613342285,
-      "learning_rate": 1.2200263504611331e-05,
-      "loss": 0.8049,
+      "grad_norm": 19.923683166503906,
+      "learning_rate": 1.8300395256916997e-05,
+      "loss": 0.6418,
       "step": 10360
     },
     {
       "epoch": 1.9518162996423865,
-      "grad_norm": 12.042229652404785,
-      "learning_rate": 1.2192734801430454e-05,
-      "loss": 0.6099,
+      "grad_norm": 14.779675483703613,
+      "learning_rate": 1.828910220214568e-05,
+      "loss": 0.6338,
       "step": 10370
     },
     {
       "epoch": 1.9536984754376059,
-      "grad_norm": 22.439245223999023,
-      "learning_rate": 1.2185206098249577e-05,
-      "loss": 0.78,
+      "grad_norm": 24.59194564819336,
+      "learning_rate": 1.8277809147374366e-05,
+      "loss": 0.6815,
       "step": 10380
     },
     {
       "epoch": 1.955580651232825,
-      "grad_norm": 9.354933738708496,
-      "learning_rate": 1.2177677395068702e-05,
-      "loss": 0.8881,
+      "grad_norm": 6.548442840576172,
+      "learning_rate": 1.8266516092603052e-05,
+      "loss": 0.9073,
       "step": 10390
     },
     {
       "epoch": 1.9574628270280443,
-      "grad_norm": 9.417488098144531,
-      "learning_rate": 1.2170148691887825e-05,
-      "loss": 0.8555,
+      "grad_norm": 25.888690948486328,
+      "learning_rate": 1.8255223037831734e-05,
+      "loss": 0.8705,
       "step": 10400
     },
     {
       "epoch": 1.9593450028232637,
-      "grad_norm": 13.155538558959961,
-      "learning_rate": 1.2162619988706946e-05,
-      "loss": 1.0589,
+      "grad_norm": 14.560491561889648,
+      "learning_rate": 1.8243929983060417e-05,
+      "loss": 1.0138,
       "step": 10410
     },
     {
       "epoch": 1.961227178618483,
-      "grad_norm": 8.894850730895996,
-      "learning_rate": 1.2155091285526069e-05,
-      "loss": 0.8917,
+      "grad_norm": 18.18019676208496,
+      "learning_rate": 1.82326369282891e-05,
+      "loss": 0.7988,
       "step": 10420
     },
     {
       "epoch": 1.9631093544137022,
-      "grad_norm": 4.504289627075195,
-      "learning_rate": 1.2147562582345192e-05,
-      "loss": 0.8084,
+      "grad_norm": 6.463456153869629,
+      "learning_rate": 1.8221343873517786e-05,
+      "loss": 0.6973,
       "step": 10430
     },
     {
       "epoch": 1.9649915302089216,
-      "grad_norm": 7.747999668121338,
-      "learning_rate": 1.2140033879164315e-05,
-      "loss": 0.7033,
+      "grad_norm": 6.2463226318359375,
+      "learning_rate": 1.8210050818746472e-05,
+      "loss": 0.6555,
       "step": 10440
     },
     {
       "epoch": 1.9668737060041408,
-      "grad_norm": 28.37241554260254,
-      "learning_rate": 1.2132505175983438e-05,
-      "loss": 0.7308,
+      "grad_norm": 18.08149528503418,
+      "learning_rate": 1.8198757763975155e-05,
+      "loss": 0.6245,
       "step": 10450
     },
     {
       "epoch": 1.96875588179936,
-      "grad_norm": 10.770339965820312,
-      "learning_rate": 1.212497647280256e-05,
-      "loss": 1.0313,
+      "grad_norm": 17.84767723083496,
+      "learning_rate": 1.818746470920384e-05,
+      "loss": 1.1495,
       "step": 10460
     },
     {
       "epoch": 1.9706380575945794,
-      "grad_norm": 16.21937370300293,
-      "learning_rate": 1.2117447769621683e-05,
-      "loss": 0.9229,
+      "grad_norm": 6.831356048583984,
+      "learning_rate": 1.8176171654432527e-05,
+      "loss": 0.8015,
       "step": 10470
     },
     {
       "epoch": 1.9725202333897986,
-      "grad_norm": 29.973360061645508,
-      "learning_rate": 1.2109919066440808e-05,
-      "loss": 0.972,
+      "grad_norm": 14.89473819732666,
+      "learning_rate": 1.816487859966121e-05,
+      "loss": 0.8459,
       "step": 10480
     },
     {
       "epoch": 1.9744024091850179,
-      "grad_norm": 34.55480194091797,
-      "learning_rate": 1.2102390363259931e-05,
-      "loss": 0.9822,
+      "grad_norm": 18.0239315032959,
+      "learning_rate": 1.8153585544889896e-05,
+      "loss": 0.891,
       "step": 10490
     },
     {
       "epoch": 1.9762845849802373,
-      "grad_norm": 18.8240909576416,
-      "learning_rate": 1.2094861660079052e-05,
-      "loss": 1.124,
+      "grad_norm": 15.742325782775879,
+      "learning_rate": 1.8142292490118575e-05,
+      "loss": 1.2079,
       "step": 10500
     },
     {
       "epoch": 1.9781667607754563,
-      "grad_norm": 5.787772178649902,
-      "learning_rate": 1.2087332956898175e-05,
-      "loss": 0.8912,
+      "grad_norm": 6.855963230133057,
+      "learning_rate": 1.813099943534726e-05,
+      "loss": 0.7506,
       "step": 10510
     },
     {
       "epoch": 1.9800489365706757,
-      "grad_norm": 11.275154113769531,
-      "learning_rate": 1.2079804253717298e-05,
-      "loss": 0.8765,
+      "grad_norm": 23.201900482177734,
+      "learning_rate": 1.8119706380575947e-05,
+      "loss": 0.854,
       "step": 10520
     },
     {
       "epoch": 1.981931112365895,
-      "grad_norm": 20.49886131286621,
-      "learning_rate": 1.2072275550536421e-05,
-      "loss": 0.9797,
+      "grad_norm": 11.044225692749023,
+      "learning_rate": 1.810841332580463e-05,
+      "loss": 0.9139,
       "step": 10530
     },
     {
       "epoch": 1.9838132881611141,
-      "grad_norm": 16.942197799682617,
-      "learning_rate": 1.2064746847355544e-05,
-      "loss": 0.8569,
+      "grad_norm": 10.52823257446289,
+      "learning_rate": 1.8097120271033316e-05,
+      "loss": 0.8894,
       "step": 10540
     },
     {
       "epoch": 1.9856954639563336,
-      "grad_norm": 7.426819801330566,
-      "learning_rate": 1.2057218144174667e-05,
-      "loss": 0.6804,
+      "grad_norm": 11.858381271362305,
+      "learning_rate": 1.8085827216262e-05,
+      "loss": 0.5292,
       "step": 10550
     },
     {
       "epoch": 1.9875776397515528,
-      "grad_norm": 16.51919174194336,
-      "learning_rate": 1.204968944099379e-05,
-      "loss": 0.6712,
+      "grad_norm": 20.706117630004883,
+      "learning_rate": 1.8074534161490685e-05,
+      "loss": 0.7082,
       "step": 10560
     },
     {
       "epoch": 1.989459815546772,
-      "grad_norm": 5.8685302734375,
-      "learning_rate": 1.2042160737812914e-05,
-      "loss": 0.9723,
+      "grad_norm": 9.068899154663086,
+      "learning_rate": 1.806324110671937e-05,
+      "loss": 0.9249,
       "step": 10570
     },
     {
       "epoch": 1.9913419913419914,
-      "grad_norm": 21.12563705444336,
-      "learning_rate": 1.2034632034632037e-05,
-      "loss": 0.8149,
+      "grad_norm": 12.875887870788574,
+      "learning_rate": 1.8051948051948053e-05,
+      "loss": 0.7443,
       "step": 10580
     },
     {
       "epoch": 1.9932241671372106,
-      "grad_norm": 18.672962188720703,
-      "learning_rate": 1.2027103331451157e-05,
-      "loss": 0.8327,
+      "grad_norm": 15.794068336486816,
+      "learning_rate": 1.8040654997176736e-05,
+      "loss": 0.7504,
       "step": 10590
     },
     {
       "epoch": 1.9951063429324298,
-      "grad_norm": 15.490560531616211,
-      "learning_rate": 1.2019574628270281e-05,
-      "loss": 0.8651,
+      "grad_norm": 9.319598197937012,
+      "learning_rate": 1.802936194240542e-05,
+      "loss": 0.7693,
       "step": 10600
     },
     {
       "epoch": 1.9969885187276493,
-      "grad_norm": 4.263712406158447,
-      "learning_rate": 1.2012045925089404e-05,
-      "loss": 0.9362,
+      "grad_norm": 7.584019660949707,
+      "learning_rate": 1.8018068887634105e-05,
+      "loss": 0.8374,
       "step": 10610
     },
     {
       "epoch": 1.9988706945228685,
-      "grad_norm": 11.25022029876709,
-      "learning_rate": 1.2004517221908527e-05,
-      "loss": 0.7861,
+      "grad_norm": 7.562906742095947,
+      "learning_rate": 1.800677583286279e-05,
+      "loss": 0.7839,
       "step": 10620
     },
     {
       "epoch": 2.0,
       "eval_accuracy": 0.9056,
-      "eval_loss": 0.4684549570083618,
-      "eval_runtime": 110.542,
-      "eval_samples_per_second": 67.848,
-      "eval_steps_per_second": 8.485,
+      "eval_loss": 0.3731466829776764,
+      "eval_runtime": 275.5436,
+      "eval_samples_per_second": 27.219,
+      "eval_steps_per_second": 3.404,
       "step": 10626
     },
     {
       "epoch": 2.0007528703180877,
-      "grad_norm": 24.8381404876709,
-      "learning_rate": 1.199698851872765e-05,
-      "loss": 0.8849,
+      "grad_norm": 26.831989288330078,
+      "learning_rate": 1.7995482778091474e-05,
+      "loss": 0.8433,
       "step": 10630
     },
     {
       "epoch": 2.002635046113307,
-      "grad_norm": 16.988882064819336,
-      "learning_rate": 1.1989459815546773e-05,
-      "loss": 0.8274,
+      "grad_norm": 14.433839797973633,
+      "learning_rate": 1.798418972332016e-05,
+      "loss": 0.874,
       "step": 10640
     },
     {
       "epoch": 2.004517221908526,
-      "grad_norm": 8.939057350158691,
-      "learning_rate": 1.1981931112365896e-05,
-      "loss": 1.0752,
+      "grad_norm": 6.329076290130615,
+      "learning_rate": 1.7972896668548842e-05,
+      "loss": 0.953,
       "step": 10650
     },
     {
       "epoch": 2.0063993977037455,
-      "grad_norm": 7.0386128425598145,
-      "learning_rate": 1.1974402409185019e-05,
-      "loss": 0.8478,
+      "grad_norm": 28.34707260131836,
+      "learning_rate": 1.796160361377753e-05,
+      "loss": 0.7231,
       "step": 10660
     },
     {
       "epoch": 2.008281573498965,
-      "grad_norm": 7.87394380569458,
-      "learning_rate": 1.196687370600414e-05,
-      "loss": 1.1662,
+      "grad_norm": 6.1005778312683105,
+      "learning_rate": 1.795031055900621e-05,
+      "loss": 1.0041,
       "step": 10670
     },
     {
       "epoch": 2.010163749294184,
-      "grad_norm": 10.221199035644531,
-      "learning_rate": 1.1959345002823263e-05,
-      "loss": 0.9248,
+      "grad_norm": 5.388144493103027,
+      "learning_rate": 1.7939017504234894e-05,
+      "loss": 0.71,
       "step": 10680
     },
     {
       "epoch": 2.0120459250894034,
-      "grad_norm": 10.589498519897461,
-      "learning_rate": 1.1951816299642388e-05,
-      "loss": 0.5548,
+      "grad_norm": 22.38304901123047,
+      "learning_rate": 1.792772444946358e-05,
+      "loss": 0.5202,
       "step": 10690
     },
     {
       "epoch": 2.013928100884623,
-      "grad_norm": 4.692480087280273,
-      "learning_rate": 1.194428759646151e-05,
-      "loss": 0.9416,
+      "grad_norm": 6.131063938140869,
+      "learning_rate": 1.7916431394692266e-05,
+      "loss": 0.8542,
       "step": 10700
     },
     {
       "epoch": 2.015810276679842,
-      "grad_norm": 12.261754989624023,
-      "learning_rate": 1.1936758893280634e-05,
-      "loss": 1.0959,
+      "grad_norm": 18.350811004638672,
+      "learning_rate": 1.790513833992095e-05,
+      "loss": 0.8689,
       "step": 10710
     },
     {
       "epoch": 2.0176924524750612,
-      "grad_norm": 3.541266441345215,
-      "learning_rate": 1.1929230190099756e-05,
-      "loss": 0.9546,
+      "grad_norm": 1.005018949508667,
+      "learning_rate": 1.7893845285149635e-05,
+      "loss": 0.8021,
       "step": 10720
     },
     {
       "epoch": 2.0195746282702802,
-      "grad_norm": 23.496295928955078,
-      "learning_rate": 1.192170148691888e-05,
-      "loss": 0.9216,
+      "grad_norm": 11.301158905029297,
+      "learning_rate": 1.7882552230378317e-05,
+      "loss": 0.7946,
       "step": 10730
     },
     {
       "epoch": 2.0214568040654997,
-      "grad_norm": 35.39616394042969,
-      "learning_rate": 1.1914172783738002e-05,
-      "loss": 0.8117,
+      "grad_norm": 22.24584197998047,
+      "learning_rate": 1.7871259175607003e-05,
+      "loss": 0.8225,
       "step": 10740
     },
     {
       "epoch": 2.023338979860719,
-      "grad_norm": 5.2047247886657715,
-      "learning_rate": 1.1906644080557125e-05,
-      "loss": 1.0543,
+      "grad_norm": 4.584529399871826,
+      "learning_rate": 1.785996612083569e-05,
+      "loss": 0.9226,
       "step": 10750
     },
     {
       "epoch": 2.025221155655938,
-      "grad_norm": 16.47121238708496,
-      "learning_rate": 1.1899115377376246e-05,
-      "loss": 0.9308,
+      "grad_norm": 9.03012466430664,
+      "learning_rate": 1.784867306606437e-05,
+      "loss": 0.7807,
       "step": 10760
     },
     {
       "epoch": 2.0271033314511575,
-      "grad_norm": 11.237469673156738,
-      "learning_rate": 1.189158667419537e-05,
-      "loss": 0.5998,
+      "grad_norm": 14.892461776733398,
+      "learning_rate": 1.7837380011293055e-05,
+      "loss": 0.5403,
       "step": 10770
     },
     {
       "epoch": 2.028985507246377,
-      "grad_norm": 15.104199409484863,
-      "learning_rate": 1.1884057971014494e-05,
-      "loss": 0.7011,
+      "grad_norm": 18.233808517456055,
+      "learning_rate": 1.7826086956521738e-05,
+      "loss": 0.7445,
       "step": 10780
     },
     {
       "epoch": 2.030867683041596,
-      "grad_norm": 12.535137176513672,
-      "learning_rate": 1.1876529267833617e-05,
-      "loss": 0.7919,
+      "grad_norm": 7.108969211578369,
+      "learning_rate": 1.7814793901750424e-05,
+      "loss": 0.7014,
       "step": 10790
     },
     {
       "epoch": 2.0327498588368154,
-      "grad_norm": 3.7836413383483887,
-      "learning_rate": 1.186900056465274e-05,
-      "loss": 0.7689,
+      "grad_norm": 3.6049118041992188,
+      "learning_rate": 1.780350084697911e-05,
+      "loss": 0.6114,
       "step": 10800
     },
     {
       "epoch": 2.034632034632035,
-      "grad_norm": 40.574642181396484,
-      "learning_rate": 1.1861471861471863e-05,
-      "loss": 0.9367,
+      "grad_norm": 8.403626441955566,
+      "learning_rate": 1.7792207792207792e-05,
+      "loss": 0.856,
       "step": 10810
     },
     {
       "epoch": 2.036514210427254,
-      "grad_norm": 14.20531177520752,
-      "learning_rate": 1.1853943158290986e-05,
-      "loss": 0.6538,
+      "grad_norm": 3.540233850479126,
+      "learning_rate": 1.778091473743648e-05,
+      "loss": 0.5674,
       "step": 10820
     },
     {
       "epoch": 2.038396386222473,
-      "grad_norm": 11.008893966674805,
-      "learning_rate": 1.1846414455110109e-05,
-      "loss": 0.9406,
+      "grad_norm": 13.208251953125,
+      "learning_rate": 1.776962168266516e-05,
+      "loss": 0.829,
       "step": 10830
     },
     {
       "epoch": 2.0402785620176926,
-      "grad_norm": 13.851802825927734,
-      "learning_rate": 1.1838885751929231e-05,
-      "loss": 0.8025,
+      "grad_norm": 3.715716600418091,
+      "learning_rate": 1.7758328627893847e-05,
+      "loss": 0.7736,
       "step": 10840
     },
     {
       "epoch": 2.0421607378129116,
-      "grad_norm": 11.819645881652832,
-      "learning_rate": 1.1831357048748353e-05,
-      "loss": 0.7047,
+      "grad_norm": 14.193865776062012,
+      "learning_rate": 1.774703557312253e-05,
+      "loss": 0.4945,
       "step": 10850
     },
     {
       "epoch": 2.044042913608131,
-      "grad_norm": 0.6497629284858704,
-      "learning_rate": 1.1823828345567476e-05,
-      "loss": 0.8094,
+      "grad_norm": 0.2950342893600464,
+      "learning_rate": 1.7735742518351213e-05,
+      "loss": 0.7292,
       "step": 10860
     },
     {
       "epoch": 2.04592508940335,
-      "grad_norm": 10.60029411315918,
-      "learning_rate": 1.18162996423866e-05,
-      "loss": 0.7603,
+      "grad_norm": 6.488024711608887,
+      "learning_rate": 1.77244494635799e-05,
+      "loss": 0.7146,
       "step": 10870
     },
     {
       "epoch": 2.0478072651985695,
-      "grad_norm": 10.798954010009766,
-      "learning_rate": 1.1808770939205723e-05,
-      "loss": 0.7551,
+      "grad_norm": 19.793678283691406,
+      "learning_rate": 1.771315640880858e-05,
+      "loss": 0.6736,
       "step": 10880
     },
     {
       "epoch": 2.049689440993789,
-      "grad_norm": 1.9502023458480835,
-      "learning_rate": 1.1801242236024846e-05,
-      "loss": 0.7977,
+      "grad_norm": 12.970588684082031,
+      "learning_rate": 1.7701863354037267e-05,
+      "loss": 0.8303,
       "step": 10890
     },
     {
       "epoch": 2.051571616789008,
-      "grad_norm": 7.010662078857422,
-      "learning_rate": 1.1793713532843969e-05,
-      "loss": 1.0109,
+      "grad_norm": 8.980803489685059,
+      "learning_rate": 1.7690570299265954e-05,
+      "loss": 0.9072,
       "step": 10900
     },
     {
       "epoch": 2.0534537925842273,
-      "grad_norm": 7.077285289764404,
-      "learning_rate": 1.1786184829663092e-05,
-      "loss": 0.72,
+      "grad_norm": 3.427503824234009,
+      "learning_rate": 1.7679277244494636e-05,
+      "loss": 0.6361,
       "step": 10910
     },
     {
       "epoch": 2.0553359683794468,
-      "grad_norm": 10.952967643737793,
-      "learning_rate": 1.1778656126482215e-05,
-      "loss": 0.7789,
+      "grad_norm": 9.372367858886719,
+      "learning_rate": 1.7667984189723322e-05,
+      "loss": 0.601,
       "step": 10920
     },
     {
       "epoch": 2.0572181441746658,
-      "grad_norm": 1.8068546056747437,
-      "learning_rate": 1.1771127423301338e-05,
-      "loss": 0.5414,
+      "grad_norm": 31.908103942871094,
+      "learning_rate": 1.7656691134952005e-05,
+      "loss": 0.4396,
       "step": 10930
     },
     {
       "epoch": 2.059100319969885,
-      "grad_norm": 9.510477066040039,
-      "learning_rate": 1.1763598720120459e-05,
-      "loss": 0.7369,
+      "grad_norm": 10.805863380432129,
+      "learning_rate": 1.7645398080180688e-05,
+      "loss": 0.6055,
       "step": 10940
     },
     {
       "epoch": 2.0609824957651046,
-      "grad_norm": 15.91817855834961,
-      "learning_rate": 1.1756070016939582e-05,
-      "loss": 0.8333,
+      "grad_norm": 9.006129264831543,
+      "learning_rate": 1.7634105025409374e-05,
+      "loss": 0.7486,
       "step": 10950
     },
     {
       "epoch": 2.0628646715603236,
-      "grad_norm": 8.938312530517578,
-      "learning_rate": 1.1748541313758705e-05,
-      "loss": 0.7071,
+      "grad_norm": 12.054622650146484,
+      "learning_rate": 1.7622811970638056e-05,
+      "loss": 0.5213,
       "step": 10960
     },
     {
       "epoch": 2.064746847355543,
-      "grad_norm": 5.337620258331299,
-      "learning_rate": 1.174101261057783e-05,
-      "loss": 0.5705,
+      "grad_norm": 4.848597526550293,
+      "learning_rate": 1.7611518915866743e-05,
+      "loss": 0.4883,
       "step": 10970
     },
     {
       "epoch": 2.0666290231507625,
-      "grad_norm": 16.76058578491211,
-      "learning_rate": 1.1733483907396952e-05,
-      "loss": 0.8773,
+      "grad_norm": 0.7347248792648315,
+      "learning_rate": 1.760022586109543e-05,
+      "loss": 0.6849,
       "step": 10980
     },
     {
       "epoch": 2.0685111989459815,
-      "grad_norm": 5.9690985679626465,
-      "learning_rate": 1.1725955204216075e-05,
-      "loss": 0.8072,
+      "grad_norm": 3.6684467792510986,
+      "learning_rate": 1.758893280632411e-05,
+      "loss": 0.7872,
       "step": 10990
     },
     {
       "epoch": 2.070393374741201,
-      "grad_norm": 1.3085390329360962,
-      "learning_rate": 1.1718426501035198e-05,
-      "loss": 1.1046,
+      "grad_norm": 1.874008059501648,
+      "learning_rate": 1.7577639751552797e-05,
+      "loss": 0.8974,
       "step": 11000
     },
     {
       "epoch": 2.0722755505364203,
-      "grad_norm": 10.01681137084961,
-      "learning_rate": 1.1710897797854321e-05,
-      "loss": 1.0356,
+      "grad_norm": 8.359776496887207,
+      "learning_rate": 1.756634669678148e-05,
+      "loss": 0.9378,
       "step": 11010
     },
     {
       "epoch": 2.0741577263316393,
-      "grad_norm": 16.690166473388672,
-      "learning_rate": 1.1703369094673442e-05,
-      "loss": 0.4895,
+      "grad_norm": 8.968391418457031,
+      "learning_rate": 1.7555053642010163e-05,
+      "loss": 0.404,
       "step": 11020
     },
     {
       "epoch": 2.0760399021268587,
-      "grad_norm": 13.10949993133545,
-      "learning_rate": 1.1695840391492565e-05,
-      "loss": 0.63,
+      "grad_norm": 40.07794952392578,
+      "learning_rate": 1.754376058723885e-05,
+      "loss": 0.5055,
       "step": 11030
     },
     {
       "epoch": 2.0779220779220777,
-      "grad_norm": 12.242857933044434,
-      "learning_rate": 1.1688311688311688e-05,
-      "loss": 0.6262,
+      "grad_norm": 45.67797088623047,
+      "learning_rate": 1.753246753246753e-05,
+      "loss": 0.5616,
       "step": 11040
     },
     {
       "epoch": 2.079804253717297,
-      "grad_norm": 7.687251567840576,
-      "learning_rate": 1.1680782985130811e-05,
-      "loss": 0.715,
+      "grad_norm": 11.17763900756836,
+      "learning_rate": 1.7521174477696218e-05,
+      "loss": 0.6244,
       "step": 11050
     },
     {
       "epoch": 2.0816864295125166,
-      "grad_norm": 32.919715881347656,
-      "learning_rate": 1.1673254281949936e-05,
-      "loss": 0.7352,
+      "grad_norm": 11.508498191833496,
+      "learning_rate": 1.75098814229249e-05,
+      "loss": 0.6413,
       "step": 11060
     },
     {
       "epoch": 2.0835686053077356,
-      "grad_norm": 4.056097507476807,
-      "learning_rate": 1.1665725578769059e-05,
-      "loss": 1.0831,
+      "grad_norm": 7.054535865783691,
+      "learning_rate": 1.7498588368153586e-05,
+      "loss": 1.0319,
       "step": 11070
     },
     {
       "epoch": 2.085450781102955,
-      "grad_norm": 2.21692156791687,
-      "learning_rate": 1.1658196875588182e-05,
-      "loss": 0.8726,
+      "grad_norm": 1.5207102298736572,
+      "learning_rate": 1.7487295313382272e-05,
+      "loss": 0.6158,
       "step": 11080
     },
     {
       "epoch": 2.0873329568981744,
-      "grad_norm": 19.382230758666992,
-      "learning_rate": 1.1650668172407305e-05,
-      "loss": 0.7071,
+      "grad_norm": 6.750110149383545,
+      "learning_rate": 1.7476002258610955e-05,
+      "loss": 0.5906,
       "step": 11090
     },
     {
       "epoch": 2.0892151326933934,
-      "grad_norm": 5.093869209289551,
-      "learning_rate": 1.1643139469226427e-05,
-      "loss": 0.8924,
+      "grad_norm": 5.071254730224609,
+      "learning_rate": 1.746470920383964e-05,
+      "loss": 0.7611,
       "step": 11100
     },
     {
       "epoch": 2.091097308488613,
-      "grad_norm": 19.17929458618164,
-      "learning_rate": 1.1635610766045549e-05,
-      "loss": 0.8245,
+      "grad_norm": 8.039938926696777,
+      "learning_rate": 1.745341614906832e-05,
+      "loss": 0.7571,
       "step": 11110
     },
     {
       "epoch": 2.0929794842838323,
-      "grad_norm": 26.134408950805664,
-      "learning_rate": 1.1628082062864672e-05,
-      "loss": 0.7664,
+      "grad_norm": 9.071053504943848,
+      "learning_rate": 1.7442123094297007e-05,
+      "loss": 0.6185,
       "step": 11120
     },
     {
       "epoch": 2.0948616600790513,
-      "grad_norm": 22.83806800842285,
-      "learning_rate": 1.1620553359683795e-05,
-      "loss": 0.901,
+      "grad_norm": 7.90115213394165,
+      "learning_rate": 1.7430830039525693e-05,
+      "loss": 0.8708,
       "step": 11130
     },
     {
       "epoch": 2.0967438358742707,
-      "grad_norm": 2.848191976547241,
-      "learning_rate": 1.1613024656502917e-05,
-      "loss": 0.8783,
+      "grad_norm": 1.540299415588379,
+      "learning_rate": 1.7419536984754375e-05,
+      "loss": 0.6717,
       "step": 11140
     },
     {
       "epoch": 2.09862601166949,
-      "grad_norm": 3.205889940261841,
-      "learning_rate": 1.1605495953322042e-05,
-      "loss": 0.8159,
+      "grad_norm": 5.848174095153809,
+      "learning_rate": 1.740824392998306e-05,
+      "loss": 0.785,
       "step": 11150
     },
     {
       "epoch": 2.100508187464709,
-      "grad_norm": 12.333131790161133,
-      "learning_rate": 1.1597967250141165e-05,
-      "loss": 0.7015,
+      "grad_norm": 12.827478408813477,
+      "learning_rate": 1.7396950875211744e-05,
+      "loss": 0.6168,
       "step": 11160
     },
     {
       "epoch": 2.1023903632599286,
-      "grad_norm": 3.75451922416687,
-      "learning_rate": 1.1590438546960288e-05,
-      "loss": 0.794,
+      "grad_norm": 2.8471341133117676,
+      "learning_rate": 1.738565782044043e-05,
+      "loss": 0.6963,
       "step": 11170
     },
     {
       "epoch": 2.1042725390551476,
-      "grad_norm": 13.827683448791504,
-      "learning_rate": 1.158290984377941e-05,
-      "loss": 0.7039,
+      "grad_norm": 12.310620307922363,
+      "learning_rate": 1.7374364765669116e-05,
+      "loss": 0.6679,
       "step": 11180
     },
     {
       "epoch": 2.106154714850367,
-      "grad_norm": 1.8480725288391113,
-      "learning_rate": 1.1575381140598534e-05,
-      "loss": 0.8531,
+      "grad_norm": 2.6820571422576904,
+      "learning_rate": 1.73630717108978e-05,
+      "loss": 0.7162,
       "step": 11190
     },
     {
       "epoch": 2.1080368906455864,
-      "grad_norm": 13.693140029907227,
-      "learning_rate": 1.1567852437417655e-05,
-      "loss": 1.3104,
+      "grad_norm": 28.728363037109375,
+      "learning_rate": 1.735177865612648e-05,
+      "loss": 1.1478,
       "step": 11200
     },
     {
       "epoch": 2.1099190664408054,
-      "grad_norm": 11.603228569030762,
-      "learning_rate": 1.1560323734236778e-05,
-      "loss": 1.1264,
+      "grad_norm": 13.076791763305664,
+      "learning_rate": 1.7340485601355168e-05,
+      "loss": 0.9537,
       "step": 11210
     },
     {
       "epoch": 2.111801242236025,
-      "grad_norm": 9.185226440429688,
-      "learning_rate": 1.15527950310559e-05,
-      "loss": 0.9391,
+      "grad_norm": 22.759828567504883,
+      "learning_rate": 1.732919254658385e-05,
+      "loss": 0.8617,
       "step": 11220
     },
     {
       "epoch": 2.1136834180312443,
-      "grad_norm": 17.381254196166992,
-      "learning_rate": 1.1545266327875024e-05,
-      "loss": 0.6598,
+      "grad_norm": 9.096440315246582,
+      "learning_rate": 1.7317899491812536e-05,
+      "loss": 0.5249,
       "step": 11230
     },
     {
       "epoch": 2.1155655938264633,
-      "grad_norm": 8.61845874786377,
-      "learning_rate": 1.1537737624694147e-05,
-      "loss": 0.7259,
+      "grad_norm": 14.975255012512207,
+      "learning_rate": 1.730660643704122e-05,
+      "loss": 0.6067,
       "step": 11240
     },
     {
       "epoch": 2.1174477696216827,
-      "grad_norm": 9.44906997680664,
-      "learning_rate": 1.1530208921513271e-05,
-      "loss": 0.6785,
+      "grad_norm": 16.731473922729492,
+      "learning_rate": 1.7295313382269905e-05,
+      "loss": 0.588,
       "step": 11250
     },
     {
       "epoch": 2.119329945416902,
-      "grad_norm": 19.07695770263672,
-      "learning_rate": 1.1522680218332394e-05,
-      "loss": 0.7864,
+      "grad_norm": 37.46254348754883,
+      "learning_rate": 1.728402032749859e-05,
+      "loss": 0.6994,
       "step": 11260
     },
     {
       "epoch": 2.121212121212121,
-      "grad_norm": 16.53068733215332,
-      "learning_rate": 1.1515151515151517e-05,
-      "loss": 0.7151,
+      "grad_norm": 4.683788299560547,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 0.5429,
       "step": 11270
     },
     {
       "epoch": 2.1230942970073405,
-      "grad_norm": 3.438575506210327,
-      "learning_rate": 1.1507622811970638e-05,
-      "loss": 0.6242,
+      "grad_norm": 1.7957135438919067,
+      "learning_rate": 1.7261434217955957e-05,
+      "loss": 0.4845,
       "step": 11280
     },
     {
       "epoch": 2.12497647280256,
-      "grad_norm": 15.59850025177002,
-      "learning_rate": 1.1500094108789761e-05,
-      "loss": 0.9812,
+      "grad_norm": 13.624578475952148,
+      "learning_rate": 1.725014116318464e-05,
+      "loss": 0.892,
       "step": 11290
     },
     {
       "epoch": 2.126858648597779,
-      "grad_norm": 37.894386291503906,
-      "learning_rate": 1.1492565405608884e-05,
-      "loss": 0.8772,
+      "grad_norm": 40.41400909423828,
+      "learning_rate": 1.7238848108413325e-05,
+      "loss": 0.8509,
       "step": 11300
     },
     {
       "epoch": 2.1287408243929984,
-      "grad_norm": 5.406034469604492,
-      "learning_rate": 1.1485036702428007e-05,
-      "loss": 0.8852,
+      "grad_norm": 16.918149948120117,
+      "learning_rate": 1.722755505364201e-05,
+      "loss": 0.9205,
       "step": 11310
     },
     {
       "epoch": 2.1306230001882174,
-      "grad_norm": 39.29657745361328,
-      "learning_rate": 1.147750799924713e-05,
-      "loss": 1.0962,
+      "grad_norm": 35.62562561035156,
+      "learning_rate": 1.7216261998870694e-05,
+      "loss": 1.0122,
       "step": 11320
     },
     {
       "epoch": 2.132505175983437,
-      "grad_norm": 8.793197631835938,
-      "learning_rate": 1.1469979296066253e-05,
-      "loss": 0.89,
+      "grad_norm": 10.24116325378418,
+      "learning_rate": 1.720496894409938e-05,
+      "loss": 0.7243,
       "step": 11330
     },
     {
       "epoch": 2.1343873517786562,
-      "grad_norm": 11.008149147033691,
-      "learning_rate": 1.1462450592885378e-05,
-      "loss": 0.9019,
+      "grad_norm": 20.230663299560547,
+      "learning_rate": 1.7193675889328063e-05,
+      "loss": 0.8808,
       "step": 11340
     },
     {
       "epoch": 2.1362695275738752,
-      "grad_norm": 9.002307891845703,
-      "learning_rate": 1.14549218897045e-05,
-      "loss": 1.0197,
+      "grad_norm": 11.159217834472656,
+      "learning_rate": 1.718238283455675e-05,
+      "loss": 0.9238,
       "step": 11350
     },
     {
       "epoch": 2.1381517033690947,
-      "grad_norm": 16.297245025634766,
-      "learning_rate": 1.1447393186523623e-05,
-      "loss": 1.0305,
+      "grad_norm": 15.367325782775879,
+      "learning_rate": 1.7171089779785435e-05,
+      "loss": 0.837,
       "step": 11360
     },
     {
       "epoch": 2.140033879164314,
-      "grad_norm": 16.021282196044922,
-      "learning_rate": 1.1439864483342745e-05,
-      "loss": 0.7957,
+      "grad_norm": 6.264622688293457,
+      "learning_rate": 1.7159796725014114e-05,
+      "loss": 0.7584,
       "step": 11370
     },
     {
       "epoch": 2.141916054959533,
-      "grad_norm": 12.168352127075195,
-      "learning_rate": 1.1432335780161868e-05,
-      "loss": 0.7707,
+      "grad_norm": 7.990375518798828,
+      "learning_rate": 1.71485036702428e-05,
+      "loss": 0.5522,
       "step": 11380
     },
     {
       "epoch": 2.1437982307547525,
-      "grad_norm": 28.62824058532715,
-      "learning_rate": 1.142480707698099e-05,
-      "loss": 0.9935,
+      "grad_norm": 13.449748039245605,
+      "learning_rate": 1.7137210615471483e-05,
+      "loss": 0.9335,
       "step": 11390
     },
     {
       "epoch": 2.145680406549972,
-      "grad_norm": 8.890032768249512,
-      "learning_rate": 1.1417278373800113e-05,
-      "loss": 0.7974,
+      "grad_norm": 16.00578498840332,
+      "learning_rate": 1.712591756070017e-05,
+      "loss": 0.6726,
       "step": 11400
     },
     {
       "epoch": 2.147562582345191,
-      "grad_norm": 7.006708145141602,
-      "learning_rate": 1.1409749670619236e-05,
-      "loss": 1.1413,
+      "grad_norm": 6.167961120605469,
+      "learning_rate": 1.7114624505928855e-05,
+      "loss": 0.9023,
       "step": 11410
     },
     {
       "epoch": 2.1494447581404104,
-      "grad_norm": 11.208281517028809,
-      "learning_rate": 1.140222096743836e-05,
-      "loss": 0.8733,
+      "grad_norm": 16.115100860595703,
+      "learning_rate": 1.7103331451157538e-05,
+      "loss": 0.9606,
       "step": 11420
     },
     {
       "epoch": 2.15132693393563,
-      "grad_norm": 37.206809997558594,
-      "learning_rate": 1.1394692264257484e-05,
-      "loss": 0.7413,
+      "grad_norm": 30.645904541015625,
+      "learning_rate": 1.7092038396386224e-05,
+      "loss": 0.6028,
       "step": 11430
     },
     {
       "epoch": 2.153209109730849,
-      "grad_norm": 4.180663585662842,
-      "learning_rate": 1.1387163561076607e-05,
-      "loss": 0.9974,
+      "grad_norm": 4.448369026184082,
+      "learning_rate": 1.7080745341614907e-05,
+      "loss": 0.9826,
       "step": 11440
     },
     {
       "epoch": 2.155091285526068,
-      "grad_norm": 13.185440063476562,
-      "learning_rate": 1.137963485789573e-05,
-      "loss": 1.164,
+      "grad_norm": 13.802371978759766,
+      "learning_rate": 1.7069452286843593e-05,
+      "loss": 1.0515,
       "step": 11450
     },
     {
       "epoch": 2.1569734613212876,
-      "grad_norm": 14.27194595336914,
-      "learning_rate": 1.1372106154714851e-05,
-      "loss": 0.6064,
+      "grad_norm": 18.508493423461914,
+      "learning_rate": 1.7058159232072276e-05,
+      "loss": 0.5478,
       "step": 11460
     },
     {
       "epoch": 2.1588556371165066,
-      "grad_norm": 34.04643249511719,
-      "learning_rate": 1.1364577451533974e-05,
-      "loss": 0.6796,
+      "grad_norm": 18.791305541992188,
+      "learning_rate": 1.7046866177300958e-05,
+      "loss": 0.6688,
       "step": 11470
     },
     {
       "epoch": 2.160737812911726,
-      "grad_norm": 19.426414489746094,
-      "learning_rate": 1.1357048748353097e-05,
-      "loss": 0.806,
+      "grad_norm": 13.487887382507324,
+      "learning_rate": 1.7035573122529644e-05,
+      "loss": 0.7173,
       "step": 11480
     },
     {
       "epoch": 2.162619988706945,
-      "grad_norm": 8.01048469543457,
-      "learning_rate": 1.134952004517222e-05,
-      "loss": 0.7656,
+      "grad_norm": 7.567160129547119,
+      "learning_rate": 1.702428006775833e-05,
+      "loss": 0.5889,
       "step": 11490
     },
     {
       "epoch": 2.1645021645021645,
-      "grad_norm": 16.02117919921875,
-      "learning_rate": 1.1341991341991343e-05,
-      "loss": 0.7809,
+      "grad_norm": 14.993462562561035,
+      "learning_rate": 1.7012987012987013e-05,
+      "loss": 0.6631,
       "step": 11500
     },
     {
       "epoch": 2.166384340297384,
-      "grad_norm": 4.275322914123535,
-      "learning_rate": 1.1334462638810466e-05,
-      "loss": 0.7501,
+      "grad_norm": 2.703808069229126,
+      "learning_rate": 1.70016939582157e-05,
+      "loss": 0.8147,
       "step": 11510
     },
     {
       "epoch": 2.168266516092603,
-      "grad_norm": 26.736955642700195,
-      "learning_rate": 1.132693393562959e-05,
-      "loss": 0.7723,
+      "grad_norm": 13.842448234558105,
+      "learning_rate": 1.6990400903444382e-05,
+      "loss": 0.6228,
       "step": 11520
     },
     {
       "epoch": 2.1701486918878223,
-      "grad_norm": 7.618874549865723,
-      "learning_rate": 1.1319405232448713e-05,
-      "loss": 0.713,
+      "grad_norm": 14.048020362854004,
+      "learning_rate": 1.6979107848673068e-05,
+      "loss": 0.6811,
       "step": 11530
     },
     {
       "epoch": 2.1720308676830418,
-      "grad_norm": 10.420034408569336,
-      "learning_rate": 1.1311876529267836e-05,
-      "loss": 0.9617,
+      "grad_norm": 9.474082946777344,
+      "learning_rate": 1.6967814793901754e-05,
+      "loss": 1.0404,
       "step": 11540
     },
     {
       "epoch": 2.1739130434782608,
-      "grad_norm": 8.347007751464844,
-      "learning_rate": 1.1304347826086957e-05,
-      "loss": 0.7585,
+      "grad_norm": 7.542745113372803,
+      "learning_rate": 1.6956521739130433e-05,
+      "loss": 0.7817,
       "step": 11550
     },
     {
       "epoch": 2.17579521927348,
-      "grad_norm": 20.805870056152344,
-      "learning_rate": 1.129681912290608e-05,
-      "loss": 0.9854,
+      "grad_norm": 14.078716278076172,
+      "learning_rate": 1.694522868435912e-05,
+      "loss": 0.9704,
       "step": 11560
     },
     {
       "epoch": 2.1776773950686996,
-      "grad_norm": 9.123351097106934,
-      "learning_rate": 1.1289290419725203e-05,
-      "loss": 0.8321,
+      "grad_norm": 19.203222274780273,
+      "learning_rate": 1.6933935629587802e-05,
+      "loss": 0.6797,
       "step": 11570
     },
     {
       "epoch": 2.1795595708639186,
-      "grad_norm": 11.314193725585938,
-      "learning_rate": 1.1281761716544326e-05,
-      "loss": 0.84,
+      "grad_norm": 10.718687057495117,
+      "learning_rate": 1.6922642574816488e-05,
+      "loss": 0.7273,
       "step": 11580
     },
     {
       "epoch": 2.181441746659138,
-      "grad_norm": 10.206731796264648,
-      "learning_rate": 1.1274233013363449e-05,
-      "loss": 0.5685,
+      "grad_norm": 7.656870365142822,
+      "learning_rate": 1.6911349520045174e-05,
+      "loss": 0.4612,
       "step": 11590
     },
     {
       "epoch": 2.183323922454357,
-      "grad_norm": 11.790088653564453,
-      "learning_rate": 1.1266704310182572e-05,
-      "loss": 0.7315,
+      "grad_norm": 9.046762466430664,
+      "learning_rate": 1.6900056465273857e-05,
+      "loss": 0.815,
       "step": 11600
     },
     {
       "epoch": 2.1852060982495765,
-      "grad_norm": 15.779023170471191,
-      "learning_rate": 1.1259175607001695e-05,
-      "loss": 0.745,
+      "grad_norm": 16.06904411315918,
+      "learning_rate": 1.6888763410502543e-05,
+      "loss": 0.7,
       "step": 11610
     },
     {
       "epoch": 2.187088274044796,
-      "grad_norm": 17.365673065185547,
-      "learning_rate": 1.125164690382082e-05,
-      "loss": 0.981,
+      "grad_norm": 17.846696853637695,
+      "learning_rate": 1.6877470355731226e-05,
+      "loss": 0.8311,
       "step": 11620
     },
     {
       "epoch": 2.188970449840015,
-      "grad_norm": 1.291931390762329,
-      "learning_rate": 1.1244118200639939e-05,
-      "loss": 1.0725,
+      "grad_norm": 0.43426594138145447,
+      "learning_rate": 1.686617730095991e-05,
+      "loss": 0.901,
       "step": 11630
     },
     {
       "epoch": 2.1908526256352343,
-      "grad_norm": 17.065105438232422,
-      "learning_rate": 1.1236589497459064e-05,
-      "loss": 1.073,
+      "grad_norm": 21.463172912597656,
+      "learning_rate": 1.6854884246188594e-05,
+      "loss": 1.1355,
       "step": 11640
     },
     {
       "epoch": 2.1927348014304537,
-      "grad_norm": 4.217034816741943,
-      "learning_rate": 1.1229060794278186e-05,
-      "loss": 0.6659,
+      "grad_norm": 4.620266914367676,
+      "learning_rate": 1.6843591191417277e-05,
+      "loss": 0.6449,
       "step": 11650
     },
     {
       "epoch": 2.1946169772256727,
-      "grad_norm": 1.8660165071487427,
-      "learning_rate": 1.122153209109731e-05,
-      "loss": 0.6897,
+      "grad_norm": 1.4968249797821045,
+      "learning_rate": 1.6832298136645963e-05,
+      "loss": 0.6416,
       "step": 11660
     },
     {
       "epoch": 2.196499153020892,
-      "grad_norm": 7.8330817222595215,
-      "learning_rate": 1.1214003387916432e-05,
-      "loss": 0.7756,
+      "grad_norm": 5.055369853973389,
+      "learning_rate": 1.6821005081874646e-05,
+      "loss": 0.6069,
       "step": 11670
     },
     {
       "epoch": 2.1983813288161116,
-      "grad_norm": 19.692419052124023,
-      "learning_rate": 1.1206474684735555e-05,
-      "loss": 1.0101,
+      "grad_norm": 17.229318618774414,
+      "learning_rate": 1.6809712027103332e-05,
+      "loss": 0.7881,
       "step": 11680
     },
     {
       "epoch": 2.2002635046113306,
-      "grad_norm": 43.92416000366211,
-      "learning_rate": 1.1198945981554678e-05,
-      "loss": 1.0972,
+      "grad_norm": 26.358625411987305,
+      "learning_rate": 1.6798418972332018e-05,
+      "loss": 0.9306,
       "step": 11690
     },
     {
       "epoch": 2.20214568040655,
-      "grad_norm": 1.8937373161315918,
-      "learning_rate": 1.1191417278373801e-05,
-      "loss": 0.8669,
+      "grad_norm": 14.272749900817871,
+      "learning_rate": 1.67871259175607e-05,
+      "loss": 0.7933,
       "step": 11700
     },
     {
       "epoch": 2.2040278562017694,
-      "grad_norm": 20.478147506713867,
-      "learning_rate": 1.1183888575192926e-05,
-      "loss": 1.0169,
+      "grad_norm": 17.075315475463867,
+      "learning_rate": 1.6775832862789387e-05,
+      "loss": 0.9922,
       "step": 11710
     },
     {
       "epoch": 2.2059100319969884,
-      "grad_norm": 18.37138557434082,
-      "learning_rate": 1.1176359872012045e-05,
-      "loss": 0.6301,
+      "grad_norm": 18.468908309936523,
+      "learning_rate": 1.676453980801807e-05,
+      "loss": 0.549,
       "step": 11720
     },
     {
       "epoch": 2.207792207792208,
-      "grad_norm": 5.762269496917725,
-      "learning_rate": 1.116883116883117e-05,
-      "loss": 0.9702,
+      "grad_norm": 7.144861698150635,
+      "learning_rate": 1.6753246753246752e-05,
+      "loss": 0.7335,
       "step": 11730
     },
     {
       "epoch": 2.2096743835874273,
-      "grad_norm": 23.213031768798828,
-      "learning_rate": 1.1161302465650293e-05,
-      "loss": 0.6551,
+      "grad_norm": 20.43572235107422,
+      "learning_rate": 1.6741953698475438e-05,
+      "loss": 0.4813,
       "step": 11740
     },
     {
       "epoch": 2.2115565593826463,
-      "grad_norm": 19.10317039489746,
-      "learning_rate": 1.1153773762469416e-05,
-      "loss": 1.0178,
+      "grad_norm": 8.59123420715332,
+      "learning_rate": 1.673066064370412e-05,
+      "loss": 0.8275,
       "step": 11750
     },
     {
       "epoch": 2.2134387351778657,
-      "grad_norm": 15.511590957641602,
-      "learning_rate": 1.1146245059288539e-05,
-      "loss": 0.799,
+      "grad_norm": 17.360458374023438,
+      "learning_rate": 1.6719367588932807e-05,
+      "loss": 0.7514,
       "step": 11760
     },
     {
       "epoch": 2.2153209109730847,
-      "grad_norm": 11.412705421447754,
-      "learning_rate": 1.1138716356107661e-05,
-      "loss": 1.0609,
+      "grad_norm": 10.042230606079102,
+      "learning_rate": 1.6708074534161493e-05,
+      "loss": 1.0016,
       "step": 11770
     },
     {
       "epoch": 2.217203086768304,
-      "grad_norm": 26.8338680267334,
-      "learning_rate": 1.1131187652926784e-05,
-      "loss": 0.9019,
+      "grad_norm": 19.570775985717773,
+      "learning_rate": 1.6696781479390176e-05,
+      "loss": 0.8327,
       "step": 11780
     },
     {
       "epoch": 2.2190852625635236,
-      "grad_norm": 9.526796340942383,
-      "learning_rate": 1.1123658949745907e-05,
-      "loss": 0.6681,
+      "grad_norm": 14.94032096862793,
+      "learning_rate": 1.6685488424618862e-05,
+      "loss": 0.6058,
       "step": 11790
     },
     {
       "epoch": 2.2209674383587426,
-      "grad_norm": 21.135456085205078,
-      "learning_rate": 1.1116130246565032e-05,
-      "loss": 0.5308,
+      "grad_norm": 15.301405906677246,
+      "learning_rate": 1.6674195369847545e-05,
+      "loss": 0.4892,
       "step": 11800
     },
     {
       "epoch": 2.222849614153962,
-      "grad_norm": 21.3120059967041,
-      "learning_rate": 1.1108601543384151e-05,
-      "loss": 0.8701,
+      "grad_norm": 18.80755615234375,
+      "learning_rate": 1.6662902315076227e-05,
+      "loss": 0.7236,
       "step": 11810
     },
     {
       "epoch": 2.2247317899491814,
-      "grad_norm": 8.938432693481445,
-      "learning_rate": 1.1101072840203274e-05,
-      "loss": 0.8294,
+      "grad_norm": 15.558100700378418,
+      "learning_rate": 1.6651609260304913e-05,
+      "loss": 0.7063,
       "step": 11820
     },
     {
       "epoch": 2.2266139657444004,
-      "grad_norm": 21.0710506439209,
-      "learning_rate": 1.1093544137022399e-05,
-      "loss": 0.5845,
+      "grad_norm": 13.449424743652344,
+      "learning_rate": 1.6640316205533596e-05,
+      "loss": 0.5049,
       "step": 11830
     },
     {
       "epoch": 2.22849614153962,
-      "grad_norm": 13.667383193969727,
-      "learning_rate": 1.1086015433841522e-05,
-      "loss": 1.137,
+      "grad_norm": 10.576664924621582,
+      "learning_rate": 1.6629023150762282e-05,
+      "loss": 1.2078,
       "step": 11840
     },
     {
       "epoch": 2.2303783173348393,
-      "grad_norm": 7.945625305175781,
-      "learning_rate": 1.1078486730660645e-05,
-      "loss": 0.774,
+      "grad_norm": 4.189072608947754,
+      "learning_rate": 1.6617730095990965e-05,
+      "loss": 0.7146,
       "step": 11850
     },
     {
       "epoch": 2.2322604931300583,
-      "grad_norm": 13.387373924255371,
-      "learning_rate": 1.1070958027479768e-05,
-      "loss": 0.6535,
+      "grad_norm": 10.161325454711914,
+      "learning_rate": 1.660643704121965e-05,
+      "loss": 0.6925,
       "step": 11860
     },
     {
       "epoch": 2.2341426689252777,
-      "grad_norm": 16.70855140686035,
-      "learning_rate": 1.106342932429889e-05,
-      "loss": 0.6098,
+      "grad_norm": 29.488740921020508,
+      "learning_rate": 1.6595143986448337e-05,
+      "loss": 0.6859,
       "step": 11870
     },
     {
       "epoch": 2.2360248447204967,
-      "grad_norm": 19.224872589111328,
-      "learning_rate": 1.1055900621118014e-05,
-      "loss": 0.8751,
+      "grad_norm": 6.638284206390381,
+      "learning_rate": 1.658385093167702e-05,
+      "loss": 0.7455,
       "step": 11880
     },
     {
       "epoch": 2.237907020515716,
-      "grad_norm": 12.4450044631958,
-      "learning_rate": 1.1048371917937135e-05,
-      "loss": 1.0692,
+      "grad_norm": 14.819140434265137,
+      "learning_rate": 1.6572557876905702e-05,
+      "loss": 0.9682,
       "step": 11890
     },
     {
       "epoch": 2.2397891963109355,
-      "grad_norm": 19.086071014404297,
-      "learning_rate": 1.1040843214756258e-05,
-      "loss": 0.9096,
+      "grad_norm": 16.32223129272461,
+      "learning_rate": 1.6561264822134385e-05,
+      "loss": 0.8409,
       "step": 11900
     },
     {
       "epoch": 2.2416713721061545,
-      "grad_norm": 18.77231788635254,
-      "learning_rate": 1.103331451157538e-05,
-      "loss": 0.8296,
+      "grad_norm": 33.57245635986328,
+      "learning_rate": 1.654997176736307e-05,
+      "loss": 0.8191,
       "step": 11910
     },
     {
       "epoch": 2.243553547901374,
-      "grad_norm": 10.611273765563965,
-      "learning_rate": 1.1025785808394505e-05,
-      "loss": 0.7729,
+      "grad_norm": 11.583626747131348,
+      "learning_rate": 1.6538678712591757e-05,
+      "loss": 0.7672,
       "step": 11920
     },
     {
       "epoch": 2.2454357236965934,
-      "grad_norm": 19.948476791381836,
-      "learning_rate": 1.1018257105213628e-05,
-      "loss": 0.6645,
+      "grad_norm": 19.42552947998047,
+      "learning_rate": 1.652738565782044e-05,
+      "loss": 0.6141,
       "step": 11930
     },
     {
       "epoch": 2.2473178994918124,
-      "grad_norm": 7.409848690032959,
-      "learning_rate": 1.1010728402032751e-05,
-      "loss": 0.6516,
+      "grad_norm": 8.740943908691406,
+      "learning_rate": 1.6516092603049126e-05,
+      "loss": 0.5423,
       "step": 11940
     },
     {
       "epoch": 2.249200075287032,
-      "grad_norm": 6.684335231781006,
-      "learning_rate": 1.1003199698851874e-05,
-      "loss": 0.986,
+      "grad_norm": 9.475835800170898,
+      "learning_rate": 1.650479954827781e-05,
+      "loss": 0.9297,
       "step": 11950
     },
     {
       "epoch": 2.2510822510822512,
-      "grad_norm": 21.333091735839844,
-      "learning_rate": 1.0995670995670997e-05,
-      "loss": 0.7682,
+      "grad_norm": 10.173590660095215,
+      "learning_rate": 1.6493506493506495e-05,
+      "loss": 0.6418,
       "step": 11960
     },
     {
       "epoch": 2.2529644268774702,
-      "grad_norm": 4.745196342468262,
-      "learning_rate": 1.098814229249012e-05,
-      "loss": 0.488,
+      "grad_norm": 5.159122467041016,
+      "learning_rate": 1.648221343873518e-05,
+      "loss": 0.4525,
       "step": 11970
     },
     {
       "epoch": 2.2548466026726897,
-      "grad_norm": 24.217205047607422,
-      "learning_rate": 1.0980613589309241e-05,
-      "loss": 0.5174,
+      "grad_norm": 20.905466079711914,
+      "learning_rate": 1.647092038396386e-05,
+      "loss": 0.4453,
       "step": 11980
     },
     {
       "epoch": 2.256728778467909,
-      "grad_norm": 9.572256088256836,
-      "learning_rate": 1.0973084886128364e-05,
-      "loss": 0.7882,
+      "grad_norm": 14.282583236694336,
+      "learning_rate": 1.6459627329192546e-05,
+      "loss": 0.6531,
       "step": 11990
     },
     {
       "epoch": 2.258610954263128,
-      "grad_norm": 2.003173828125,
-      "learning_rate": 1.0965556182947487e-05,
-      "loss": 0.6915,
+      "grad_norm": 4.331676006317139,
+      "learning_rate": 1.6448334274421232e-05,
+      "loss": 0.6144,
       "step": 12000
     },
     {
       "epoch": 2.2604931300583475,
-      "grad_norm": 39.94436264038086,
-      "learning_rate": 1.0958027479766612e-05,
-      "loss": 0.9829,
+      "grad_norm": 18.63337516784668,
+      "learning_rate": 1.6437041219649915e-05,
+      "loss": 0.9553,
       "step": 12010
     },
     {
       "epoch": 2.262375305853567,
-      "grad_norm": 16.391490936279297,
-      "learning_rate": 1.0950498776585735e-05,
-      "loss": 0.792,
+      "grad_norm": 8.563776016235352,
+      "learning_rate": 1.64257481648786e-05,
+      "loss": 0.5918,
       "step": 12020
     },
     {
       "epoch": 2.264257481648786,
-      "grad_norm": 19.83889389038086,
-      "learning_rate": 1.0942970073404857e-05,
-      "loss": 0.8692,
+      "grad_norm": 26.9168758392334,
+      "learning_rate": 1.6414455110107284e-05,
+      "loss": 0.9051,
       "step": 12030
     },
     {
       "epoch": 2.2661396574440054,
-      "grad_norm": 10.875959396362305,
-      "learning_rate": 1.093544137022398e-05,
-      "loss": 0.6063,
+      "grad_norm": 20.256752014160156,
+      "learning_rate": 1.640316205533597e-05,
+      "loss": 0.6102,
       "step": 12040
     },
     {
       "epoch": 2.2680218332392243,
-      "grad_norm": 0.8151828646659851,
-      "learning_rate": 1.0927912667043103e-05,
-      "loss": 0.5955,
+      "grad_norm": 0.3493950366973877,
+      "learning_rate": 1.6391869000564656e-05,
+      "loss": 0.5125,
       "step": 12050
     },
     {
       "epoch": 2.269904009034444,
-      "grad_norm": 4.661582946777344,
-      "learning_rate": 1.0920383963862226e-05,
-      "loss": 0.8415,
+      "grad_norm": 2.2542970180511475,
+      "learning_rate": 1.638057594579334e-05,
+      "loss": 0.7501,
       "step": 12060
     },
     {
       "epoch": 2.271786184829663,
-      "grad_norm": 6.517459392547607,
-      "learning_rate": 1.0912855260681347e-05,
-      "loss": 0.703,
+      "grad_norm": 8.313566207885742,
+      "learning_rate": 1.636928289102202e-05,
+      "loss": 0.6996,
       "step": 12070
     },
     {
       "epoch": 2.273668360624882,
-      "grad_norm": 8.205318450927734,
-      "learning_rate": 1.090532655750047e-05,
-      "loss": 0.6762,
+      "grad_norm": 7.590768337249756,
+      "learning_rate": 1.6357989836250704e-05,
+      "loss": 0.6913,
       "step": 12080
     },
     {
       "epoch": 2.2755505364201016,
-      "grad_norm": 46.9409065246582,
-      "learning_rate": 1.0897797854319593e-05,
-      "loss": 0.6102,
+      "grad_norm": 17.979164123535156,
+      "learning_rate": 1.634669678147939e-05,
+      "loss": 0.5275,
       "step": 12090
     },
     {
       "epoch": 2.277432712215321,
-      "grad_norm": 23.41012954711914,
-      "learning_rate": 1.0890269151138718e-05,
-      "loss": 0.787,
+      "grad_norm": 30.397066116333008,
+      "learning_rate": 1.6335403726708076e-05,
+      "loss": 0.8555,
       "step": 12100
     },
     {
       "epoch": 2.27931488801054,
-      "grad_norm": 7.724615097045898,
-      "learning_rate": 1.088274044795784e-05,
-      "loss": 0.9145,
+      "grad_norm": 8.03955364227295,
+      "learning_rate": 1.632411067193676e-05,
+      "loss": 0.6765,
       "step": 12110
     },
     {
       "epoch": 2.2811970638057595,
-      "grad_norm": 10.711122512817383,
-      "learning_rate": 1.0875211744776964e-05,
-      "loss": 0.7992,
+      "grad_norm": 8.310486793518066,
+      "learning_rate": 1.6312817617165445e-05,
+      "loss": 0.8106,
       "step": 12120
     },
     {
       "epoch": 2.283079239600979,
-      "grad_norm": 6.124819278717041,
-      "learning_rate": 1.0867683041596087e-05,
-      "loss": 0.6344,
+      "grad_norm": 5.86186408996582,
+      "learning_rate": 1.6301524562394127e-05,
+      "loss": 0.4987,
       "step": 12130
     },
     {
       "epoch": 2.284961415396198,
-      "grad_norm": 27.855012893676758,
-      "learning_rate": 1.086015433841521e-05,
-      "loss": 0.6282,
+      "grad_norm": 19.821392059326172,
+      "learning_rate": 1.6290231507622814e-05,
+      "loss": 0.5728,
       "step": 12140
     },
     {
       "epoch": 2.2868435911914173,
-      "grad_norm": 7.3847832679748535,
-      "learning_rate": 1.0852625635234333e-05,
-      "loss": 0.7378,
+      "grad_norm": 12.544803619384766,
+      "learning_rate": 1.62789384528515e-05,
+      "loss": 0.7435,
       "step": 12150
     },
     {
       "epoch": 2.2887257669866363,
-      "grad_norm": 13.684946060180664,
-      "learning_rate": 1.0845096932053454e-05,
-      "loss": 0.5045,
+      "grad_norm": 13.145238876342773,
+      "learning_rate": 1.626764539808018e-05,
+      "loss": 0.4866,
       "step": 12160
     },
     {
       "epoch": 2.2906079427818558,
-      "grad_norm": 14.742804527282715,
-      "learning_rate": 1.0837568228872577e-05,
-      "loss": 1.0028,
+      "grad_norm": 15.785924911499023,
+      "learning_rate": 1.6256352343308865e-05,
+      "loss": 0.8673,
       "step": 12170
     },
     {
       "epoch": 2.292490118577075,
-      "grad_norm": 10.077812194824219,
-      "learning_rate": 1.08300395256917e-05,
-      "loss": 0.7794,
+      "grad_norm": 9.898990631103516,
+      "learning_rate": 1.6245059288537548e-05,
+      "loss": 0.7766,
       "step": 12180
     },
     {
       "epoch": 2.2943722943722946,
-      "grad_norm": 33.63645553588867,
-      "learning_rate": 1.0822510822510823e-05,
-      "loss": 0.9631,
+      "grad_norm": 14.689043998718262,
+      "learning_rate": 1.6233766233766234e-05,
+      "loss": 0.9227,
       "step": 12190
     },
     {
       "epoch": 2.2962544701675136,
-      "grad_norm": 9.247525215148926,
-      "learning_rate": 1.0814982119329947e-05,
-      "loss": 0.5459,
+      "grad_norm": 6.211693286895752,
+      "learning_rate": 1.622247317899492e-05,
+      "loss": 0.5012,
       "step": 12200
     },
     {
       "epoch": 2.298136645962733,
-      "grad_norm": 5.322352409362793,
-      "learning_rate": 1.080745341614907e-05,
-      "loss": 0.8692,
+      "grad_norm": 6.014758586883545,
+      "learning_rate": 1.6211180124223603e-05,
+      "loss": 0.8023,
       "step": 12210
     },
     {
       "epoch": 2.300018821757952,
-      "grad_norm": 29.076679229736328,
-      "learning_rate": 1.0799924712968193e-05,
-      "loss": 0.8797,
+      "grad_norm": 32.03548812866211,
+      "learning_rate": 1.619988706945229e-05,
+      "loss": 0.86,
       "step": 12220
     },
     {
       "epoch": 2.3019009975531715,
-      "grad_norm": 18.286100387573242,
-      "learning_rate": 1.0792396009787316e-05,
-      "loss": 0.9282,
+      "grad_norm": 25.236791610717773,
+      "learning_rate": 1.6188594014680975e-05,
+      "loss": 0.7572,
       "step": 12230
     },
     {
       "epoch": 2.303783173348391,
-      "grad_norm": 8.712194442749023,
-      "learning_rate": 1.0784867306606437e-05,
-      "loss": 0.7751,
+      "grad_norm": 7.06341552734375,
+      "learning_rate": 1.6177300959909654e-05,
+      "loss": 0.6443,
       "step": 12240
     },
     {
       "epoch": 2.30566534914361,
-      "grad_norm": 2.8198611736297607,
-      "learning_rate": 1.077733860342556e-05,
-      "loss": 0.7057,
+      "grad_norm": 3.819842576980591,
+      "learning_rate": 1.616600790513834e-05,
+      "loss": 0.6632,
       "step": 12250
     },
     {
       "epoch": 2.3075475249388293,
-      "grad_norm": 23.1691951751709,
-      "learning_rate": 1.0769809900244683e-05,
-      "loss": 0.8707,
+      "grad_norm": 11.308211326599121,
+      "learning_rate": 1.6154714850367023e-05,
+      "loss": 0.7541,
       "step": 12260
     },
     {
       "epoch": 2.3094297007340487,
-      "grad_norm": 10.105401992797852,
-      "learning_rate": 1.0762281197063806e-05,
-      "loss": 0.9284,
+      "grad_norm": 9.683237075805664,
+      "learning_rate": 1.614342179559571e-05,
+      "loss": 0.9261,
       "step": 12270
     },
     {
       "epoch": 2.3113118765292677,
-      "grad_norm": 29.39499855041504,
-      "learning_rate": 1.0754752493882929e-05,
-      "loss": 0.6555,
+      "grad_norm": 21.996740341186523,
+      "learning_rate": 1.6132128740824395e-05,
+      "loss": 0.6662,
       "step": 12280
     },
     {
       "epoch": 2.313194052324487,
-      "grad_norm": 8.851414680480957,
-      "learning_rate": 1.0747223790702053e-05,
-      "loss": 0.9129,
+      "grad_norm": 8.742904663085938,
+      "learning_rate": 1.6120835686053078e-05,
+      "loss": 0.7921,
       "step": 12290
     },
     {
       "epoch": 2.3150762281197066,
-      "grad_norm": 1.137457251548767,
-      "learning_rate": 1.0739695087521176e-05,
-      "loss": 0.9343,
+      "grad_norm": 1.3880571126937866,
+      "learning_rate": 1.6109542631281764e-05,
+      "loss": 0.8751,
       "step": 12300
     },
     {
       "epoch": 2.3169584039149256,
-      "grad_norm": 25.591732025146484,
-      "learning_rate": 1.07321663843403e-05,
-      "loss": 0.682,
+      "grad_norm": 26.320383071899414,
+      "learning_rate": 1.6098249576510446e-05,
+      "loss": 0.6595,
       "step": 12310
     },
     {
       "epoch": 2.318840579710145,
-      "grad_norm": 26.291501998901367,
-      "learning_rate": 1.0724637681159422e-05,
-      "loss": 0.8744,
+      "grad_norm": 34.07208251953125,
+      "learning_rate": 1.6086956521739132e-05,
+      "loss": 0.7587,
       "step": 12320
     },
     {
       "epoch": 2.320722755505364,
-      "grad_norm": 15.204377174377441,
-      "learning_rate": 1.0717108977978543e-05,
-      "loss": 0.6868,
+      "grad_norm": 7.523997783660889,
+      "learning_rate": 1.6075663466967815e-05,
+      "loss": 0.6733,
       "step": 12330
     },
     {
       "epoch": 2.3226049313005834,
-      "grad_norm": 26.76555633544922,
-      "learning_rate": 1.0709580274797666e-05,
-      "loss": 0.8418,
+      "grad_norm": 18.36374855041504,
+      "learning_rate": 1.6064370412196498e-05,
+      "loss": 0.8033,
       "step": 12340
     },
     {
       "epoch": 2.324487107095803,
-      "grad_norm": 22.4197940826416,
-      "learning_rate": 1.070205157161679e-05,
-      "loss": 1.1285,
+      "grad_norm": 13.632719039916992,
+      "learning_rate": 1.6053077357425184e-05,
+      "loss": 0.9909,
       "step": 12350
     },
     {
       "epoch": 2.326369282891022,
-      "grad_norm": 20.703449249267578,
-      "learning_rate": 1.0694522868435912e-05,
-      "loss": 0.841,
+      "grad_norm": 13.183788299560547,
+      "learning_rate": 1.6041784302653867e-05,
+      "loss": 0.8184,
       "step": 12360
     },
     {
       "epoch": 2.3282514586862413,
-      "grad_norm": 27.92598533630371,
-      "learning_rate": 1.0686994165255035e-05,
-      "loss": 0.572,
+      "grad_norm": 2.855896472930908,
+      "learning_rate": 1.6030491247882553e-05,
+      "loss": 0.5908,
       "step": 12370
     },
     {
       "epoch": 2.3301336344814607,
-      "grad_norm": 1.4635416269302368,
-      "learning_rate": 1.067946546207416e-05,
-      "loss": 0.7061,
+      "grad_norm": 2.077853202819824,
+      "learning_rate": 1.601919819311124e-05,
+      "loss": 0.5688,
       "step": 12380
     },
     {
       "epoch": 2.3320158102766797,
-      "grad_norm": 12.686742782592773,
-      "learning_rate": 1.0671936758893283e-05,
-      "loss": 0.7331,
+      "grad_norm": 10.118651390075684,
+      "learning_rate": 1.600790513833992e-05,
+      "loss": 0.7142,
       "step": 12390
     },
     {
       "epoch": 2.333897986071899,
-      "grad_norm": 4.224462509155273,
-      "learning_rate": 1.0664408055712406e-05,
-      "loss": 1.2657,
+      "grad_norm": 11.305803298950195,
+      "learning_rate": 1.5996612083568607e-05,
+      "loss": 1.1076,
       "step": 12400
     },
     {
       "epoch": 2.3357801618671186,
-      "grad_norm": 13.47118091583252,
-      "learning_rate": 1.0656879352531528e-05,
-      "loss": 0.8943,
+      "grad_norm": 10.361531257629395,
+      "learning_rate": 1.598531902879729e-05,
+      "loss": 0.7602,
       "step": 12410
     },
     {
       "epoch": 2.3376623376623376,
-      "grad_norm": 17.81936264038086,
-      "learning_rate": 1.064935064935065e-05,
-      "loss": 0.7193,
+      "grad_norm": 20.7995662689209,
+      "learning_rate": 1.5974025974025973e-05,
+      "loss": 0.6195,
       "step": 12420
     },
     {
       "epoch": 2.339544513457557,
-      "grad_norm": 8.82054328918457,
-      "learning_rate": 1.0641821946169773e-05,
-      "loss": 0.9461,
+      "grad_norm": 8.287638664245605,
+      "learning_rate": 1.596273291925466e-05,
+      "loss": 0.8501,
       "step": 12430
     },
     {
       "epoch": 2.341426689252776,
-      "grad_norm": 26.209758758544922,
-      "learning_rate": 1.0634293242988896e-05,
-      "loss": 0.8565,
+      "grad_norm": 20.09409523010254,
+      "learning_rate": 1.595143986448334e-05,
+      "loss": 0.8378,
       "step": 12440
     },
     {
       "epoch": 2.3433088650479954,
-      "grad_norm": 13.844365119934082,
-      "learning_rate": 1.0626764539808018e-05,
-      "loss": 0.9119,
+      "grad_norm": 26.83869171142578,
+      "learning_rate": 1.5940146809712028e-05,
+      "loss": 0.9183,
       "step": 12450
     },
     {
       "epoch": 2.345191040843215,
-      "grad_norm": 0.3456774353981018,
-      "learning_rate": 1.0619235836627141e-05,
-      "loss": 1.1152,
+      "grad_norm": 0.30550795793533325,
+      "learning_rate": 1.592885375494071e-05,
+      "loss": 1.0637,
       "step": 12460
     },
     {
       "epoch": 2.3470732166384343,
-      "grad_norm": 1.4162580966949463,
-      "learning_rate": 1.0611707133446264e-05,
-      "loss": 0.6846,
+      "grad_norm": 2.1787874698638916,
+      "learning_rate": 1.5917560700169396e-05,
+      "loss": 0.7166,
       "step": 12470
     },
     {
       "epoch": 2.3489553924336533,
-      "grad_norm": 9.784071922302246,
-      "learning_rate": 1.0604178430265389e-05,
-      "loss": 1.1308,
+      "grad_norm": 9.78334903717041,
+      "learning_rate": 1.5906267645398083e-05,
+      "loss": 1.08,
       "step": 12480
     },
     {
       "epoch": 2.3508375682288727,
-      "grad_norm": 65.54403686523438,
-      "learning_rate": 1.0596649727084512e-05,
-      "loss": 0.6059,
+      "grad_norm": 26.472990036010742,
+      "learning_rate": 1.5894974590626765e-05,
+      "loss": 0.5742,
       "step": 12490
     },
     {
       "epoch": 2.3527197440240917,
-      "grad_norm": 18.23605728149414,
-      "learning_rate": 1.0589121023903633e-05,
-      "loss": 0.5438,
+      "grad_norm": 7.039992332458496,
+      "learning_rate": 1.5883681535855448e-05,
+      "loss": 0.4755,
       "step": 12500
     },
     {
       "epoch": 2.354601919819311,
-      "grad_norm": 19.891637802124023,
-      "learning_rate": 1.0581592320722756e-05,
-      "loss": 0.9351,
+      "grad_norm": 12.98085880279541,
+      "learning_rate": 1.5872388481084134e-05,
+      "loss": 0.9198,
       "step": 12510
     },
     {
       "epoch": 2.3564840956145305,
-      "grad_norm": 20.713233947753906,
-      "learning_rate": 1.0574063617541879e-05,
-      "loss": 0.6646,
+      "grad_norm": 19.38686752319336,
+      "learning_rate": 1.5861095426312817e-05,
+      "loss": 0.512,
       "step": 12520
     },
     {
       "epoch": 2.3583662714097495,
-      "grad_norm": 14.681770324707031,
-      "learning_rate": 1.0566534914361002e-05,
-      "loss": 0.8591,
+      "grad_norm": 16.51802635192871,
+      "learning_rate": 1.5849802371541503e-05,
+      "loss": 0.7799,
       "step": 12530
     },
     {
       "epoch": 2.360248447204969,
-      "grad_norm": 6.22154426574707,
-      "learning_rate": 1.0559006211180125e-05,
-      "loss": 0.8724,
+      "grad_norm": 5.628615379333496,
+      "learning_rate": 1.5838509316770185e-05,
+      "loss": 0.8057,
       "step": 12540
     },
     {
       "epoch": 2.3621306230001884,
-      "grad_norm": 28.546422958374023,
-      "learning_rate": 1.0551477507999248e-05,
-      "loss": 0.8389,
+      "grad_norm": 25.433387756347656,
+      "learning_rate": 1.582721626199887e-05,
+      "loss": 0.6807,
       "step": 12550
     },
     {
       "epoch": 2.3640127987954074,
-      "grad_norm": 8.232221603393555,
-      "learning_rate": 1.054394880481837e-05,
-      "loss": 0.7999,
+      "grad_norm": 11.819149017333984,
+      "learning_rate": 1.5815923207227558e-05,
+      "loss": 0.6426,
       "step": 12560
     },
     {
       "epoch": 2.365894974590627,
-      "grad_norm": 25.91827964782715,
-      "learning_rate": 1.0536420101637495e-05,
-      "loss": 0.5889,
+      "grad_norm": 18.320758819580078,
+      "learning_rate": 1.580463015245624e-05,
+      "loss": 0.5381,
       "step": 12570
     },
     {
       "epoch": 2.3677771503858462,
-      "grad_norm": 7.587324142456055,
-      "learning_rate": 1.0528891398456618e-05,
-      "loss": 0.6031,
+      "grad_norm": 21.337982177734375,
+      "learning_rate": 1.5793337097684926e-05,
+      "loss": 0.6222,
       "step": 12580
     },
     {
       "epoch": 2.3696593261810652,
-      "grad_norm": 16.203954696655273,
-      "learning_rate": 1.052136269527574e-05,
-      "loss": 0.7866,
+      "grad_norm": 30.114587783813477,
+      "learning_rate": 1.5782044042913606e-05,
+      "loss": 0.8209,
       "step": 12590
     },
     {
       "epoch": 2.3715415019762847,
-      "grad_norm": 16.91657066345215,
-      "learning_rate": 1.0513833992094862e-05,
-      "loss": 0.9203,
+      "grad_norm": 13.773634910583496,
+      "learning_rate": 1.5770750988142292e-05,
+      "loss": 0.7724,
       "step": 12600
     },
     {
       "epoch": 2.3734236777715036,
-      "grad_norm": 12.624746322631836,
-      "learning_rate": 1.0506305288913985e-05,
-      "loss": 0.9539,
+      "grad_norm": 7.200165748596191,
+      "learning_rate": 1.5759457933370978e-05,
+      "loss": 0.9503,
       "step": 12610
     },
     {
       "epoch": 2.375305853566723,
-      "grad_norm": 12.346266746520996,
-      "learning_rate": 1.0498776585733108e-05,
-      "loss": 0.5703,
+      "grad_norm": 16.634714126586914,
+      "learning_rate": 1.574816487859966e-05,
+      "loss": 0.5145,
       "step": 12620
     },
     {
       "epoch": 2.3771880293619425,
-      "grad_norm": 12.931204795837402,
-      "learning_rate": 1.0491247882552231e-05,
-      "loss": 0.656,
+      "grad_norm": 13.062280654907227,
+      "learning_rate": 1.5736871823828347e-05,
+      "loss": 0.584,
       "step": 12630
     },
     {
       "epoch": 2.3790702051571615,
-      "grad_norm": 24.599132537841797,
-      "learning_rate": 1.0483719179371354e-05,
-      "loss": 1.234,
+      "grad_norm": 15.320236206054688,
+      "learning_rate": 1.572557876905703e-05,
+      "loss": 1.109,
       "step": 12640
     },
     {
       "epoch": 2.380952380952381,
-      "grad_norm": 12.25839614868164,
-      "learning_rate": 1.0476190476190477e-05,
-      "loss": 0.6236,
+      "grad_norm": 18.621957778930664,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.5475,
       "step": 12650
     },
     {
       "epoch": 2.3828345567476004,
-      "grad_norm": 16.587228775024414,
-      "learning_rate": 1.0468661773009602e-05,
-      "loss": 0.6196,
+      "grad_norm": 21.138986587524414,
+      "learning_rate": 1.57029926595144e-05,
+      "loss": 0.6148,
       "step": 12660
     },
     {
       "epoch": 2.3847167325428194,
-      "grad_norm": 8.345686912536621,
-      "learning_rate": 1.0461133069828724e-05,
-      "loss": 0.7598,
+      "grad_norm": 7.1945929527282715,
+      "learning_rate": 1.5691699604743084e-05,
+      "loss": 0.6836,
       "step": 12670
     },
     {
       "epoch": 2.386598908338039,
-      "grad_norm": 12.944352149963379,
-      "learning_rate": 1.0453604366647846e-05,
-      "loss": 1.0841,
+      "grad_norm": 7.824986457824707,
+      "learning_rate": 1.5680406549971767e-05,
+      "loss": 1.0433,
       "step": 12680
     },
     {
       "epoch": 2.388481084133258,
-      "grad_norm": 26.68334197998047,
-      "learning_rate": 1.0446075663466969e-05,
-      "loss": 1.0751,
+      "grad_norm": 26.856182098388672,
+      "learning_rate": 1.566911349520045e-05,
+      "loss": 0.9668,
       "step": 12690
     },
     {
       "epoch": 2.390363259928477,
-      "grad_norm": 1.97270929813385,
-      "learning_rate": 1.0438546960286091e-05,
-      "loss": 0.6852,
+      "grad_norm": 1.3294514417648315,
+      "learning_rate": 1.5657820440429136e-05,
+      "loss": 0.6016,
       "step": 12700
     },
     {
       "epoch": 2.3922454357236966,
-      "grad_norm": 10.04570198059082,
-      "learning_rate": 1.0431018257105214e-05,
-      "loss": 0.6548,
+      "grad_norm": 25.570005416870117,
+      "learning_rate": 1.564652738565782e-05,
+      "loss": 0.6145,
       "step": 12710
     },
     {
       "epoch": 2.3941276115189156,
-      "grad_norm": 17.121076583862305,
-      "learning_rate": 1.0423489553924337e-05,
-      "loss": 0.7934,
+      "grad_norm": 22.4234676361084,
+      "learning_rate": 1.5635234330886504e-05,
+      "loss": 0.7482,
       "step": 12720
     },
     {
       "epoch": 2.396009787314135,
-      "grad_norm": 21.22518539428711,
-      "learning_rate": 1.041596085074346e-05,
-      "loss": 1.0878,
+      "grad_norm": 14.264887809753418,
+      "learning_rate": 1.562394127611519e-05,
+      "loss": 0.9606,
       "step": 12730
     },
     {
       "epoch": 2.3978919631093545,
-      "grad_norm": 14.835243225097656,
-      "learning_rate": 1.0408432147562583e-05,
-      "loss": 0.6626,
+      "grad_norm": 23.162593841552734,
+      "learning_rate": 1.5612648221343876e-05,
+      "loss": 0.6843,
       "step": 12740
     },
     {
       "epoch": 2.399774138904574,
-      "grad_norm": 24.89073371887207,
-      "learning_rate": 1.0400903444381708e-05,
-      "loss": 0.9652,
+      "grad_norm": 30.500329971313477,
+      "learning_rate": 1.560135516657256e-05,
+      "loss": 0.8691,
       "step": 12750
     },
     {
       "epoch": 2.401656314699793,
-      "grad_norm": 13.480345726013184,
-      "learning_rate": 1.0393374741200827e-05,
-      "loss": 0.758,
+      "grad_norm": 10.762563705444336,
+      "learning_rate": 1.5590062111801242e-05,
+      "loss": 0.7518,
       "step": 12760
     },
     {
       "epoch": 2.4035384904950123,
-      "grad_norm": 26.335533142089844,
-      "learning_rate": 1.038584603801995e-05,
-      "loss": 0.9258,
+      "grad_norm": 38.12343978881836,
+      "learning_rate": 1.5578769057029925e-05,
+      "loss": 0.8324,
       "step": 12770
     },
     {
       "epoch": 2.4054206662902313,
-      "grad_norm": 17.967731475830078,
-      "learning_rate": 1.0378317334839075e-05,
-      "loss": 1.1077,
+      "grad_norm": 26.032785415649414,
+      "learning_rate": 1.556747600225861e-05,
+      "loss": 0.9027,
       "step": 12780
     },
     {
       "epoch": 2.4073028420854508,
-      "grad_norm": 10.612001419067383,
-      "learning_rate": 1.0370788631658198e-05,
-      "loss": 0.7788,
+      "grad_norm": 12.26636028289795,
+      "learning_rate": 1.5556182947487297e-05,
+      "loss": 0.6673,
       "step": 12790
     },
     {
       "epoch": 2.40918501788067,
-      "grad_norm": 25.963218688964844,
-      "learning_rate": 1.036325992847732e-05,
-      "loss": 1.0074,
+      "grad_norm": 22.655187606811523,
+      "learning_rate": 1.554488989271598e-05,
+      "loss": 0.9456,
       "step": 12800
     },
     {
       "epoch": 2.411067193675889,
-      "grad_norm": 9.171457290649414,
-      "learning_rate": 1.0355731225296444e-05,
-      "loss": 0.8116,
+      "grad_norm": 8.51685619354248,
+      "learning_rate": 1.5533596837944665e-05,
+      "loss": 0.6657,
       "step": 12810
     },
     {
       "epoch": 2.4129493694711086,
-      "grad_norm": 5.5040459632873535,
-      "learning_rate": 1.0348202522115567e-05,
-      "loss": 0.7237,
+      "grad_norm": 8.790712356567383,
+      "learning_rate": 1.5522303783173348e-05,
+      "loss": 0.6555,
       "step": 12820
     },
     {
       "epoch": 2.414831545266328,
-      "grad_norm": 12.755102157592773,
-      "learning_rate": 1.034067381893469e-05,
-      "loss": 0.7111,
+      "grad_norm": 24.235885620117188,
+      "learning_rate": 1.5511010728402034e-05,
+      "loss": 0.5839,
       "step": 12830
     },
     {
       "epoch": 2.416713721061547,
-      "grad_norm": 12.672661781311035,
-      "learning_rate": 1.0333145115753812e-05,
-      "loss": 0.5718,
+      "grad_norm": 11.751230239868164,
+      "learning_rate": 1.549971767363072e-05,
+      "loss": 0.53,
       "step": 12840
     },
     {
       "epoch": 2.4185958968567665,
-      "grad_norm": 14.201343536376953,
-      "learning_rate": 1.0325616412572934e-05,
-      "loss": 0.8472,
+      "grad_norm": 11.033337593078613,
+      "learning_rate": 1.54884246188594e-05,
+      "loss": 0.7369,
       "step": 12850
     },
     {
       "epoch": 2.420478072651986,
-      "grad_norm": 11.615950584411621,
-      "learning_rate": 1.0318087709392057e-05,
-      "loss": 0.6249,
+      "grad_norm": 11.11131763458252,
+      "learning_rate": 1.5477131564088086e-05,
+      "loss": 0.5856,
       "step": 12860
     },
     {
       "epoch": 2.422360248447205,
-      "grad_norm": 33.66123962402344,
-      "learning_rate": 1.0310559006211181e-05,
-      "loss": 0.784,
+      "grad_norm": 31.728572845458984,
+      "learning_rate": 1.546583850931677e-05,
+      "loss": 0.8794,
       "step": 12870
     },
     {
       "epoch": 2.4242424242424243,
-      "grad_norm": 20.7972354888916,
-      "learning_rate": 1.0303030303030304e-05,
-      "loss": 0.8194,
+      "grad_norm": 15.746371269226074,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 0.6741,
       "step": 12880
     },
     {
       "epoch": 2.4261246000376433,
-      "grad_norm": 14.012480735778809,
-      "learning_rate": 1.0295501599849427e-05,
-      "loss": 1.0207,
+      "grad_norm": 7.750115394592285,
+      "learning_rate": 1.544325239977414e-05,
+      "loss": 0.9826,
       "step": 12890
     },
     {
       "epoch": 2.4280067758328627,
-      "grad_norm": 28.325319290161133,
-      "learning_rate": 1.028797289666855e-05,
-      "loss": 0.8505,
+      "grad_norm": 12.74660873413086,
+      "learning_rate": 1.5431959345002823e-05,
+      "loss": 0.7174,
       "step": 12900
     },
     {
       "epoch": 2.429888951628082,
-      "grad_norm": 12.772529602050781,
-      "learning_rate": 1.0280444193487673e-05,
-      "loss": 0.7771,
+      "grad_norm": 10.540325164794922,
+      "learning_rate": 1.542066629023151e-05,
+      "loss": 0.6789,
       "step": 12910
     },
     {
       "epoch": 2.431771127423301,
-      "grad_norm": 15.617521286010742,
-      "learning_rate": 1.0272915490306796e-05,
-      "loss": 0.6914,
+      "grad_norm": 16.440242767333984,
+      "learning_rate": 1.5409373235460192e-05,
+      "loss": 0.5718,
       "step": 12920
     },
     {
       "epoch": 2.4336533032185206,
-      "grad_norm": 9.929617881774902,
-      "learning_rate": 1.0265386787125919e-05,
-      "loss": 0.8166,
+      "grad_norm": 6.608902931213379,
+      "learning_rate": 1.5398080180688878e-05,
+      "loss": 0.7068,
       "step": 12930
     },
     {
       "epoch": 2.43553547901374,
-      "grad_norm": 9.618364334106445,
-      "learning_rate": 1.025785808394504e-05,
-      "loss": 0.7019,
+      "grad_norm": 9.328181266784668,
+      "learning_rate": 1.538678712591756e-05,
+      "loss": 0.5848,
       "step": 12940
     },
     {
       "epoch": 2.437417654808959,
-      "grad_norm": 22.540725708007812,
-      "learning_rate": 1.0250329380764163e-05,
-      "loss": 0.8511,
+      "grad_norm": 3.718935966491699,
+      "learning_rate": 1.5375494071146243e-05,
+      "loss": 0.7668,
       "step": 12950
     },
     {
       "epoch": 2.4392998306041784,
-      "grad_norm": 7.289876461029053,
-      "learning_rate": 1.0242800677583287e-05,
-      "loss": 0.6807,
+      "grad_norm": 12.87177848815918,
+      "learning_rate": 1.536420101637493e-05,
+      "loss": 0.5857,
       "step": 12960
     },
     {
       "epoch": 2.441182006399398,
-      "grad_norm": 18.872398376464844,
-      "learning_rate": 1.023527197440241e-05,
-      "loss": 0.9898,
+      "grad_norm": 2.911306142807007,
+      "learning_rate": 1.5352907961603616e-05,
+      "loss": 0.8994,
       "step": 12970
     },
     {
       "epoch": 2.443064182194617,
-      "grad_norm": 14.431037902832031,
-      "learning_rate": 1.0227743271221533e-05,
-      "loss": 0.961,
+      "grad_norm": 10.852980613708496,
+      "learning_rate": 1.5341614906832298e-05,
+      "loss": 0.8624,
       "step": 12980
     },
     {
       "epoch": 2.4449463579898363,
-      "grad_norm": 19.559343338012695,
-      "learning_rate": 1.0220214568040656e-05,
-      "loss": 0.7496,
+      "grad_norm": 5.891432285308838,
+      "learning_rate": 1.5330321852060984e-05,
+      "loss": 0.7049,
       "step": 12990
     },
     {
       "epoch": 2.4468285337850557,
-      "grad_norm": 5.90576171875,
-      "learning_rate": 1.0212685864859779e-05,
-      "loss": 0.7529,
+      "grad_norm": 7.124139785766602,
+      "learning_rate": 1.5319028797289667e-05,
+      "loss": 0.7183,
       "step": 13000
     },
     {
       "epoch": 2.4487107095802747,
-      "grad_norm": 18.241514205932617,
-      "learning_rate": 1.0205157161678902e-05,
-      "loss": 0.6666,
+      "grad_norm": 11.209503173828125,
+      "learning_rate": 1.5307735742518353e-05,
+      "loss": 0.5121,
       "step": 13010
     },
     {
       "epoch": 2.450592885375494,
-      "grad_norm": 30.32737159729004,
-      "learning_rate": 1.0197628458498025e-05,
-      "loss": 0.7313,
+      "grad_norm": 12.793279647827148,
+      "learning_rate": 1.529644268774704e-05,
+      "loss": 0.6083,
       "step": 13020
     },
     {
       "epoch": 2.4524750611707136,
-      "grad_norm": 8.771208763122559,
-      "learning_rate": 1.0190099755317146e-05,
-      "loss": 0.4378,
+      "grad_norm": 16.64006996154785,
+      "learning_rate": 1.528514963297572e-05,
+      "loss": 0.3401,
       "step": 13030
     },
     {
       "epoch": 2.4543572369659326,
-      "grad_norm": 21.53113555908203,
-      "learning_rate": 1.0182571052136269e-05,
-      "loss": 0.6384,
+      "grad_norm": 3.3853352069854736,
+      "learning_rate": 1.5273856578204405e-05,
+      "loss": 0.5073,
       "step": 13040
     },
     {
       "epoch": 2.456239412761152,
-      "grad_norm": 14.424870491027832,
-      "learning_rate": 1.0175042348955392e-05,
-      "loss": 0.5578,
+      "grad_norm": 11.800168991088867,
+      "learning_rate": 1.5262563523433087e-05,
+      "loss": 0.4107,
       "step": 13050
     },
     {
       "epoch": 2.458121588556371,
-      "grad_norm": 21.395099639892578,
-      "learning_rate": 1.0167513645774517e-05,
-      "loss": 0.5507,
+      "grad_norm": 17.713973999023438,
+      "learning_rate": 1.5251270468661773e-05,
+      "loss": 0.4237,
       "step": 13060
     },
     {
       "epoch": 2.4600037643515904,
-      "grad_norm": 18.84751319885254,
-      "learning_rate": 1.015998494259364e-05,
-      "loss": 0.9829,
+      "grad_norm": 14.999812126159668,
+      "learning_rate": 1.5239977413890458e-05,
+      "loss": 0.8685,
       "step": 13070
     },
     {
       "epoch": 2.46188594014681,
-      "grad_norm": 7.269599437713623,
-      "learning_rate": 1.0152456239412763e-05,
-      "loss": 0.8354,
+      "grad_norm": 10.52674674987793,
+      "learning_rate": 1.5228684359119142e-05,
+      "loss": 0.8917,
       "step": 13080
     },
     {
       "epoch": 2.463768115942029,
-      "grad_norm": 14.933770179748535,
-      "learning_rate": 1.0144927536231885e-05,
-      "loss": 0.7661,
+      "grad_norm": 21.042428970336914,
+      "learning_rate": 1.5217391304347828e-05,
+      "loss": 0.7128,
       "step": 13090
     },
     {
       "epoch": 2.4656502917372483,
-      "grad_norm": 0.9064294695854187,
-      "learning_rate": 1.0137398833051008e-05,
-      "loss": 0.9594,
+      "grad_norm": 0.1580163836479187,
+      "learning_rate": 1.5206098249576513e-05,
+      "loss": 0.9413,
       "step": 13100
     },
     {
       "epoch": 2.4675324675324677,
-      "grad_norm": 14.727800369262695,
-      "learning_rate": 1.012987012987013e-05,
-      "loss": 0.5136,
+      "grad_norm": 7.958024501800537,
+      "learning_rate": 1.5194805194805194e-05,
+      "loss": 0.4504,
       "step": 13110
     },
     {
       "epoch": 2.4694146433276867,
-      "grad_norm": 20.03411293029785,
-      "learning_rate": 1.0122341426689252e-05,
-      "loss": 1.0009,
+      "grad_norm": 21.001428604125977,
+      "learning_rate": 1.5183512140033878e-05,
+      "loss": 0.9076,
       "step": 13120
     },
     {
       "epoch": 2.471296819122906,
-      "grad_norm": 19.584400177001953,
-      "learning_rate": 1.0114812723508375e-05,
-      "loss": 0.8759,
+      "grad_norm": 14.668991088867188,
+      "learning_rate": 1.5172219085262562e-05,
+      "loss": 0.7698,
       "step": 13130
     },
     {
       "epoch": 2.4731789949181255,
-      "grad_norm": 9.938185691833496,
-      "learning_rate": 1.0107284020327498e-05,
-      "loss": 0.7959,
+      "grad_norm": 9.113917350769043,
+      "learning_rate": 1.5160926030491248e-05,
+      "loss": 0.8785,
       "step": 13140
     },
     {
       "epoch": 2.4750611707133445,
-      "grad_norm": 23.411479949951172,
-      "learning_rate": 1.0099755317146623e-05,
-      "loss": 0.7344,
+      "grad_norm": 16.353412628173828,
+      "learning_rate": 1.5149632975719933e-05,
+      "loss": 0.7005,
       "step": 13150
     },
     {
       "epoch": 2.476943346508564,
-      "grad_norm": 12.910082817077637,
-      "learning_rate": 1.0092226613965746e-05,
-      "loss": 0.6742,
+      "grad_norm": 7.100589275360107,
+      "learning_rate": 1.5138339920948617e-05,
+      "loss": 0.5536,
       "step": 13160
     },
     {
       "epoch": 2.478825522303783,
-      "grad_norm": 19.10327911376953,
-      "learning_rate": 1.0084697910784869e-05,
-      "loss": 0.8607,
+      "grad_norm": 10.338568687438965,
+      "learning_rate": 1.5127046866177302e-05,
+      "loss": 0.794,
       "step": 13170
     },
     {
       "epoch": 2.4807076980990024,
-      "grad_norm": 8.324233055114746,
-      "learning_rate": 1.0077169207603992e-05,
-      "loss": 0.874,
+      "grad_norm": 19.2764949798584,
+      "learning_rate": 1.5115753811405988e-05,
+      "loss": 0.7741,
       "step": 13180
     },
     {
       "epoch": 2.482589873894222,
-      "grad_norm": 44.7740592956543,
-      "learning_rate": 1.0069640504423115e-05,
-      "loss": 0.6132,
+      "grad_norm": 41.238563537597656,
+      "learning_rate": 1.5104460756634672e-05,
+      "loss": 0.5174,
       "step": 13190
     },
     {
       "epoch": 2.4844720496894412,
-      "grad_norm": 24.905864715576172,
-      "learning_rate": 1.0062111801242236e-05,
-      "loss": 0.7348,
+      "grad_norm": 14.756248474121094,
+      "learning_rate": 1.5093167701863353e-05,
+      "loss": 0.6969,
       "step": 13200
     },
     {
       "epoch": 2.4863542254846602,
-      "grad_norm": 39.80661392211914,
-      "learning_rate": 1.0054583098061359e-05,
-      "loss": 0.7712,
+      "grad_norm": 25.224035263061523,
+      "learning_rate": 1.5081874647092037e-05,
+      "loss": 0.7373,
       "step": 13210
     },
     {
       "epoch": 2.4882364012798797,
-      "grad_norm": 14.922135353088379,
-      "learning_rate": 1.0047054394880482e-05,
-      "loss": 0.7724,
+      "grad_norm": 27.549787521362305,
+      "learning_rate": 1.5070581592320722e-05,
+      "loss": 0.7657,
       "step": 13220
     },
     {
       "epoch": 2.4901185770750986,
-      "grad_norm": 8.277451515197754,
-      "learning_rate": 1.0039525691699605e-05,
-      "loss": 0.8076,
+      "grad_norm": 13.31575870513916,
+      "learning_rate": 1.5059288537549408e-05,
+      "loss": 0.8013,
       "step": 13230
     },
     {
       "epoch": 2.492000752870318,
-      "grad_norm": 12.632638931274414,
-      "learning_rate": 1.003199698851873e-05,
-      "loss": 0.8422,
+      "grad_norm": 15.506152153015137,
+      "learning_rate": 1.5047995482778092e-05,
+      "loss": 0.8362,
       "step": 13240
     },
     {
       "epoch": 2.4938829286655375,
-      "grad_norm": 10.459776878356934,
-      "learning_rate": 1.0024468285337852e-05,
-      "loss": 0.7523,
+      "grad_norm": 12.64037036895752,
+      "learning_rate": 1.5036702428006777e-05,
+      "loss": 0.7907,
       "step": 13250
     },
     {
       "epoch": 2.4957651044607565,
-      "grad_norm": 7.631857395172119,
-      "learning_rate": 1.0016939582156975e-05,
-      "loss": 0.7041,
+      "grad_norm": 7.721663475036621,
+      "learning_rate": 1.5025409373235461e-05,
+      "loss": 0.6027,
       "step": 13260
     },
     {
       "epoch": 2.497647280255976,
-      "grad_norm": 25.433801651000977,
-      "learning_rate": 1.0009410878976098e-05,
-      "loss": 0.8024,
+      "grad_norm": 20.326704025268555,
+      "learning_rate": 1.5014116318464145e-05,
+      "loss": 0.7932,
       "step": 13270
     },
     {
       "epoch": 2.4995294560511954,
-      "grad_norm": 2.3861024379730225,
-      "learning_rate": 1.0001882175795221e-05,
-      "loss": 0.7084,
+      "grad_norm": 3.6941041946411133,
+      "learning_rate": 1.5002823263692831e-05,
+      "loss": 0.6615,
       "step": 13280
     },
     {
       "epoch": 2.5014116318464144,
-      "grad_norm": 32.96226119995117,
-      "learning_rate": 9.994353472614344e-06,
-      "loss": 0.8429,
+      "grad_norm": 23.14988899230957,
+      "learning_rate": 1.4991530208921514e-05,
+      "loss": 0.6778,
       "step": 13290
     },
     {
       "epoch": 2.503293807641634,
-      "grad_norm": 22.255521774291992,
-      "learning_rate": 9.986824769433467e-06,
-      "loss": 0.5662,
+      "grad_norm": 8.961627960205078,
+      "learning_rate": 1.4980237154150198e-05,
+      "loss": 0.5089,
       "step": 13300
     },
     {
       "epoch": 2.505175983436853,
-      "grad_norm": 17.061006546020508,
-      "learning_rate": 9.979296066252588e-06,
-      "loss": 0.8483,
+      "grad_norm": 17.82870864868164,
+      "learning_rate": 1.4968944099378881e-05,
+      "loss": 0.8967,
       "step": 13310
     },
     {
       "epoch": 2.507058159232072,
-      "grad_norm": 16.66559410095215,
-      "learning_rate": 9.971767363071711e-06,
-      "loss": 0.6624,
+      "grad_norm": 3.7372190952301025,
+      "learning_rate": 1.4957651044607567e-05,
+      "loss": 0.6563,
       "step": 13320
     },
     {
       "epoch": 2.5089403350272916,
-      "grad_norm": 16.982269287109375,
-      "learning_rate": 9.964238659890836e-06,
-      "loss": 0.9074,
+      "grad_norm": 17.692962646484375,
+      "learning_rate": 1.4946357989836252e-05,
+      "loss": 0.9301,
       "step": 13330
     },
     {
       "epoch": 2.5108225108225106,
-      "grad_norm": 8.781649589538574,
-      "learning_rate": 9.956709956709958e-06,
-      "loss": 0.9302,
+      "grad_norm": 11.346606254577637,
+      "learning_rate": 1.4935064935064936e-05,
+      "loss": 0.689,
       "step": 13340
     },
     {
       "epoch": 2.51270468661773,
-      "grad_norm": 9.977226257324219,
-      "learning_rate": 9.94918125352908e-06,
-      "loss": 0.7217,
+      "grad_norm": 25.42557144165039,
+      "learning_rate": 1.4923771880293619e-05,
+      "loss": 0.6755,
       "step": 13350
     },
     {
       "epoch": 2.5145868624129495,
-      "grad_norm": 20.91251564025879,
-      "learning_rate": 9.941652550348203e-06,
-      "loss": 0.7181,
+      "grad_norm": 16.371252059936523,
+      "learning_rate": 1.4912478825522303e-05,
+      "loss": 0.6872,
       "step": 13360
     },
     {
       "epoch": 2.516469038208169,
-      "grad_norm": 15.54866886138916,
-      "learning_rate": 9.934123847167326e-06,
-      "loss": 0.4848,
+      "grad_norm": 10.551392555236816,
+      "learning_rate": 1.4901185770750989e-05,
+      "loss": 0.3674,
       "step": 13370
     },
     {
       "epoch": 2.518351214003388,
-      "grad_norm": 11.833428382873535,
-      "learning_rate": 9.92659514398645e-06,
-      "loss": 0.4919,
+      "grad_norm": 7.954850673675537,
+      "learning_rate": 1.4889892715979674e-05,
+      "loss": 0.445,
       "step": 13380
     },
     {
       "epoch": 2.5202333897986073,
-      "grad_norm": 7.332128524780273,
-      "learning_rate": 9.919066440805573e-06,
-      "loss": 0.7247,
+      "grad_norm": 6.890119552612305,
+      "learning_rate": 1.4878599661208358e-05,
+      "loss": 0.5636,
       "step": 13390
     },
     {
       "epoch": 2.5221155655938263,
-      "grad_norm": 13.253988265991211,
-      "learning_rate": 9.911537737624694e-06,
-      "loss": 0.7385,
+      "grad_norm": 11.65666389465332,
+      "learning_rate": 1.486730660643704e-05,
+      "loss": 0.6063,
       "step": 13400
     },
     {
       "epoch": 2.5239977413890458,
-      "grad_norm": 24.234617233276367,
-      "learning_rate": 9.904009034443817e-06,
-      "loss": 0.7513,
+      "grad_norm": 17.123275756835938,
+      "learning_rate": 1.4856013551665727e-05,
+      "loss": 0.615,
       "step": 13410
     },
     {
       "epoch": 2.525879917184265,
-      "grad_norm": 24.290355682373047,
-      "learning_rate": 9.89648033126294e-06,
-      "loss": 1.0585,
+      "grad_norm": 24.541301727294922,
+      "learning_rate": 1.4844720496894411e-05,
+      "loss": 0.9433,
       "step": 13420
     },
     {
       "epoch": 2.527762092979484,
-      "grad_norm": 7.267026901245117,
-      "learning_rate": 9.888951628082065e-06,
-      "loss": 0.574,
+      "grad_norm": 11.410900115966797,
+      "learning_rate": 1.4833427442123095e-05,
+      "loss": 0.5644,
       "step": 13430
     },
     {
       "epoch": 2.5296442687747036,
-      "grad_norm": 5.2897443771362305,
-      "learning_rate": 9.881422924901186e-06,
-      "loss": 0.6149,
+      "grad_norm": 7.17426872253418,
+      "learning_rate": 1.4822134387351778e-05,
+      "loss": 0.5348,
       "step": 13440
     },
     {
       "epoch": 2.5315264445699226,
-      "grad_norm": 27.750844955444336,
-      "learning_rate": 9.873894221720309e-06,
-      "loss": 0.9253,
+      "grad_norm": 19.93268585205078,
+      "learning_rate": 1.4810841332580463e-05,
+      "loss": 0.9548,
       "step": 13450
     },
     {
       "epoch": 2.533408620365142,
-      "grad_norm": 12.20700454711914,
-      "learning_rate": 9.866365518539432e-06,
-      "loss": 0.7781,
+      "grad_norm": 12.935050964355469,
+      "learning_rate": 1.4799548277809149e-05,
+      "loss": 0.6989,
       "step": 13460
     },
     {
       "epoch": 2.5352907961603615,
-      "grad_norm": 13.96152400970459,
-      "learning_rate": 9.858836815358556e-06,
-      "loss": 0.8337,
+      "grad_norm": 19.485132217407227,
+      "learning_rate": 1.4788255223037833e-05,
+      "loss": 0.7934,
       "step": 13470
     },
     {
       "epoch": 2.537172971955581,
-      "grad_norm": 12.576264381408691,
-      "learning_rate": 9.851308112177678e-06,
-      "loss": 0.6645,
+      "grad_norm": 28.300479888916016,
+      "learning_rate": 1.4776962168266516e-05,
+      "loss": 0.6659,
       "step": 13480
     },
     {
       "epoch": 2.5390551477508,
-      "grad_norm": 25.116222381591797,
-      "learning_rate": 9.8437794089968e-06,
-      "loss": 1.2199,
+      "grad_norm": 27.766523361206055,
+      "learning_rate": 1.47656691134952e-05,
+      "loss": 1.0873,
       "step": 13490
     },
     {
       "epoch": 2.5409373235460193,
-      "grad_norm": 18.44822120666504,
-      "learning_rate": 9.836250705815924e-06,
-      "loss": 0.5354,
+      "grad_norm": 23.600576400756836,
+      "learning_rate": 1.4754376058723884e-05,
+      "loss": 0.5181,
       "step": 13500
     },
     {
       "epoch": 2.5428194993412383,
-      "grad_norm": 12.855209350585938,
-      "learning_rate": 9.828722002635046e-06,
-      "loss": 0.7728,
+      "grad_norm": 12.327131271362305,
+      "learning_rate": 1.474308300395257e-05,
+      "loss": 0.7655,
       "step": 13510
     },
     {
       "epoch": 2.5447016751364577,
-      "grad_norm": 25.006006240844727,
-      "learning_rate": 9.821193299454171e-06,
-      "loss": 0.8413,
+      "grad_norm": 19.353673934936523,
+      "learning_rate": 1.4731789949181255e-05,
+      "loss": 0.8238,
       "step": 13520
     },
     {
       "epoch": 2.546583850931677,
-      "grad_norm": 27.276927947998047,
-      "learning_rate": 9.813664596273292e-06,
-      "loss": 1.0349,
+      "grad_norm": 23.07147979736328,
+      "learning_rate": 1.4720496894409938e-05,
+      "loss": 0.8695,
       "step": 13530
     },
     {
       "epoch": 2.548466026726896,
-      "grad_norm": 13.0643949508667,
-      "learning_rate": 9.806135893092415e-06,
-      "loss": 0.7487,
+      "grad_norm": 19.188295364379883,
+      "learning_rate": 1.4709203839638622e-05,
+      "loss": 0.7139,
       "step": 13540
     },
     {
       "epoch": 2.5503482025221156,
-      "grad_norm": 15.080717086791992,
-      "learning_rate": 9.798607189911538e-06,
-      "loss": 0.9849,
+      "grad_norm": 8.434711456298828,
+      "learning_rate": 1.4697910784867308e-05,
+      "loss": 0.8626,
       "step": 13550
     },
     {
       "epoch": 2.5522303783173346,
-      "grad_norm": 7.237006664276123,
-      "learning_rate": 9.791078486730661e-06,
-      "loss": 0.8056,
+      "grad_norm": 4.494293212890625,
+      "learning_rate": 1.4686617730095992e-05,
+      "loss": 0.7623,
       "step": 13560
     },
     {
       "epoch": 2.554112554112554,
-      "grad_norm": 12.461982727050781,
-      "learning_rate": 9.783549783549784e-06,
-      "loss": 0.8994,
+      "grad_norm": 17.76078224182129,
+      "learning_rate": 1.4675324675324675e-05,
+      "loss": 0.7771,
       "step": 13570
     },
     {
       "epoch": 2.5559947299077734,
-      "grad_norm": 13.99405288696289,
-      "learning_rate": 9.776021080368907e-06,
-      "loss": 0.7762,
+      "grad_norm": 18.283334732055664,
+      "learning_rate": 1.466403162055336e-05,
+      "loss": 0.615,
       "step": 13580
     },
     {
       "epoch": 2.557876905702993,
-      "grad_norm": 12.452048301696777,
-      "learning_rate": 9.76849237718803e-06,
-      "loss": 1.1317,
+      "grad_norm": 15.662129402160645,
+      "learning_rate": 1.4652738565782044e-05,
+      "loss": 1.0202,
       "step": 13590
     },
     {
       "epoch": 2.559759081498212,
-      "grad_norm": 18.15245819091797,
-      "learning_rate": 9.760963674007153e-06,
-      "loss": 0.6629,
+      "grad_norm": 11.736997604370117,
+      "learning_rate": 1.464144551101073e-05,
+      "loss": 0.6311,
       "step": 13600
     },
     {
       "epoch": 2.5616412572934313,
-      "grad_norm": 20.07953453063965,
-      "learning_rate": 9.753434970826276e-06,
-      "loss": 0.6838,
+      "grad_norm": 26.242910385131836,
+      "learning_rate": 1.4630152456239413e-05,
+      "loss": 0.5408,
       "step": 13610
     },
     {
       "epoch": 2.5635234330886503,
-      "grad_norm": 34.626930236816406,
-      "learning_rate": 9.745906267645399e-06,
-      "loss": 0.634,
+      "grad_norm": 34.35280227661133,
+      "learning_rate": 1.4618859401468097e-05,
+      "loss": 0.5723,
       "step": 13620
     },
     {
       "epoch": 2.5654056088838697,
-      "grad_norm": 11.463188171386719,
-      "learning_rate": 9.738377564464521e-06,
-      "loss": 1.0214,
+      "grad_norm": 12.662686347961426,
+      "learning_rate": 1.4607566346696781e-05,
+      "loss": 0.9724,
       "step": 13630
     },
     {
       "epoch": 2.567287784679089,
-      "grad_norm": 9.072608947753906,
-      "learning_rate": 9.730848861283644e-06,
-      "loss": 0.7319,
+      "grad_norm": 7.174187660217285,
+      "learning_rate": 1.4596273291925466e-05,
+      "loss": 0.7687,
       "step": 13640
     },
     {
       "epoch": 2.5691699604743086,
-      "grad_norm": 22.94637107849121,
-      "learning_rate": 9.723320158102767e-06,
-      "loss": 0.68,
+      "grad_norm": 13.642120361328125,
+      "learning_rate": 1.4584980237154152e-05,
+      "loss": 0.5166,
       "step": 13650
     },
     {
       "epoch": 2.5710521362695276,
-      "grad_norm": 5.962002754211426,
-      "learning_rate": 9.71579145492189e-06,
-      "loss": 0.8871,
+      "grad_norm": 18.49991798400879,
+      "learning_rate": 1.4573687182382835e-05,
+      "loss": 0.889,
       "step": 13660
     },
     {
       "epoch": 2.572934312064747,
-      "grad_norm": 16.218671798706055,
-      "learning_rate": 9.708262751741013e-06,
-      "loss": 0.88,
+      "grad_norm": 15.515763282775879,
+      "learning_rate": 1.4562394127611519e-05,
+      "loss": 0.7712,
       "step": 13670
     },
     {
       "epoch": 2.574816487859966,
-      "grad_norm": 7.1879777908325195,
-      "learning_rate": 9.700734048560136e-06,
-      "loss": 1.2115,
+      "grad_norm": 9.801226615905762,
+      "learning_rate": 1.4551101072840203e-05,
+      "loss": 1.0405,
       "step": 13680
     },
     {
       "epoch": 2.5766986636551854,
-      "grad_norm": 7.769774913787842,
-      "learning_rate": 9.693205345379259e-06,
-      "loss": 0.4464,
+      "grad_norm": 8.876042366027832,
+      "learning_rate": 1.453980801806889e-05,
+      "loss": 0.3575,
       "step": 13690
     },
     {
       "epoch": 2.578580839450405,
-      "grad_norm": 31.54520606994629,
-      "learning_rate": 9.685676642198382e-06,
-      "loss": 1.1748,
+      "grad_norm": 23.405811309814453,
+      "learning_rate": 1.4528514963297572e-05,
+      "loss": 1.0442,
       "step": 13700
     },
     {
       "epoch": 2.580463015245624,
-      "grad_norm": 36.15426254272461,
-      "learning_rate": 9.678147939017505e-06,
-      "loss": 0.8529,
+      "grad_norm": 19.530874252319336,
+      "learning_rate": 1.4517221908526256e-05,
+      "loss": 0.7402,
       "step": 13710
     },
     {
       "epoch": 2.5823451910408433,
-      "grad_norm": 19.819948196411133,
-      "learning_rate": 9.670619235836628e-06,
-      "loss": 0.658,
+      "grad_norm": 14.73330020904541,
+      "learning_rate": 1.450592885375494e-05,
+      "loss": 0.6519,
       "step": 13720
     },
     {
       "epoch": 2.5842273668360622,
-      "grad_norm": 38.99443435668945,
-      "learning_rate": 9.66309053265575e-06,
-      "loss": 1.0109,
+      "grad_norm": 31.803998947143555,
+      "learning_rate": 1.4494635798983625e-05,
+      "loss": 1.0237,
       "step": 13730
     },
     {
       "epoch": 2.5861095426312817,
-      "grad_norm": 38.460819244384766,
-      "learning_rate": 9.655561829474874e-06,
-      "loss": 0.9315,
+      "grad_norm": 25.37823486328125,
+      "learning_rate": 1.448334274421231e-05,
+      "loss": 0.9164,
       "step": 13740
     },
     {
       "epoch": 2.587991718426501,
-      "grad_norm": 30.626075744628906,
-      "learning_rate": 9.648033126293997e-06,
-      "loss": 0.9099,
+      "grad_norm": 25.509199142456055,
+      "learning_rate": 1.4472049689440994e-05,
+      "loss": 0.76,
       "step": 13750
     },
     {
       "epoch": 2.5898738942217205,
-      "grad_norm": 2.2698826789855957,
-      "learning_rate": 9.64050442311312e-06,
-      "loss": 0.357,
+      "grad_norm": 4.799928665161133,
+      "learning_rate": 1.4460756634669678e-05,
+      "loss": 0.2943,
       "step": 13760
     },
     {
       "epoch": 2.5917560700169395,
-      "grad_norm": 13.714137077331543,
-      "learning_rate": 9.632975719932242e-06,
-      "loss": 0.8274,
+      "grad_norm": 12.444110870361328,
+      "learning_rate": 1.4449463579898363e-05,
+      "loss": 0.8041,
       "step": 13770
     },
     {
       "epoch": 2.593638245812159,
-      "grad_norm": 13.168296813964844,
-      "learning_rate": 9.625447016751365e-06,
-      "loss": 0.6799,
+      "grad_norm": 16.842971801757812,
+      "learning_rate": 1.4438170525127047e-05,
+      "loss": 0.6848,
       "step": 13780
     },
     {
       "epoch": 2.595520421607378,
-      "grad_norm": 22.481964111328125,
-      "learning_rate": 9.617918313570488e-06,
-      "loss": 0.8352,
+      "grad_norm": 21.784448623657227,
+      "learning_rate": 1.4426877470355732e-05,
+      "loss": 0.7807,
       "step": 13790
     },
     {
       "epoch": 2.5974025974025974,
-      "grad_norm": 2.651874303817749,
-      "learning_rate": 9.610389610389611e-06,
-      "loss": 0.5312,
+      "grad_norm": 1.9101155996322632,
+      "learning_rate": 1.4415584415584416e-05,
+      "loss": 0.4079,
       "step": 13800
     },
     {
       "epoch": 2.599284773197817,
-      "grad_norm": 24.131620407104492,
-      "learning_rate": 9.602860907208734e-06,
-      "loss": 1.0113,
+      "grad_norm": 10.34659194946289,
+      "learning_rate": 1.44042913608131e-05,
+      "loss": 0.9427,
       "step": 13810
     },
     {
       "epoch": 2.601166948993036,
-      "grad_norm": 28.216411590576172,
-      "learning_rate": 9.595332204027857e-06,
-      "loss": 0.9916,
+      "grad_norm": 32.64616394042969,
+      "learning_rate": 1.4392998306041785e-05,
+      "loss": 0.968,
       "step": 13820
     },
     {
       "epoch": 2.6030491247882552,
-      "grad_norm": 33.7442626953125,
-      "learning_rate": 9.58780350084698e-06,
-      "loss": 1.0541,
+      "grad_norm": 23.128334045410156,
+      "learning_rate": 1.4381705251270469e-05,
+      "loss": 0.8594,
       "step": 13830
     },
     {
       "epoch": 2.6049313005834747,
-      "grad_norm": 26.970806121826172,
-      "learning_rate": 9.580274797666103e-06,
-      "loss": 0.5681,
+      "grad_norm": 13.279285430908203,
+      "learning_rate": 1.4370412196499153e-05,
+      "loss": 0.7133,
       "step": 13840
     },
     {
       "epoch": 2.6068134763786937,
-      "grad_norm": 4.043606758117676,
-      "learning_rate": 9.572746094485226e-06,
-      "loss": 0.5892,
+      "grad_norm": 3.632760763168335,
+      "learning_rate": 1.4359119141727838e-05,
+      "loss": 0.4146,
       "step": 13850
     },
     {
       "epoch": 2.608695652173913,
-      "grad_norm": 1.2837311029434204,
-      "learning_rate": 9.565217391304349e-06,
-      "loss": 0.4397,
+      "grad_norm": 5.402322769165039,
+      "learning_rate": 1.4347826086956522e-05,
+      "loss": 0.4458,
       "step": 13860
     },
     {
       "epoch": 2.6105778279691325,
-      "grad_norm": 12.38525104522705,
-      "learning_rate": 9.557688688123472e-06,
-      "loss": 0.6613,
+      "grad_norm": 18.91129493713379,
+      "learning_rate": 1.4336533032185207e-05,
+      "loss": 0.7429,
       "step": 13870
     },
     {
       "epoch": 2.6124600037643515,
-      "grad_norm": 23.795654296875,
-      "learning_rate": 9.550159984942595e-06,
-      "loss": 0.6335,
+      "grad_norm": 29.642852783203125,
+      "learning_rate": 1.4325239977413891e-05,
+      "loss": 0.6789,
       "step": 13880
     },
     {
       "epoch": 2.614342179559571,
-      "grad_norm": 28.89879035949707,
-      "learning_rate": 9.542631281761717e-06,
-      "loss": 0.7167,
+      "grad_norm": 26.60822868347168,
+      "learning_rate": 1.4313946922642575e-05,
+      "loss": 0.708,
       "step": 13890
     },
     {
       "epoch": 2.61622435535479,
-      "grad_norm": 8.0195894241333,
-      "learning_rate": 9.53510257858084e-06,
-      "loss": 1.1297,
+      "grad_norm": 8.151140213012695,
+      "learning_rate": 1.430265386787126e-05,
+      "loss": 0.9136,
       "step": 13900
     },
     {
       "epoch": 2.6181065311500094,
-      "grad_norm": 10.090495109558105,
-      "learning_rate": 9.527573875399963e-06,
-      "loss": 0.858,
+      "grad_norm": 9.463515281677246,
+      "learning_rate": 1.4291360813099944e-05,
+      "loss": 0.7923,
       "step": 13910
     },
     {
       "epoch": 2.619988706945229,
-      "grad_norm": 24.793590545654297,
-      "learning_rate": 9.520045172219086e-06,
-      "loss": 0.8893,
+      "grad_norm": 21.169139862060547,
+      "learning_rate": 1.4280067758328628e-05,
+      "loss": 0.7719,
       "step": 13920
     },
     {
       "epoch": 2.621870882740448,
-      "grad_norm": 3.7464537620544434,
-      "learning_rate": 9.512516469038209e-06,
-      "loss": 0.5312,
+      "grad_norm": 2.78265643119812,
+      "learning_rate": 1.4268774703557313e-05,
+      "loss": 0.564,
       "step": 13930
     },
     {
       "epoch": 2.623753058535667,
-      "grad_norm": 11.590424537658691,
-      "learning_rate": 9.504987765857332e-06,
-      "loss": 0.7581,
+      "grad_norm": 10.70055103302002,
+      "learning_rate": 1.4257481648785997e-05,
+      "loss": 0.7906,
       "step": 13940
     },
     {
       "epoch": 2.6256352343308866,
-      "grad_norm": 23.143346786499023,
-      "learning_rate": 9.497459062676455e-06,
-      "loss": 0.581,
+      "grad_norm": 6.613348007202148,
+      "learning_rate": 1.4246188594014682e-05,
+      "loss": 0.4984,
       "step": 13950
     },
     {
       "epoch": 2.6275174101261056,
-      "grad_norm": 5.5682477951049805,
-      "learning_rate": 9.489930359495578e-06,
-      "loss": 0.9783,
+      "grad_norm": 13.339488983154297,
+      "learning_rate": 1.4234895539243364e-05,
+      "loss": 0.9332,
       "step": 13960
     },
     {
       "epoch": 2.629399585921325,
-      "grad_norm": 14.035380363464355,
-      "learning_rate": 9.4824016563147e-06,
-      "loss": 0.5775,
+      "grad_norm": 11.467644691467285,
+      "learning_rate": 1.422360248447205e-05,
+      "loss": 0.5156,
       "step": 13970
     },
     {
       "epoch": 2.6312817617165445,
-      "grad_norm": 28.95306968688965,
-      "learning_rate": 9.474872953133824e-06,
-      "loss": 0.5147,
+      "grad_norm": 29.80975914001465,
+      "learning_rate": 1.4212309429700735e-05,
+      "loss": 0.5243,
       "step": 13980
     },
     {
       "epoch": 2.6331639375117635,
-      "grad_norm": 14.128450393676758,
-      "learning_rate": 9.467344249952947e-06,
-      "loss": 1.1285,
+      "grad_norm": 18.506032943725586,
+      "learning_rate": 1.4201016374929419e-05,
+      "loss": 0.9703,
       "step": 13990
     },
     {
       "epoch": 2.635046113306983,
-      "grad_norm": 23.115079879760742,
-      "learning_rate": 9.45981554677207e-06,
-      "loss": 0.9068,
+      "grad_norm": 12.98060131072998,
+      "learning_rate": 1.4189723320158104e-05,
+      "loss": 0.9162,
       "step": 14000
     },
     {
       "epoch": 2.636928289102202,
-      "grad_norm": 4.194851875305176,
-      "learning_rate": 9.452286843591193e-06,
-      "loss": 0.7157,
+      "grad_norm": 3.217790365219116,
+      "learning_rate": 1.4178430265386786e-05,
+      "loss": 0.6168,
       "step": 14010
     },
     {
       "epoch": 2.6388104648974213,
-      "grad_norm": 6.765182018280029,
-      "learning_rate": 9.444758140410315e-06,
-      "loss": 0.6346,
+      "grad_norm": 12.240312576293945,
+      "learning_rate": 1.4167137210615472e-05,
+      "loss": 0.587,
       "step": 14020
     },
     {
       "epoch": 2.6406926406926408,
-      "grad_norm": 5.432229518890381,
-      "learning_rate": 9.437229437229438e-06,
-      "loss": 0.5878,
+      "grad_norm": 17.046951293945312,
+      "learning_rate": 1.4155844155844157e-05,
+      "loss": 0.4854,
       "step": 14030
     },
     {
       "epoch": 2.64257481648786,
-      "grad_norm": 27.054845809936523,
-      "learning_rate": 9.429700734048561e-06,
-      "loss": 0.517,
+      "grad_norm": 20.683822631835938,
+      "learning_rate": 1.4144551101072841e-05,
+      "loss": 0.5354,
       "step": 14040
     },
     {
       "epoch": 2.644456992283079,
-      "grad_norm": 9.652582168579102,
-      "learning_rate": 9.422172030867684e-06,
-      "loss": 0.6961,
+      "grad_norm": 19.47835350036621,
+      "learning_rate": 1.4133258046301524e-05,
+      "loss": 0.5272,
       "step": 14050
     },
     {
       "epoch": 2.6463391680782986,
-      "grad_norm": 2.1461129188537598,
-      "learning_rate": 9.414643327686807e-06,
-      "loss": 0.7967,
+      "grad_norm": 5.546383380889893,
+      "learning_rate": 1.412196499153021e-05,
+      "loss": 0.7365,
       "step": 14060
     },
     {
       "epoch": 2.6482213438735176,
-      "grad_norm": 61.811004638671875,
-      "learning_rate": 9.40711462450593e-06,
-      "loss": 0.4514,
+      "grad_norm": 19.25607681274414,
+      "learning_rate": 1.4110671936758894e-05,
+      "loss": 0.4252,
       "step": 14070
     },
     {
       "epoch": 2.650103519668737,
-      "grad_norm": 5.977787017822266,
-      "learning_rate": 9.399585921325053e-06,
-      "loss": 0.6916,
+      "grad_norm": 5.1752448081970215,
+      "learning_rate": 1.4099378881987579e-05,
+      "loss": 0.7042,
       "step": 14080
     },
     {
       "epoch": 2.6519856954639565,
-      "grad_norm": 1.1817024946212769,
-      "learning_rate": 9.392057218144174e-06,
-      "loss": 0.4419,
+      "grad_norm": 1.2071647644042969,
+      "learning_rate": 1.4088085827216261e-05,
+      "loss": 0.5272,
       "step": 14090
     },
     {
       "epoch": 2.6538678712591754,
-      "grad_norm": 1.6775157451629639,
-      "learning_rate": 9.384528514963299e-06,
-      "loss": 0.7471,
+      "grad_norm": 33.30137252807617,
+      "learning_rate": 1.4076792772444946e-05,
+      "loss": 0.7419,
       "step": 14100
     },
     {
       "epoch": 2.655750047054395,
-      "grad_norm": 10.702045440673828,
-      "learning_rate": 9.376999811782422e-06,
-      "loss": 0.727,
+      "grad_norm": 2.014209508895874,
+      "learning_rate": 1.4065499717673632e-05,
+      "loss": 0.7304,
       "step": 14110
     },
     {
       "epoch": 2.6576322228496143,
-      "grad_norm": 26.477373123168945,
-      "learning_rate": 9.369471108601545e-06,
-      "loss": 1.0107,
+      "grad_norm": 24.12761116027832,
+      "learning_rate": 1.4054206662902316e-05,
+      "loss": 0.859,
       "step": 14120
     },
     {
       "epoch": 2.6595143986448333,
-      "grad_norm": 7.560026168823242,
-      "learning_rate": 9.361942405420668e-06,
-      "loss": 0.7974,
+      "grad_norm": 13.404014587402344,
+      "learning_rate": 1.4042913608131e-05,
+      "loss": 0.7206,
       "step": 14130
     },
     {
       "epoch": 2.6613965744400527,
-      "grad_norm": 29.21369743347168,
-      "learning_rate": 9.354413702239789e-06,
-      "loss": 0.8084,
+      "grad_norm": 28.344799041748047,
+      "learning_rate": 1.4031620553359683e-05,
+      "loss": 0.7208,
       "step": 14140
     },
     {
       "epoch": 2.663278750235272,
-      "grad_norm": 19.456628799438477,
-      "learning_rate": 9.346884999058913e-06,
-      "loss": 0.9577,
+      "grad_norm": 23.86800765991211,
+      "learning_rate": 1.4020327498588368e-05,
+      "loss": 0.8126,
       "step": 14150
     },
     {
       "epoch": 2.665160926030491,
-      "grad_norm": 18.852127075195312,
-      "learning_rate": 9.339356295878036e-06,
-      "loss": 0.6632,
+      "grad_norm": 14.190437316894531,
+      "learning_rate": 1.4009034443817054e-05,
+      "loss": 0.6967,
       "step": 14160
     },
     {
       "epoch": 2.6670431018257106,
-      "grad_norm": 2.706190824508667,
-      "learning_rate": 9.33182759269716e-06,
-      "loss": 0.5317,
+      "grad_norm": 0.3645530045032501,
+      "learning_rate": 1.3997741389045738e-05,
+      "loss": 0.4742,
       "step": 14170
     },
     {
       "epoch": 2.6689252776209296,
-      "grad_norm": 11.295425415039062,
-      "learning_rate": 9.32429888951628e-06,
-      "loss": 1.0268,
+      "grad_norm": 15.038154602050781,
+      "learning_rate": 1.398644833427442e-05,
+      "loss": 1.0225,
       "step": 14180
     },
     {
       "epoch": 2.670807453416149,
-      "grad_norm": 16.14836311340332,
-      "learning_rate": 9.316770186335405e-06,
-      "loss": 0.6663,
+      "grad_norm": 12.963705062866211,
+      "learning_rate": 1.3975155279503105e-05,
+      "loss": 0.6174,
       "step": 14190
     },
     {
       "epoch": 2.6726896292113684,
-      "grad_norm": 11.459263801574707,
-      "learning_rate": 9.309241483154528e-06,
-      "loss": 0.7522,
+      "grad_norm": 15.526269912719727,
+      "learning_rate": 1.3963862224731791e-05,
+      "loss": 0.6071,
       "step": 14200
     },
     {
       "epoch": 2.674571805006588,
-      "grad_norm": 15.684527397155762,
-      "learning_rate": 9.301712779973651e-06,
-      "loss": 1.1521,
+      "grad_norm": 12.310412406921387,
+      "learning_rate": 1.3952569169960476e-05,
+      "loss": 1.0341,
       "step": 14210
     },
     {
       "epoch": 2.676453980801807,
-      "grad_norm": 5.818396091461182,
-      "learning_rate": 9.294184076792772e-06,
-      "loss": 0.8579,
+      "grad_norm": 17.386167526245117,
+      "learning_rate": 1.3941276115189158e-05,
+      "loss": 0.9449,
       "step": 14220
     },
     {
       "epoch": 2.6783361565970263,
-      "grad_norm": 4.9605326652526855,
-      "learning_rate": 9.286655373611895e-06,
-      "loss": 0.4771,
+      "grad_norm": 3.0282480716705322,
+      "learning_rate": 1.3929983060417843e-05,
+      "loss": 0.4089,
       "step": 14230
     },
     {
       "epoch": 2.6802183323922453,
-      "grad_norm": 3.3847286701202393,
-      "learning_rate": 9.27912667043102e-06,
-      "loss": 0.5427,
+      "grad_norm": 1.6062228679656982,
+      "learning_rate": 1.3918690005646527e-05,
+      "loss": 0.4825,
       "step": 14240
     },
     {
       "epoch": 2.6821005081874647,
-      "grad_norm": 8.891020774841309,
-      "learning_rate": 9.271597967250143e-06,
-      "loss": 0.6936,
+      "grad_norm": 1.4682273864746094,
+      "learning_rate": 1.3907396950875213e-05,
+      "loss": 0.597,
       "step": 14250
     },
     {
       "epoch": 2.683982683982684,
-      "grad_norm": 17.484031677246094,
-      "learning_rate": 9.264069264069266e-06,
-      "loss": 0.9668,
+      "grad_norm": 19.706525802612305,
+      "learning_rate": 1.3896103896103897e-05,
+      "loss": 0.9402,
       "step": 14260
     },
     {
       "epoch": 2.685864859777903,
-      "grad_norm": 11.090484619140625,
-      "learning_rate": 9.256540560888387e-06,
-      "loss": 1.2104,
+      "grad_norm": 13.49492073059082,
+      "learning_rate": 1.388481084133258e-05,
+      "loss": 1.1029,
       "step": 14270
     },
     {
       "epoch": 2.6877470355731226,
-      "grad_norm": 17.802000045776367,
-      "learning_rate": 9.24901185770751e-06,
-      "loss": 0.8714,
+      "grad_norm": 27.556337356567383,
+      "learning_rate": 1.3873517786561265e-05,
+      "loss": 0.789,
       "step": 14280
     },
     {
       "epoch": 2.6896292113683415,
-      "grad_norm": 12.713444709777832,
-      "learning_rate": 9.241483154526634e-06,
-      "loss": 0.6936,
+      "grad_norm": 16.97081756591797,
+      "learning_rate": 1.3862224731789949e-05,
+      "loss": 0.5639,
       "step": 14290
     },
     {
       "epoch": 2.691511387163561,
-      "grad_norm": 15.754109382629395,
-      "learning_rate": 9.233954451345757e-06,
-      "loss": 0.8509,
+      "grad_norm": 24.679975509643555,
+      "learning_rate": 1.3850931677018635e-05,
+      "loss": 0.7003,
       "step": 14300
     },
     {
       "epoch": 2.6933935629587804,
-      "grad_norm": 25.97137451171875,
-      "learning_rate": 9.226425748164878e-06,
-      "loss": 0.6812,
+      "grad_norm": 24.730314254760742,
+      "learning_rate": 1.3839638622247318e-05,
+      "loss": 0.6968,
       "step": 14310
     },
     {
       "epoch": 2.695275738754,
-      "grad_norm": 5.161522388458252,
-      "learning_rate": 9.218897044984001e-06,
-      "loss": 0.7603,
+      "grad_norm": 13.802581787109375,
+      "learning_rate": 1.3828345567476002e-05,
+      "loss": 0.7957,
       "step": 14320
     },
     {
       "epoch": 2.697157914549219,
-      "grad_norm": 9.959467887878418,
-      "learning_rate": 9.211368341803126e-06,
-      "loss": 0.4748,
+      "grad_norm": 21.049495697021484,
+      "learning_rate": 1.3817052512704686e-05,
+      "loss": 0.3892,
       "step": 14330
     },
     {
       "epoch": 2.6990400903444383,
-      "grad_norm": 5.075318336486816,
-      "learning_rate": 9.203839638622249e-06,
-      "loss": 0.7132,
+      "grad_norm": 4.196599006652832,
+      "learning_rate": 1.3805759457933373e-05,
+      "loss": 0.6734,
       "step": 14340
     },
     {
       "epoch": 2.7009222661396572,
-      "grad_norm": 47.152984619140625,
-      "learning_rate": 9.19631093544137e-06,
-      "loss": 0.9679,
+      "grad_norm": 37.20542526245117,
+      "learning_rate": 1.3794466403162055e-05,
+      "loss": 0.9126,
       "step": 14350
     },
     {
       "epoch": 2.7028044419348767,
-      "grad_norm": 5.405478000640869,
-      "learning_rate": 9.188782232260493e-06,
-      "loss": 0.6459,
+      "grad_norm": 11.342184066772461,
+      "learning_rate": 1.378317334839074e-05,
+      "loss": 0.5825,
       "step": 14360
     },
     {
       "epoch": 2.704686617730096,
-      "grad_norm": 8.182209014892578,
-      "learning_rate": 9.181253529079616e-06,
-      "loss": 0.7378,
+      "grad_norm": 9.5272798538208,
+      "learning_rate": 1.3771880293619424e-05,
+      "loss": 0.6676,
       "step": 14370
     },
     {
       "epoch": 2.7065687935253155,
-      "grad_norm": 16.320960998535156,
-      "learning_rate": 9.17372482589874e-06,
-      "loss": 0.9392,
+      "grad_norm": 14.55465030670166,
+      "learning_rate": 1.3760587238848108e-05,
+      "loss": 0.899,
       "step": 14380
     },
     {
       "epoch": 2.7084509693205345,
-      "grad_norm": 26.094823837280273,
-      "learning_rate": 9.166196122717864e-06,
-      "loss": 0.5237,
+      "grad_norm": 16.193349838256836,
+      "learning_rate": 1.3749294184076794e-05,
+      "loss": 0.5359,
       "step": 14390
     },
     {
       "epoch": 2.710333145115754,
-      "grad_norm": 1.0742714405059814,
-      "learning_rate": 9.158667419536985e-06,
-      "loss": 0.7379,
+      "grad_norm": 0.4011596441268921,
+      "learning_rate": 1.3738001129305477e-05,
+      "loss": 0.6422,
       "step": 14400
     },
     {
       "epoch": 2.712215320910973,
-      "grad_norm": 19.00723648071289,
-      "learning_rate": 9.151138716356108e-06,
-      "loss": 0.7913,
+      "grad_norm": 18.894222259521484,
+      "learning_rate": 1.3726708074534161e-05,
+      "loss": 0.6957,
       "step": 14410
     },
     {
       "epoch": 2.7140974967061924,
-      "grad_norm": 14.363834381103516,
-      "learning_rate": 9.143610013175232e-06,
-      "loss": 0.9245,
+      "grad_norm": 23.98822593688965,
+      "learning_rate": 1.3715415019762846e-05,
+      "loss": 0.9018,
       "step": 14420
     },
     {
       "epoch": 2.715979672501412,
-      "grad_norm": 20.376310348510742,
-      "learning_rate": 9.136081309994355e-06,
-      "loss": 0.749,
+      "grad_norm": 24.73630142211914,
+      "learning_rate": 1.3704121964991532e-05,
+      "loss": 0.6684,
       "step": 14430
     },
     {
       "epoch": 2.717861848296631,
-      "grad_norm": 0.6434114575386047,
-      "learning_rate": 9.128552606813476e-06,
-      "loss": 0.7697,
+      "grad_norm": 0.7143511772155762,
+      "learning_rate": 1.3692828910220215e-05,
+      "loss": 0.7074,
       "step": 14440
     },
     {
       "epoch": 2.7197440240918502,
-      "grad_norm": 12.685359001159668,
-      "learning_rate": 9.1210239036326e-06,
-      "loss": 0.6835,
+      "grad_norm": 13.2362699508667,
+      "learning_rate": 1.3681535855448899e-05,
+      "loss": 0.6953,
       "step": 14450
     },
     {
       "epoch": 2.721626199887069,
-      "grad_norm": 2.339576005935669,
-      "learning_rate": 9.113495200451722e-06,
-      "loss": 0.7144,
+      "grad_norm": 0.7118558287620544,
+      "learning_rate": 1.3670242800677583e-05,
+      "loss": 0.6667,
       "step": 14460
     },
     {
       "epoch": 2.7235083756822887,
-      "grad_norm": 10.208086013793945,
-      "learning_rate": 9.105966497270847e-06,
-      "loss": 0.7922,
+      "grad_norm": 10.795101165771484,
+      "learning_rate": 1.3658949745906268e-05,
+      "loss": 0.8088,
       "step": 14470
     },
     {
       "epoch": 2.725390551477508,
-      "grad_norm": 12.116181373596191,
-      "learning_rate": 9.09843779408997e-06,
-      "loss": 0.5502,
+      "grad_norm": 21.050437927246094,
+      "learning_rate": 1.3647656691134954e-05,
+      "loss": 0.6499,
       "step": 14480
     },
     {
       "epoch": 2.7272727272727275,
-      "grad_norm": 18.400135040283203,
-      "learning_rate": 9.090909090909091e-06,
-      "loss": 0.9157,
+      "grad_norm": 15.647856712341309,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.769,
       "step": 14490
     },
     {
       "epoch": 2.7291549030679465,
-      "grad_norm": 20.76301383972168,
-      "learning_rate": 9.083380387728214e-06,
-      "loss": 0.8151,
+      "grad_norm": 18.59046173095703,
+      "learning_rate": 1.3625070581592321e-05,
+      "loss": 0.8155,
       "step": 14500
     },
     {
       "epoch": 2.731037078863166,
-      "grad_norm": 4.120022296905518,
-      "learning_rate": 9.075851684547337e-06,
-      "loss": 0.6814,
+      "grad_norm": 0.6544022560119629,
+      "learning_rate": 1.3613777526821005e-05,
+      "loss": 0.5492,
       "step": 14510
     },
     {
       "epoch": 2.732919254658385,
-      "grad_norm": 10.123649597167969,
-      "learning_rate": 9.068322981366461e-06,
-      "loss": 0.7769,
+      "grad_norm": 5.2929182052612305,
+      "learning_rate": 1.360248447204969e-05,
+      "loss": 0.6304,
       "step": 14520
     },
     {
       "epoch": 2.7348014304536044,
-      "grad_norm": 16.534719467163086,
-      "learning_rate": 9.060794278185583e-06,
-      "loss": 0.8379,
+      "grad_norm": 18.719219207763672,
+      "learning_rate": 1.3591191417278374e-05,
+      "loss": 0.7405,
       "step": 14530
     },
     {
       "epoch": 2.736683606248824,
-      "grad_norm": 18.45408058166504,
-      "learning_rate": 9.053265575004706e-06,
-      "loss": 0.5487,
+      "grad_norm": 14.632789611816406,
+      "learning_rate": 1.3579898362507058e-05,
+      "loss": 0.6674,
       "step": 14540
     },
     {
       "epoch": 2.7385657820440428,
-      "grad_norm": 4.100691318511963,
-      "learning_rate": 9.045736871823829e-06,
-      "loss": 0.7074,
+      "grad_norm": 1.8742183446884155,
+      "learning_rate": 1.3568605307735743e-05,
+      "loss": 0.7499,
       "step": 14550
     },
     {
       "epoch": 2.740447957839262,
-      "grad_norm": 27.86737632751465,
-      "learning_rate": 9.038208168642953e-06,
-      "loss": 0.9491,
+      "grad_norm": 14.247894287109375,
+      "learning_rate": 1.3557312252964427e-05,
+      "loss": 0.8707,
       "step": 14560
     },
     {
       "epoch": 2.742330133634481,
-      "grad_norm": 10.956998825073242,
-      "learning_rate": 9.030679465462074e-06,
-      "loss": 0.9141,
+      "grad_norm": 12.40461254119873,
+      "learning_rate": 1.3546019198193112e-05,
+      "loss": 0.6975,
       "step": 14570
     },
     {
       "epoch": 2.7442123094297006,
-      "grad_norm": 24.600666046142578,
-      "learning_rate": 9.023150762281197e-06,
-      "loss": 0.9744,
+      "grad_norm": 3.423053503036499,
+      "learning_rate": 1.3534726143421796e-05,
+      "loss": 0.9175,
       "step": 14580
     },
     {
       "epoch": 2.74609448522492,
-      "grad_norm": 5.019098281860352,
-      "learning_rate": 9.01562205910032e-06,
-      "loss": 0.7633,
+      "grad_norm": 6.041767597198486,
+      "learning_rate": 1.352343308865048e-05,
+      "loss": 0.6782,
       "step": 14590
     },
     {
       "epoch": 2.7479766610201395,
-      "grad_norm": 15.814457893371582,
-      "learning_rate": 9.008093355919443e-06,
-      "loss": 0.6359,
+      "grad_norm": 29.75321388244629,
+      "learning_rate": 1.3512140033879165e-05,
+      "loss": 0.6639,
       "step": 14600
     },
     {
       "epoch": 2.7498588368153585,
-      "grad_norm": 14.39587116241455,
-      "learning_rate": 9.000564652738568e-06,
-      "loss": 0.7094,
+      "grad_norm": 18.750484466552734,
+      "learning_rate": 1.3500846979107849e-05,
+      "loss": 0.6613,
       "step": 14610
     },
     {
       "epoch": 2.751741012610578,
-      "grad_norm": 19.537870407104492,
-      "learning_rate": 8.993035949557689e-06,
-      "loss": 0.6722,
+      "grad_norm": 15.438997268676758,
+      "learning_rate": 1.3489553924336534e-05,
+      "loss": 0.551,
       "step": 14620
     },
     {
       "epoch": 2.753623188405797,
-      "grad_norm": 22.65885353088379,
-      "learning_rate": 8.985507246376812e-06,
-      "loss": 0.8115,
+      "grad_norm": 14.487974166870117,
+      "learning_rate": 1.3478260869565218e-05,
+      "loss": 0.7749,
       "step": 14630
     },
     {
       "epoch": 2.7555053642010163,
-      "grad_norm": 3.483638048171997,
-      "learning_rate": 8.977978543195935e-06,
-      "loss": 0.5659,
+      "grad_norm": 3.726011276245117,
+      "learning_rate": 1.3466967814793902e-05,
+      "loss": 0.4723,
       "step": 14640
     },
     {
       "epoch": 2.7573875399962358,
-      "grad_norm": 20.589828491210938,
-      "learning_rate": 8.970449840015058e-06,
-      "loss": 0.8191,
+      "grad_norm": 12.075268745422363,
+      "learning_rate": 1.3455674760022587e-05,
+      "loss": 0.7522,
       "step": 14650
     },
     {
       "epoch": 2.759269715791455,
-      "grad_norm": 22.62665557861328,
-      "learning_rate": 8.96292113683418e-06,
-      "loss": 0.7916,
+      "grad_norm": 6.138002872467041,
+      "learning_rate": 1.344438170525127e-05,
+      "loss": 0.7348,
       "step": 14660
     },
     {
       "epoch": 2.761151891586674,
-      "grad_norm": 21.46197509765625,
-      "learning_rate": 8.955392433653304e-06,
-      "loss": 0.8109,
+      "grad_norm": 1.5578501224517822,
+      "learning_rate": 1.3433088650479955e-05,
+      "loss": 0.7188,
       "step": 14670
     },
     {
       "epoch": 2.7630340673818936,
-      "grad_norm": 32.4882698059082,
-      "learning_rate": 8.947863730472427e-06,
-      "loss": 1.0646,
+      "grad_norm": 24.87708282470703,
+      "learning_rate": 1.342179559570864e-05,
+      "loss": 0.8655,
       "step": 14680
     },
     {
       "epoch": 2.7649162431771126,
-      "grad_norm": 7.668648719787598,
-      "learning_rate": 8.94033502729155e-06,
-      "loss": 0.6514,
+      "grad_norm": 13.054447174072266,
+      "learning_rate": 1.3410502540937324e-05,
+      "loss": 0.6821,
       "step": 14690
     },
     {
       "epoch": 2.766798418972332,
-      "grad_norm": 4.4062581062316895,
-      "learning_rate": 8.932806324110672e-06,
-      "loss": 0.7437,
+      "grad_norm": 0.5486339926719666,
+      "learning_rate": 1.3399209486166007e-05,
+      "loss": 0.5678,
       "step": 14700
     },
     {
       "epoch": 2.7686805947675515,
-      "grad_norm": 10.707076072692871,
-      "learning_rate": 8.925277620929795e-06,
-      "loss": 0.7874,
+      "grad_norm": 16.782821655273438,
+      "learning_rate": 1.3387916431394693e-05,
+      "loss": 0.8178,
       "step": 14710
     },
     {
       "epoch": 2.7705627705627704,
-      "grad_norm": 10.452109336853027,
-      "learning_rate": 8.917748917748918e-06,
-      "loss": 0.5787,
+      "grad_norm": 5.81451416015625,
+      "learning_rate": 1.3376623376623377e-05,
+      "loss": 0.5588,
       "step": 14720
     },
     {
       "epoch": 2.77244494635799,
-      "grad_norm": 5.401888370513916,
-      "learning_rate": 8.910220214568041e-06,
-      "loss": 0.824,
+      "grad_norm": 3.6618595123291016,
+      "learning_rate": 1.3365330321852062e-05,
+      "loss": 0.7615,
       "step": 14730
     },
     {
       "epoch": 2.774327122153209,
-      "grad_norm": 7.998766899108887,
-      "learning_rate": 8.902691511387164e-06,
-      "loss": 0.7369,
+      "grad_norm": 8.5778169631958,
+      "learning_rate": 1.3354037267080746e-05,
+      "loss": 0.576,
       "step": 14740
     },
     {
       "epoch": 2.7762092979484283,
-      "grad_norm": 18.264177322387695,
-      "learning_rate": 8.895162808206287e-06,
-      "loss": 0.6624,
+      "grad_norm": 13.566119194030762,
+      "learning_rate": 1.3342744212309429e-05,
+      "loss": 0.5681,
       "step": 14750
     },
     {
       "epoch": 2.7780914737436477,
-      "grad_norm": 2.7713396549224854,
-      "learning_rate": 8.88763410502541e-06,
-      "loss": 0.7402,
+      "grad_norm": 15.843841552734375,
+      "learning_rate": 1.3331451157538115e-05,
+      "loss": 0.6232,
       "step": 14760
     },
     {
       "epoch": 2.779973649538867,
-      "grad_norm": 6.356698989868164,
-      "learning_rate": 8.880105401844533e-06,
-      "loss": 0.5583,
+      "grad_norm": 8.985896110534668,
+      "learning_rate": 1.33201581027668e-05,
+      "loss": 0.538,
       "step": 14770
     },
     {
       "epoch": 2.781855825334086,
-      "grad_norm": 15.435123443603516,
-      "learning_rate": 8.872576698663656e-06,
-      "loss": 0.8285,
+      "grad_norm": 17.511743545532227,
+      "learning_rate": 1.3308865047995484e-05,
+      "loss": 0.8905,
       "step": 14780
     },
     {
       "epoch": 2.7837380011293056,
-      "grad_norm": 18.149913787841797,
-      "learning_rate": 8.865047995482779e-06,
-      "loss": 1.1242,
+      "grad_norm": 23.286775588989258,
+      "learning_rate": 1.3297571993224166e-05,
+      "loss": 0.8913,
       "step": 14790
     },
     {
       "epoch": 2.7856201769245246,
-      "grad_norm": 37.68449783325195,
-      "learning_rate": 8.857519292301902e-06,
-      "loss": 0.7911,
+      "grad_norm": 23.496522903442383,
+      "learning_rate": 1.3286278938452852e-05,
+      "loss": 0.6562,
       "step": 14800
     },
     {
       "epoch": 2.787502352719744,
-      "grad_norm": 12.546880722045898,
-      "learning_rate": 8.849990589121025e-06,
-      "loss": 0.7478,
+      "grad_norm": 11.973505973815918,
+      "learning_rate": 1.3274985883681537e-05,
+      "loss": 0.6814,
       "step": 14810
     },
     {
       "epoch": 2.7893845285149634,
-      "grad_norm": 13.983786582946777,
-      "learning_rate": 8.842461885940147e-06,
-      "loss": 0.4683,
+      "grad_norm": 19.595155715942383,
+      "learning_rate": 1.3263692828910221e-05,
+      "loss": 0.4778,
       "step": 14820
     },
     {
       "epoch": 2.7912667043101824,
-      "grad_norm": 13.797372817993164,
-      "learning_rate": 8.83493318275927e-06,
-      "loss": 0.8398,
+      "grad_norm": 10.94100284576416,
+      "learning_rate": 1.3252399774138904e-05,
+      "loss": 0.8547,
       "step": 14830
     },
     {
       "epoch": 2.793148880105402,
-      "grad_norm": 14.855267524719238,
-      "learning_rate": 8.827404479578393e-06,
-      "loss": 0.9749,
+      "grad_norm": 9.172542572021484,
+      "learning_rate": 1.3241106719367588e-05,
+      "loss": 0.8835,
       "step": 14840
     },
     {
       "epoch": 2.795031055900621,
-      "grad_norm": 7.507328987121582,
-      "learning_rate": 8.819875776397516e-06,
-      "loss": 0.7643,
+      "grad_norm": 49.06556701660156,
+      "learning_rate": 1.3229813664596274e-05,
+      "loss": 0.7209,
       "step": 14850
     },
     {
       "epoch": 2.7969132316958403,
-      "grad_norm": 6.816436290740967,
-      "learning_rate": 8.812347073216639e-06,
-      "loss": 0.7515,
+      "grad_norm": 2.8272016048431396,
+      "learning_rate": 1.3218520609824959e-05,
+      "loss": 0.6661,
       "step": 14860
     },
     {
       "epoch": 2.7987954074910597,
-      "grad_norm": 0.9153765439987183,
-      "learning_rate": 8.804818370035762e-06,
-      "loss": 0.4428,
+      "grad_norm": 0.3546334505081177,
+      "learning_rate": 1.3207227555053643e-05,
+      "loss": 0.4056,
       "step": 14870
     },
     {
       "epoch": 2.800677583286279,
-      "grad_norm": 6.6985673904418945,
-      "learning_rate": 8.797289666854885e-06,
-      "loss": 0.8417,
+      "grad_norm": 3.786240816116333,
+      "learning_rate": 1.3195934500282326e-05,
+      "loss": 0.964,
       "step": 14880
     },
     {
       "epoch": 2.802559759081498,
-      "grad_norm": 8.799405097961426,
-      "learning_rate": 8.789760963674008e-06,
-      "loss": 0.7036,
+      "grad_norm": 5.78407096862793,
+      "learning_rate": 1.318464144551101e-05,
+      "loss": 0.6051,
       "step": 14890
     },
     {
       "epoch": 2.8044419348767176,
-      "grad_norm": 11.702898025512695,
-      "learning_rate": 8.78223226049313e-06,
-      "loss": 0.6547,
+      "grad_norm": 6.689875602722168,
+      "learning_rate": 1.3173348390739696e-05,
+      "loss": 0.6458,
       "step": 14900
     },
     {
       "epoch": 2.8063241106719365,
-      "grad_norm": 8.840179443359375,
-      "learning_rate": 8.774703557312254e-06,
-      "loss": 0.8857,
+      "grad_norm": 12.659087181091309,
+      "learning_rate": 1.316205533596838e-05,
+      "loss": 0.8754,
       "step": 14910
     },
     {
       "epoch": 2.808206286467156,
-      "grad_norm": 17.277437210083008,
-      "learning_rate": 8.767174854131377e-06,
-      "loss": 0.9799,
+      "grad_norm": 14.06718921661377,
+      "learning_rate": 1.3150762281197063e-05,
+      "loss": 0.9916,
       "step": 14920
     },
     {
       "epoch": 2.8100884622623754,
-      "grad_norm": 40.69013977050781,
-      "learning_rate": 8.7596461509505e-06,
-      "loss": 0.8676,
+      "grad_norm": 15.806565284729004,
+      "learning_rate": 1.3139469226425748e-05,
+      "loss": 0.7995,
       "step": 14930
     },
     {
       "epoch": 2.811970638057595,
-      "grad_norm": 2.312723398208618,
-      "learning_rate": 8.752117447769623e-06,
-      "loss": 0.4275,
+      "grad_norm": 1.6029272079467773,
+      "learning_rate": 1.3128176171654434e-05,
+      "loss": 0.4056,
       "step": 14940
     },
     {
       "epoch": 2.813852813852814,
-      "grad_norm": 30.11204719543457,
-      "learning_rate": 8.744588744588745e-06,
-      "loss": 0.8106,
+      "grad_norm": 11.386082649230957,
+      "learning_rate": 1.3116883116883118e-05,
+      "loss": 0.6312,
       "step": 14950
     },
     {
       "epoch": 2.8157349896480333,
-      "grad_norm": 10.502528190612793,
-      "learning_rate": 8.737060041407868e-06,
-      "loss": 0.8292,
+      "grad_norm": 12.017684936523438,
+      "learning_rate": 1.31055900621118e-05,
+      "loss": 0.8528,
       "step": 14960
     },
     {
       "epoch": 2.8176171654432522,
-      "grad_norm": 20.59739875793457,
-      "learning_rate": 8.729531338226991e-06,
-      "loss": 0.5314,
+      "grad_norm": 10.35236644744873,
+      "learning_rate": 1.3094297007340485e-05,
+      "loss": 0.63,
       "step": 14970
     },
     {
       "epoch": 2.8194993412384717,
-      "grad_norm": 8.57652473449707,
-      "learning_rate": 8.722002635046114e-06,
-      "loss": 0.7936,
+      "grad_norm": 18.851110458374023,
+      "learning_rate": 1.308300395256917e-05,
+      "loss": 0.7506,
       "step": 14980
     },
     {
       "epoch": 2.821381517033691,
-      "grad_norm": 8.064692497253418,
-      "learning_rate": 8.714473931865237e-06,
-      "loss": 0.536,
+      "grad_norm": 7.350698947906494,
+      "learning_rate": 1.3071710897797856e-05,
+      "loss": 0.5077,
       "step": 14990
     },
     {
       "epoch": 2.82326369282891,
-      "grad_norm": 9.103423118591309,
-      "learning_rate": 8.70694522868436e-06,
-      "loss": 0.9126,
+      "grad_norm": 9.5648775100708,
+      "learning_rate": 1.306041784302654e-05,
+      "loss": 0.6446,
       "step": 15000
     },
     {
       "epoch": 2.8251458686241295,
-      "grad_norm": 25.442306518554688,
-      "learning_rate": 8.699416525503483e-06,
-      "loss": 0.6525,
+      "grad_norm": 46.55702209472656,
+      "learning_rate": 1.3049124788255223e-05,
+      "loss": 0.6478,
       "step": 15010
     },
     {
       "epoch": 2.8270280444193485,
-      "grad_norm": 17.37743377685547,
-      "learning_rate": 8.691887822322606e-06,
-      "loss": 0.4816,
+      "grad_norm": 22.763042449951172,
+      "learning_rate": 1.3037831733483907e-05,
+      "loss": 0.5402,
       "step": 15020
     },
     {
       "epoch": 2.828910220214568,
-      "grad_norm": 22.00581932067871,
-      "learning_rate": 8.684359119141729e-06,
-      "loss": 0.6523,
+      "grad_norm": 23.328462600708008,
+      "learning_rate": 1.3026538678712591e-05,
+      "loss": 0.7589,
       "step": 15030
     },
     {
       "epoch": 2.8307923960097874,
-      "grad_norm": 13.935546875,
-      "learning_rate": 8.676830415960852e-06,
-      "loss": 0.5463,
+      "grad_norm": 9.546448707580566,
+      "learning_rate": 1.3015245623941278e-05,
+      "loss": 0.4769,
       "step": 15040
     },
     {
       "epoch": 2.832674571805007,
-      "grad_norm": 7.106715679168701,
-      "learning_rate": 8.669301712779975e-06,
-      "loss": 0.7296,
+      "grad_norm": 9.505398750305176,
+      "learning_rate": 1.300395256916996e-05,
+      "loss": 0.6579,
       "step": 15050
     },
     {
       "epoch": 2.834556747600226,
-      "grad_norm": 8.322189331054688,
-      "learning_rate": 8.661773009599098e-06,
-      "loss": 0.7273,
+      "grad_norm": 10.608928680419922,
+      "learning_rate": 1.2992659514398645e-05,
+      "loss": 0.6198,
       "step": 15060
     },
     {
       "epoch": 2.8364389233954452,
-      "grad_norm": 15.4451322555542,
-      "learning_rate": 8.65424430641822e-06,
-      "loss": 0.8192,
+      "grad_norm": 13.843592643737793,
+      "learning_rate": 1.2981366459627329e-05,
+      "loss": 0.74,
       "step": 15070
     },
     {
       "epoch": 2.838321099190664,
-      "grad_norm": 24.261255264282227,
-      "learning_rate": 8.646715603237343e-06,
-      "loss": 0.4825,
+      "grad_norm": 18.67850112915039,
+      "learning_rate": 1.2970073404856015e-05,
+      "loss": 0.4672,
       "step": 15080
     },
     {
       "epoch": 2.8402032749858837,
-      "grad_norm": 6.4923095703125,
-      "learning_rate": 8.639186900056465e-06,
-      "loss": 0.6175,
+      "grad_norm": 6.967514991760254,
+      "learning_rate": 1.2958780350084698e-05,
+      "loss": 0.4409,
       "step": 15090
     },
     {
       "epoch": 2.842085450781103,
-      "grad_norm": 7.1945085525512695,
-      "learning_rate": 8.63165819687559e-06,
-      "loss": 0.8712,
+      "grad_norm": 6.820723533630371,
+      "learning_rate": 1.2947487295313382e-05,
+      "loss": 0.7573,
       "step": 15100
     },
     {
       "epoch": 2.843967626576322,
-      "grad_norm": 9.958353996276855,
-      "learning_rate": 8.624129493694712e-06,
-      "loss": 0.8539,
+      "grad_norm": 18.13920783996582,
+      "learning_rate": 1.2936194240542067e-05,
+      "loss": 0.7914,
       "step": 15110
     },
     {
       "epoch": 2.8458498023715415,
-      "grad_norm": 4.765786647796631,
-      "learning_rate": 8.616600790513835e-06,
-      "loss": 0.6327,
+      "grad_norm": 7.37209939956665,
+      "learning_rate": 1.2924901185770751e-05,
+      "loss": 0.6351,
       "step": 15120
     },
     {
       "epoch": 2.847731978166761,
-      "grad_norm": 0.4407392740249634,
-      "learning_rate": 8.609072087332958e-06,
-      "loss": 0.6043,
+      "grad_norm": 0.3693366050720215,
+      "learning_rate": 1.2913608130999437e-05,
+      "loss": 0.5822,
       "step": 15130
     },
     {
       "epoch": 2.84961415396198,
-      "grad_norm": 12.082877159118652,
-      "learning_rate": 8.601543384152081e-06,
-      "loss": 0.7255,
+      "grad_norm": 21.349327087402344,
+      "learning_rate": 1.290231507622812e-05,
+      "loss": 0.6001,
       "step": 15140
     },
     {
       "epoch": 2.8514963297571994,
-      "grad_norm": 11.160933494567871,
-      "learning_rate": 8.594014680971204e-06,
-      "loss": 0.5107,
+      "grad_norm": 8.462204933166504,
+      "learning_rate": 1.2891022021456804e-05,
+      "loss": 0.4354,
       "step": 15150
     },
     {
       "epoch": 2.853378505552419,
-      "grad_norm": 7.909982681274414,
-      "learning_rate": 8.586485977790327e-06,
-      "loss": 0.7659,
+      "grad_norm": 9.716917991638184,
+      "learning_rate": 1.2879728966685488e-05,
+      "loss": 0.7532,
       "step": 15160
     },
     {
       "epoch": 2.8552606813476378,
-      "grad_norm": 15.593622207641602,
-      "learning_rate": 8.57895727460945e-06,
-      "loss": 0.6901,
+      "grad_norm": 4.279250144958496,
+      "learning_rate": 1.2868435911914173e-05,
+      "loss": 0.7497,
       "step": 15170
     },
     {
       "epoch": 2.857142857142857,
-      "grad_norm": 31.364673614501953,
-      "learning_rate": 8.571428571428571e-06,
-      "loss": 0.5985,
+      "grad_norm": 27.534521102905273,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 0.5156,
       "step": 15180
     },
     {
       "epoch": 2.859025032938076,
-      "grad_norm": 26.913175582885742,
-      "learning_rate": 8.563899868247696e-06,
-      "loss": 0.8037,
+      "grad_norm": 10.771012306213379,
+      "learning_rate": 1.2845849802371542e-05,
+      "loss": 0.7265,
       "step": 15190
     },
     {
       "epoch": 2.8609072087332956,
-      "grad_norm": 15.38956069946289,
-      "learning_rate": 8.556371165066818e-06,
-      "loss": 0.7523,
+      "grad_norm": 18.646312713623047,
+      "learning_rate": 1.2834556747600226e-05,
+      "loss": 0.6741,
       "step": 15200
     },
     {
       "epoch": 2.862789384528515,
-      "grad_norm": 22.167905807495117,
-      "learning_rate": 8.548842461885941e-06,
-      "loss": 1.1415,
+      "grad_norm": 19.257171630859375,
+      "learning_rate": 1.282326369282891e-05,
+      "loss": 1.1365,
       "step": 15210
     },
     {
       "epoch": 2.8646715603237345,
-      "grad_norm": 5.6959075927734375,
-      "learning_rate": 8.541313758705064e-06,
-      "loss": 0.7276,
+      "grad_norm": 5.380772590637207,
+      "learning_rate": 1.2811970638057596e-05,
+      "loss": 0.8546,
       "step": 15220
     },
     {
       "epoch": 2.8665537361189535,
-      "grad_norm": 15.122590065002441,
-      "learning_rate": 8.533785055524186e-06,
-      "loss": 0.7565,
+      "grad_norm": 23.16881561279297,
+      "learning_rate": 1.2800677583286279e-05,
+      "loss": 0.7943,
       "step": 15230
     },
     {
       "epoch": 2.868435911914173,
-      "grad_norm": 5.523094177246094,
-      "learning_rate": 8.52625635234331e-06,
-      "loss": 0.7688,
+      "grad_norm": 3.4778006076812744,
+      "learning_rate": 1.2789384528514964e-05,
+      "loss": 0.6461,
       "step": 15240
     },
     {
       "epoch": 2.870318087709392,
-      "grad_norm": 13.523476600646973,
-      "learning_rate": 8.518727649162433e-06,
-      "loss": 0.8942,
+      "grad_norm": 6.4198994636535645,
+      "learning_rate": 1.2778091473743648e-05,
+      "loss": 0.8293,
       "step": 15250
     },
     {
       "epoch": 2.8722002635046113,
-      "grad_norm": 13.976105690002441,
-      "learning_rate": 8.511198945981556e-06,
-      "loss": 0.7804,
+      "grad_norm": 9.544682502746582,
+      "learning_rate": 1.2766798418972332e-05,
+      "loss": 0.7434,
       "step": 15260
     },
     {
       "epoch": 2.8740824392998308,
-      "grad_norm": 23.4466552734375,
-      "learning_rate": 8.503670242800677e-06,
-      "loss": 0.7666,
+      "grad_norm": 20.545665740966797,
+      "learning_rate": 1.2755505364201017e-05,
+      "loss": 0.6885,
       "step": 15270
     },
     {
       "epoch": 2.8759646150950497,
-      "grad_norm": 20.092809677124023,
-      "learning_rate": 8.496141539619802e-06,
-      "loss": 0.9101,
+      "grad_norm": 21.490854263305664,
+      "learning_rate": 1.2744212309429701e-05,
+      "loss": 0.9683,
       "step": 15280
     },
     {
       "epoch": 2.877846790890269,
-      "grad_norm": 2.974151849746704,
-      "learning_rate": 8.488612836438925e-06,
-      "loss": 0.4844,
+      "grad_norm": 3.391572952270508,
+      "learning_rate": 1.2732919254658385e-05,
+      "loss": 0.421,
       "step": 15290
     },
     {
       "epoch": 2.879728966685488,
-      "grad_norm": 24.60965919494629,
-      "learning_rate": 8.481084133258048e-06,
-      "loss": 0.6742,
+      "grad_norm": 6.413089275360107,
+      "learning_rate": 1.272162619988707e-05,
+      "loss": 0.6439,
       "step": 15300
     },
     {
       "epoch": 2.8816111424807076,
-      "grad_norm": 6.7468485832214355,
-      "learning_rate": 8.473555430077169e-06,
-      "loss": 0.8118,
+      "grad_norm": 9.638285636901855,
+      "learning_rate": 1.2710333145115754e-05,
+      "loss": 0.6487,
       "step": 15310
     },
     {
       "epoch": 2.883493318275927,
-      "grad_norm": 16.004629135131836,
-      "learning_rate": 8.466026726896292e-06,
-      "loss": 0.8272,
+      "grad_norm": 19.76523208618164,
+      "learning_rate": 1.2699040090344439e-05,
+      "loss": 0.7171,
       "step": 15320
     },
     {
       "epoch": 2.8853754940711465,
-      "grad_norm": 9.687468528747559,
-      "learning_rate": 8.458498023715416e-06,
-      "loss": 0.6065,
+      "grad_norm": 18.670621871948242,
+      "learning_rate": 1.2687747035573123e-05,
+      "loss": 0.4666,
       "step": 15330
     },
     {
       "epoch": 2.8872576698663655,
-      "grad_norm": 18.13496971130371,
-      "learning_rate": 8.45096932053454e-06,
-      "loss": 0.7087,
+      "grad_norm": 11.9641695022583,
+      "learning_rate": 1.2676453980801807e-05,
+      "loss": 0.5461,
       "step": 15340
     },
     {
       "epoch": 2.889139845661585,
-      "grad_norm": 5.200069427490234,
-      "learning_rate": 8.443440617353662e-06,
-      "loss": 0.4101,
+      "grad_norm": 1.4184321165084839,
+      "learning_rate": 1.2665160926030492e-05,
+      "loss": 0.407,
       "step": 15350
     },
     {
       "epoch": 2.891022021456804,
-      "grad_norm": 18.469839096069336,
-      "learning_rate": 8.435911914172784e-06,
-      "loss": 0.4233,
+      "grad_norm": 19.386415481567383,
+      "learning_rate": 1.2653867871259176e-05,
+      "loss": 0.4194,
       "step": 15360
     },
     {
       "epoch": 2.8929041972520233,
-      "grad_norm": 27.56670379638672,
-      "learning_rate": 8.428383210991906e-06,
-      "loss": 0.9112,
+      "grad_norm": 25.078582763671875,
+      "learning_rate": 1.264257481648786e-05,
+      "loss": 0.8741,
       "step": 15370
     },
     {
       "epoch": 2.8947863730472427,
-      "grad_norm": 18.819438934326172,
-      "learning_rate": 8.420854507811031e-06,
-      "loss": 0.92,
+      "grad_norm": 18.054174423217773,
+      "learning_rate": 1.2631281761716545e-05,
+      "loss": 0.8567,
       "step": 15380
     },
     {
       "epoch": 2.8966685488424617,
-      "grad_norm": 12.235003471374512,
-      "learning_rate": 8.413325804630154e-06,
-      "loss": 0.9199,
+      "grad_norm": 9.56291675567627,
+      "learning_rate": 1.261998870694523e-05,
+      "loss": 1.0115,
       "step": 15390
     },
     {
       "epoch": 2.898550724637681,
-      "grad_norm": 12.527748107910156,
-      "learning_rate": 8.405797101449275e-06,
-      "loss": 0.6267,
+      "grad_norm": 10.484170913696289,
+      "learning_rate": 1.2608695652173912e-05,
+      "loss": 0.5493,
       "step": 15400
     },
     {
       "epoch": 2.9004329004329006,
-      "grad_norm": 20.517087936401367,
-      "learning_rate": 8.398268398268398e-06,
-      "loss": 0.9461,
+      "grad_norm": 17.50212287902832,
+      "learning_rate": 1.2597402597402598e-05,
+      "loss": 0.8972,
       "step": 15410
     },
     {
       "epoch": 2.9023150762281196,
-      "grad_norm": 7.900396347045898,
-      "learning_rate": 8.390739695087523e-06,
-      "loss": 0.5878,
+      "grad_norm": 11.050121307373047,
+      "learning_rate": 1.2586109542631282e-05,
+      "loss": 0.572,
       "step": 15420
     },
     {
       "epoch": 2.904197252023339,
-      "grad_norm": 12.748804092407227,
-      "learning_rate": 8.383210991906646e-06,
-      "loss": 0.6466,
+      "grad_norm": 19.2404727935791,
+      "learning_rate": 1.2574816487859967e-05,
+      "loss": 0.6457,
       "step": 15430
     },
     {
       "epoch": 2.9060794278185584,
-      "grad_norm": 15.995560646057129,
-      "learning_rate": 8.375682288725767e-06,
-      "loss": 0.5473,
+      "grad_norm": 7.877418518066406,
+      "learning_rate": 1.256352343308865e-05,
+      "loss": 0.5262,
       "step": 15440
     },
     {
       "epoch": 2.9079616036137774,
-      "grad_norm": 25.992582321166992,
-      "learning_rate": 8.36815358554489e-06,
-      "loss": 0.9877,
+      "grad_norm": 25.962966918945312,
+      "learning_rate": 1.2552230378317336e-05,
+      "loss": 1.0387,
       "step": 15450
     },
     {
       "epoch": 2.909843779408997,
-      "grad_norm": 3.78088641166687,
-      "learning_rate": 8.360624882364013e-06,
-      "loss": 0.4233,
+      "grad_norm": 18.10149383544922,
+      "learning_rate": 1.254093732354602e-05,
+      "loss": 0.3264,
       "step": 15460
     },
     {
       "epoch": 2.911725955204216,
-      "grad_norm": 10.031113624572754,
-      "learning_rate": 8.353096179183137e-06,
-      "loss": 0.5838,
+      "grad_norm": 12.390528678894043,
+      "learning_rate": 1.2529644268774704e-05,
+      "loss": 0.5733,
       "step": 15470
     },
     {
       "epoch": 2.9136081309994353,
-      "grad_norm": 2.0009913444519043,
-      "learning_rate": 8.34556747600226e-06,
-      "loss": 0.975,
+      "grad_norm": 6.335510730743408,
+      "learning_rate": 1.2518351214003389e-05,
+      "loss": 0.9647,
       "step": 15480
     },
     {
       "epoch": 2.9154903067946547,
-      "grad_norm": 7.168464660644531,
-      "learning_rate": 8.338038772821381e-06,
-      "loss": 0.6154,
+      "grad_norm": 8.489152908325195,
+      "learning_rate": 1.2507058159232071e-05,
+      "loss": 0.5086,
       "step": 15490
     },
     {
       "epoch": 2.917372482589874,
-      "grad_norm": 18.54973793029785,
-      "learning_rate": 8.330510069640504e-06,
-      "loss": 0.974,
+      "grad_norm": 5.623629570007324,
+      "learning_rate": 1.2495765104460757e-05,
+      "loss": 0.8003,
       "step": 15500
     },
     {
       "epoch": 2.919254658385093,
-      "grad_norm": 1.8958022594451904,
-      "learning_rate": 8.322981366459629e-06,
-      "loss": 0.57,
+      "grad_norm": 7.3673787117004395,
+      "learning_rate": 1.2484472049689442e-05,
+      "loss": 0.5053,
       "step": 15510
     },
     {
       "epoch": 2.9211368341803126,
-      "grad_norm": 10.597874641418457,
-      "learning_rate": 8.315452663278752e-06,
-      "loss": 0.6006,
+      "grad_norm": 25.399425506591797,
+      "learning_rate": 1.2473178994918126e-05,
+      "loss": 0.6455,
       "step": 15520
     },
     {
       "epoch": 2.9230190099755315,
-      "grad_norm": 15.698101043701172,
-      "learning_rate": 8.307923960097873e-06,
-      "loss": 0.5481,
+      "grad_norm": 15.276825904846191,
+      "learning_rate": 1.2461885940146809e-05,
+      "loss": 0.5071,
       "step": 15530
     },
     {
       "epoch": 2.924901185770751,
-      "grad_norm": 5.330072402954102,
-      "learning_rate": 8.300395256916996e-06,
-      "loss": 0.7784,
+      "grad_norm": 11.166011810302734,
+      "learning_rate": 1.2450592885375493e-05,
+      "loss": 0.8913,
       "step": 15540
     },
     {
       "epoch": 2.9267833615659704,
-      "grad_norm": 7.590254783630371,
-      "learning_rate": 8.292866553736119e-06,
-      "loss": 1.1277,
+      "grad_norm": 16.122472763061523,
+      "learning_rate": 1.243929983060418e-05,
+      "loss": 1.0398,
       "step": 15550
     },
     {
       "epoch": 2.9286655373611894,
-      "grad_norm": 12.026838302612305,
-      "learning_rate": 8.285337850555244e-06,
-      "loss": 0.661,
+      "grad_norm": 21.32600212097168,
+      "learning_rate": 1.2428006775832864e-05,
+      "loss": 0.6833,
       "step": 15560
     },
     {
       "epoch": 2.930547713156409,
-      "grad_norm": 1.9510034322738647,
-      "learning_rate": 8.277809147374365e-06,
-      "loss": 0.7917,
+      "grad_norm": 3.505770206451416,
+      "learning_rate": 1.2416713721061546e-05,
+      "loss": 0.7454,
       "step": 15570
     },
     {
       "epoch": 2.932429888951628,
-      "grad_norm": 24.171648025512695,
-      "learning_rate": 8.270280444193488e-06,
-      "loss": 0.6675,
+      "grad_norm": 18.41620445251465,
+      "learning_rate": 1.240542066629023e-05,
+      "loss": 0.5258,
       "step": 15580
     },
     {
       "epoch": 2.9343120647468472,
-      "grad_norm": 19.12851333618164,
-      "learning_rate": 8.26275174101261e-06,
-      "loss": 0.6863,
+      "grad_norm": 25.811620712280273,
+      "learning_rate": 1.2394127611518917e-05,
+      "loss": 0.607,
       "step": 15590
     },
     {
       "epoch": 2.9361942405420667,
-      "grad_norm": 49.385536193847656,
-      "learning_rate": 8.255223037831734e-06,
-      "loss": 1.0035,
+      "grad_norm": 40.27906036376953,
+      "learning_rate": 1.2382834556747601e-05,
+      "loss": 0.8219,
       "step": 15600
     },
     {
       "epoch": 2.938076416337286,
-      "grad_norm": 13.706058502197266,
-      "learning_rate": 8.247694334650858e-06,
-      "loss": 0.9184,
+      "grad_norm": 24.22795867919922,
+      "learning_rate": 1.2371541501976286e-05,
+      "loss": 0.943,
       "step": 15610
     },
     {
       "epoch": 2.939958592132505,
-      "grad_norm": 14.227673530578613,
-      "learning_rate": 8.24016563146998e-06,
-      "loss": 0.8337,
+      "grad_norm": 25.92055892944336,
+      "learning_rate": 1.2360248447204968e-05,
+      "loss": 0.8276,
       "step": 15620
     },
     {
       "epoch": 2.9418407679277245,
-      "grad_norm": 11.792285919189453,
-      "learning_rate": 8.232636928289102e-06,
-      "loss": 0.7102,
+      "grad_norm": 8.324178695678711,
+      "learning_rate": 1.2348955392433653e-05,
+      "loss": 0.6895,
       "step": 15630
     },
     {
       "epoch": 2.9437229437229435,
-      "grad_norm": 11.827733039855957,
-      "learning_rate": 8.225108225108225e-06,
-      "loss": 1.0902,
+      "grad_norm": 22.274948120117188,
+      "learning_rate": 1.2337662337662339e-05,
+      "loss": 1.0351,
       "step": 15640
     },
     {
       "epoch": 2.945605119518163,
-      "grad_norm": 22.433387756347656,
-      "learning_rate": 8.21757952192735e-06,
-      "loss": 0.6356,
+      "grad_norm": 40.22727584838867,
+      "learning_rate": 1.2326369282891023e-05,
+      "loss": 0.5756,
       "step": 15650
     },
     {
       "epoch": 2.9474872953133824,
-      "grad_norm": 2.886334180831909,
-      "learning_rate": 8.210050818746471e-06,
-      "loss": 0.6147,
+      "grad_norm": 2.7665634155273438,
+      "learning_rate": 1.2315076228119706e-05,
+      "loss": 0.5928,
       "step": 15660
     },
     {
       "epoch": 2.949369471108602,
-      "grad_norm": 20.091821670532227,
-      "learning_rate": 8.202522115565594e-06,
-      "loss": 0.8424,
+      "grad_norm": 20.086702346801758,
+      "learning_rate": 1.230378317334839e-05,
+      "loss": 0.7117,
       "step": 15670
     },
     {
       "epoch": 2.951251646903821,
-      "grad_norm": 11.964825630187988,
-      "learning_rate": 8.194993412384717e-06,
-      "loss": 0.8177,
+      "grad_norm": 13.280003547668457,
+      "learning_rate": 1.2292490118577076e-05,
+      "loss": 0.7403,
       "step": 15680
     },
     {
       "epoch": 2.9531338226990402,
-      "grad_norm": 24.54062271118164,
-      "learning_rate": 8.18746470920384e-06,
-      "loss": 1.1969,
+      "grad_norm": 37.245201110839844,
+      "learning_rate": 1.228119706380576e-05,
+      "loss": 0.8726,
       "step": 15690
     },
     {
       "epoch": 2.955015998494259,
-      "grad_norm": 26.192880630493164,
-      "learning_rate": 8.179936006022963e-06,
-      "loss": 0.6056,
+      "grad_norm": 6.580207824707031,
+      "learning_rate": 1.2269904009034443e-05,
+      "loss": 0.6027,
       "step": 15700
     },
     {
       "epoch": 2.9568981742894787,
-      "grad_norm": 13.566141128540039,
-      "learning_rate": 8.172407302842086e-06,
-      "loss": 0.7419,
+      "grad_norm": 1.7421516180038452,
+      "learning_rate": 1.2258610954263128e-05,
+      "loss": 0.6724,
       "step": 15710
     },
     {
       "epoch": 2.958780350084698,
-      "grad_norm": 37.28998565673828,
-      "learning_rate": 8.164878599661209e-06,
-      "loss": 0.5588,
+      "grad_norm": 29.77842140197754,
+      "learning_rate": 1.2247317899491812e-05,
+      "loss": 0.5805,
       "step": 15720
     },
     {
       "epoch": 2.960662525879917,
-      "grad_norm": 16.60234260559082,
-      "learning_rate": 8.157349896480332e-06,
-      "loss": 0.72,
+      "grad_norm": 26.471179962158203,
+      "learning_rate": 1.2236024844720498e-05,
+      "loss": 0.7421,
       "step": 15730
     },
     {
       "epoch": 2.9625447016751365,
-      "grad_norm": 24.26984977722168,
-      "learning_rate": 8.149821193299455e-06,
-      "loss": 0.7045,
+      "grad_norm": 8.780817031860352,
+      "learning_rate": 1.2224731789949183e-05,
+      "loss": 0.7289,
       "step": 15740
     },
     {
       "epoch": 2.9644268774703555,
-      "grad_norm": 6.713445663452148,
-      "learning_rate": 8.142292490118577e-06,
-      "loss": 0.5017,
+      "grad_norm": 14.556037902832031,
+      "learning_rate": 1.2213438735177865e-05,
+      "loss": 0.4937,
       "step": 15750
     },
     {
       "epoch": 2.966309053265575,
-      "grad_norm": 32.575897216796875,
-      "learning_rate": 8.1347637869377e-06,
-      "loss": 0.8543,
+      "grad_norm": 25.941247940063477,
+      "learning_rate": 1.220214568040655e-05,
+      "loss": 0.8835,
       "step": 15760
     },
     {
       "epoch": 2.9681912290607944,
-      "grad_norm": 3.385469913482666,
-      "learning_rate": 8.127235083756823e-06,
-      "loss": 0.7149,
+      "grad_norm": 6.276753902435303,
+      "learning_rate": 1.2190852625635234e-05,
+      "loss": 0.6071,
       "step": 15770
     },
     {
       "epoch": 2.970073404856014,
-      "grad_norm": 29.050565719604492,
-      "learning_rate": 8.119706380575946e-06,
-      "loss": 1.0051,
+      "grad_norm": 24.19154930114746,
+      "learning_rate": 1.217955957086392e-05,
+      "loss": 0.907,
       "step": 15780
     },
     {
       "epoch": 2.9719555806512328,
-      "grad_norm": 2.618756055831909,
-      "learning_rate": 8.112177677395069e-06,
-      "loss": 0.6576,
+      "grad_norm": 5.083404064178467,
+      "learning_rate": 1.2168266516092603e-05,
+      "loss": 0.5493,
       "step": 15790
     },
     {
       "epoch": 2.973837756446452,
-      "grad_norm": 11.132329940795898,
-      "learning_rate": 8.104648974214192e-06,
-      "loss": 0.703,
+      "grad_norm": 9.346044540405273,
+      "learning_rate": 1.2156973461321287e-05,
+      "loss": 0.6961,
       "step": 15800
     },
     {
       "epoch": 2.975719932241671,
-      "grad_norm": 1.447586178779602,
-      "learning_rate": 8.097120271033315e-06,
-      "loss": 0.7784,
+      "grad_norm": 1.4552770853042603,
+      "learning_rate": 1.2145680406549972e-05,
+      "loss": 0.6451,
       "step": 15810
     },
     {
       "epoch": 2.9776021080368906,
-      "grad_norm": 13.071996688842773,
-      "learning_rate": 8.089591567852438e-06,
-      "loss": 0.7363,
+      "grad_norm": 17.02495765686035,
+      "learning_rate": 1.2134387351778658e-05,
+      "loss": 0.5733,
       "step": 15820
     },
     {
       "epoch": 2.97948428383211,
-      "grad_norm": 8.223648071289062,
-      "learning_rate": 8.08206286467156e-06,
-      "loss": 0.7832,
+      "grad_norm": 2.2290263175964355,
+      "learning_rate": 1.2123094297007342e-05,
+      "loss": 0.7652,
       "step": 15830
     },
     {
       "epoch": 2.981366459627329,
-      "grad_norm": 7.399938583374023,
-      "learning_rate": 8.074534161490684e-06,
-      "loss": 0.8243,
+      "grad_norm": 20.987024307250977,
+      "learning_rate": 1.2111801242236025e-05,
+      "loss": 0.7364,
       "step": 15840
     },
     {
       "epoch": 2.9832486354225485,
-      "grad_norm": 23.568254470825195,
-      "learning_rate": 8.067005458309807e-06,
-      "loss": 1.0995,
+      "grad_norm": 18.569358825683594,
+      "learning_rate": 1.2100508187464709e-05,
+      "loss": 1.0693,
       "step": 15850
     },
     {
       "epoch": 2.9851308112177675,
-      "grad_norm": 18.40585708618164,
-      "learning_rate": 8.05947675512893e-06,
-      "loss": 0.8484,
+      "grad_norm": 22.19905662536621,
+      "learning_rate": 1.2089215132693394e-05,
+      "loss": 0.8939,
       "step": 15860
     },
     {
       "epoch": 2.987012987012987,
-      "grad_norm": 7.398470401763916,
-      "learning_rate": 8.051948051948052e-06,
-      "loss": 0.7103,
+      "grad_norm": 20.058185577392578,
+      "learning_rate": 1.207792207792208e-05,
+      "loss": 0.5969,
       "step": 15870
     },
     {
       "epoch": 2.9888951628082063,
-      "grad_norm": 41.91292190551758,
-      "learning_rate": 8.044419348767175e-06,
-      "loss": 0.376,
+      "grad_norm": 14.325201034545898,
+      "learning_rate": 1.2066629023150762e-05,
+      "loss": 0.4321,
       "step": 15880
     },
     {
       "epoch": 2.9907773386034258,
-      "grad_norm": 22.689844131469727,
-      "learning_rate": 8.036890645586298e-06,
-      "loss": 0.8109,
+      "grad_norm": 3.7948050498962402,
+      "learning_rate": 1.2055335968379447e-05,
+      "loss": 0.6696,
       "step": 15890
     },
     {
       "epoch": 2.9926595143986447,
-      "grad_norm": 18.084108352661133,
-      "learning_rate": 8.029361942405421e-06,
-      "loss": 0.7224,
+      "grad_norm": 10.829690933227539,
+      "learning_rate": 1.2044042913608131e-05,
+      "loss": 0.6765,
       "step": 15900
     },
     {
       "epoch": 2.994541690193864,
-      "grad_norm": 20.86298179626465,
-      "learning_rate": 8.021833239224544e-06,
-      "loss": 0.7326,
+      "grad_norm": 16.70258903503418,
+      "learning_rate": 1.2032749858836815e-05,
+      "loss": 0.5554,
       "step": 15910
     },
     {
       "epoch": 2.996423865989083,
-      "grad_norm": 15.901738166809082,
-      "learning_rate": 8.014304536043667e-06,
-      "loss": 0.5458,
+      "grad_norm": 8.688410758972168,
+      "learning_rate": 1.20214568040655e-05,
+      "loss": 0.542,
       "step": 15920
     },
     {
       "epoch": 2.9983060417843026,
-      "grad_norm": 28.315481185913086,
-      "learning_rate": 8.00677583286279e-06,
-      "loss": 0.732,
+      "grad_norm": 7.034015655517578,
+      "learning_rate": 1.2010163749294184e-05,
+      "loss": 0.5749,
       "step": 15930
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.9138666666666667,
-      "eval_loss": 0.3551448881626129,
-      "eval_runtime": 124.8647,
-      "eval_samples_per_second": 60.065,
-      "eval_steps_per_second": 7.512,
+      "eval_accuracy": 0.9153333333333333,
+      "eval_loss": 0.32141387462615967,
+      "eval_runtime": 319.8636,
+      "eval_samples_per_second": 23.447,
+      "eval_steps_per_second": 2.932,
       "step": 15939
     },
     {
       "epoch": 3.000188217579522,
-      "grad_norm": 24.23802947998047,
-      "learning_rate": 7.999247129681913e-06,
-      "loss": 0.6504,
+      "grad_norm": 13.433454513549805,
+      "learning_rate": 1.1998870694522869e-05,
+      "loss": 0.4825,
       "step": 15940
     },
     {
       "epoch": 3.002070393374741,
-      "grad_norm": 23.428064346313477,
-      "learning_rate": 7.991718426501036e-06,
-      "loss": 0.8482,
+      "grad_norm": 18.549455642700195,
+      "learning_rate": 1.1987577639751553e-05,
+      "loss": 0.7692,
       "step": 15950
     },
     {
       "epoch": 3.0039525691699605,
-      "grad_norm": 17.107192993164062,
-      "learning_rate": 7.984189723320159e-06,
-      "loss": 0.8003,
+      "grad_norm": 22.799373626708984,
+      "learning_rate": 1.1976284584980239e-05,
+      "loss": 0.6418,
       "step": 15960
     },
     {
       "epoch": 3.00583474496518,
-      "grad_norm": 23.277639389038086,
-      "learning_rate": 7.976661020139282e-06,
-      "loss": 0.669,
+      "grad_norm": 21.21796989440918,
+      "learning_rate": 1.1964991530208922e-05,
+      "loss": 0.5274,
       "step": 15970
     },
     {
       "epoch": 3.007716920760399,
-      "grad_norm": 24.554271697998047,
-      "learning_rate": 7.969132316958405e-06,
-      "loss": 0.6349,
+      "grad_norm": 25.831186294555664,
+      "learning_rate": 1.1953698475437606e-05,
+      "loss": 0.5339,
       "step": 15980
     },
     {
       "epoch": 3.0095990965556183,
-      "grad_norm": 16.340686798095703,
-      "learning_rate": 7.961603613777528e-06,
-      "loss": 0.9792,
+      "grad_norm": 9.926560401916504,
+      "learning_rate": 1.194240542066629e-05,
+      "loss": 0.8981,
       "step": 15990
     },
     {
       "epoch": 3.0114812723508377,
-      "grad_norm": 22.233644485473633,
-      "learning_rate": 7.95407491059665e-06,
-      "loss": 0.8113,
+      "grad_norm": 22.619985580444336,
+      "learning_rate": 1.1931112365894975e-05,
+      "loss": 0.7825,
       "step": 16000
     },
     {
       "epoch": 3.0133634481460567,
-      "grad_norm": 6.466579914093018,
-      "learning_rate": 7.946546207415773e-06,
-      "loss": 0.4841,
+      "grad_norm": 12.743892669677734,
+      "learning_rate": 1.191981931112366e-05,
+      "loss": 0.4295,
       "step": 16010
     },
     {
       "epoch": 3.015245623941276,
-      "grad_norm": 7.007059574127197,
-      "learning_rate": 7.939017504234896e-06,
-      "loss": 0.9804,
+      "grad_norm": 13.726835250854492,
+      "learning_rate": 1.1908526256352344e-05,
+      "loss": 0.8777,
       "step": 16020
     },
     {
       "epoch": 3.0171277997364956,
-      "grad_norm": 9.863162994384766,
-      "learning_rate": 7.93148880105402e-06,
-      "loss": 0.5055,
+      "grad_norm": 14.803107261657715,
+      "learning_rate": 1.1897233201581028e-05,
+      "loss": 0.3934,
       "step": 16030
     },
     {
       "epoch": 3.0190099755317146,
-      "grad_norm": 27.062816619873047,
-      "learning_rate": 7.923960097873142e-06,
-      "loss": 0.6065,
+      "grad_norm": 41.224464416503906,
+      "learning_rate": 1.1885940146809712e-05,
+      "loss": 0.459,
       "step": 16040
     },
     {
       "epoch": 3.020892151326934,
-      "grad_norm": 8.803095817565918,
-      "learning_rate": 7.916431394692265e-06,
-      "loss": 0.3608,
+      "grad_norm": 14.260553359985352,
+      "learning_rate": 1.1874647092038397e-05,
+      "loss": 0.3349,
       "step": 16050
     },
     {
       "epoch": 3.022774327122153,
-      "grad_norm": 2.467694044113159,
-      "learning_rate": 7.908902691511388e-06,
-      "loss": 0.7722,
+      "grad_norm": 2.4585812091827393,
+      "learning_rate": 1.1863354037267081e-05,
+      "loss": 0.8261,
       "step": 16060
     },
     {
       "epoch": 3.0246565029173724,
-      "grad_norm": 8.088591575622559,
-      "learning_rate": 7.901373988330511e-06,
-      "loss": 0.5181,
+      "grad_norm": 15.067172050476074,
+      "learning_rate": 1.1852060982495766e-05,
+      "loss": 0.4319,
       "step": 16070
     },
     {
       "epoch": 3.026538678712592,
-      "grad_norm": 18.756155014038086,
-      "learning_rate": 7.893845285149634e-06,
-      "loss": 0.7348,
+      "grad_norm": 16.329282760620117,
+      "learning_rate": 1.184076792772445e-05,
+      "loss": 0.7257,
       "step": 16080
     },
     {
       "epoch": 3.028420854507811,
-      "grad_norm": 16.53417205810547,
-      "learning_rate": 7.886316581968757e-06,
-      "loss": 0.4264,
+      "grad_norm": 12.787527084350586,
+      "learning_rate": 1.1829474872953134e-05,
+      "loss": 0.3546,
       "step": 16090
     },
     {
       "epoch": 3.0303030303030303,
-      "grad_norm": 6.082591533660889,
-      "learning_rate": 7.87878787878788e-06,
-      "loss": 0.6317,
+      "grad_norm": 3.467725992202759,
+      "learning_rate": 1.1818181818181819e-05,
+      "loss": 0.6743,
       "step": 16100
     },
     {
       "epoch": 3.0321852060982497,
-      "grad_norm": 29.989810943603516,
-      "learning_rate": 7.871259175607003e-06,
-      "loss": 0.4669,
+      "grad_norm": 26.06414794921875,
+      "learning_rate": 1.1806888763410503e-05,
+      "loss": 0.3207,
       "step": 16110
     },
     {
       "epoch": 3.0340673818934687,
-      "grad_norm": 9.16569995880127,
-      "learning_rate": 7.863730472426126e-06,
-      "loss": 0.6771,
+      "grad_norm": 18.53022003173828,
+      "learning_rate": 1.1795595708639187e-05,
+      "loss": 0.6154,
       "step": 16120
     },
     {
       "epoch": 3.035949557688688,
-      "grad_norm": 9.343409538269043,
-      "learning_rate": 7.856201769245248e-06,
-      "loss": 0.3368,
+      "grad_norm": 7.747043132781982,
+      "learning_rate": 1.1784302653867872e-05,
+      "loss": 0.2841,
       "step": 16130
     },
     {
       "epoch": 3.0378317334839076,
-      "grad_norm": 15.900422096252441,
-      "learning_rate": 7.848673066064371e-06,
-      "loss": 0.6708,
+      "grad_norm": 8.015532493591309,
+      "learning_rate": 1.1773009599096555e-05,
+      "loss": 0.5264,
       "step": 16140
     },
     {
       "epoch": 3.0397139092791265,
-      "grad_norm": 4.078092098236084,
-      "learning_rate": 7.841144362883494e-06,
-      "loss": 0.5507,
+      "grad_norm": 7.965919494628906,
+      "learning_rate": 1.176171654432524e-05,
+      "loss": 0.566,
       "step": 16150
     },
     {
       "epoch": 3.041596085074346,
-      "grad_norm": 1.3915058374404907,
-      "learning_rate": 7.833615659702617e-06,
-      "loss": 0.5978,
+      "grad_norm": 0.4477425813674927,
+      "learning_rate": 1.1750423489553925e-05,
+      "loss": 0.5456,
       "step": 16160
     },
     {
       "epoch": 3.0434782608695654,
-      "grad_norm": 5.797540187835693,
-      "learning_rate": 7.82608695652174e-06,
-      "loss": 0.8542,
+      "grad_norm": 6.554590702056885,
+      "learning_rate": 1.173913043478261e-05,
+      "loss": 1.0002,
       "step": 16170
     },
     {
       "epoch": 3.0453604366647844,
-      "grad_norm": 11.26925277709961,
-      "learning_rate": 7.818558253340861e-06,
-      "loss": 0.7604,
+      "grad_norm": 12.476119041442871,
+      "learning_rate": 1.1727837380011292e-05,
+      "loss": 0.8468,
       "step": 16180
     },
     {
       "epoch": 3.047242612460004,
-      "grad_norm": 11.07471752166748,
-      "learning_rate": 7.811029550159986e-06,
-      "loss": 0.8315,
+      "grad_norm": 13.832260131835938,
+      "learning_rate": 1.1716544325239978e-05,
+      "loss": 0.7469,
       "step": 16190
     },
     {
       "epoch": 3.049124788255223,
-      "grad_norm": 2.0462937355041504,
-      "learning_rate": 7.803500846979109e-06,
-      "loss": 0.7766,
+      "grad_norm": 1.2430438995361328,
+      "learning_rate": 1.1705251270468663e-05,
+      "loss": 0.7722,
       "step": 16200
     },
     {
       "epoch": 3.0510069640504422,
-      "grad_norm": 14.226692199707031,
-      "learning_rate": 7.795972143798232e-06,
-      "loss": 0.9066,
+      "grad_norm": 14.875078201293945,
+      "learning_rate": 1.1693958215697347e-05,
+      "loss": 0.7945,
       "step": 16210
     },
     {
       "epoch": 3.0528891398456617,
-      "grad_norm": 19.502113342285156,
-      "learning_rate": 7.788443440617355e-06,
-      "loss": 0.8325,
+      "grad_norm": 5.486425399780273,
+      "learning_rate": 1.1682665160926031e-05,
+      "loss": 0.696,
       "step": 16220
     },
     {
       "epoch": 3.0547713156408807,
-      "grad_norm": 9.413043022155762,
-      "learning_rate": 7.780914737436478e-06,
-      "loss": 0.5072,
+      "grad_norm": 12.678677558898926,
+      "learning_rate": 1.1671372106154714e-05,
+      "loss": 0.339,
       "step": 16230
     },
     {
       "epoch": 3.0566534914361,
-      "grad_norm": 18.830753326416016,
-      "learning_rate": 7.7733860342556e-06,
-      "loss": 0.6074,
+      "grad_norm": 21.625308990478516,
+      "learning_rate": 1.16600790513834e-05,
+      "loss": 0.6191,
       "step": 16240
     },
     {
       "epoch": 3.0585356672313195,
-      "grad_norm": 21.968721389770508,
-      "learning_rate": 7.765857331074724e-06,
-      "loss": 0.7097,
+      "grad_norm": 41.28392791748047,
+      "learning_rate": 1.1648785996612084e-05,
+      "loss": 0.6735,
       "step": 16250
     },
     {
       "epoch": 3.0604178430265385,
-      "grad_norm": 0.984065592288971,
-      "learning_rate": 7.758328627893846e-06,
-      "loss": 0.7093,
+      "grad_norm": 2.554927349090576,
+      "learning_rate": 1.1637492941840769e-05,
+      "loss": 0.7547,
       "step": 16260
     },
     {
       "epoch": 3.062300018821758,
-      "grad_norm": 1.9321556091308594,
-      "learning_rate": 7.750799924712968e-06,
-      "loss": 0.7335,
+      "grad_norm": 0.5858107805252075,
+      "learning_rate": 1.1626199887069451e-05,
+      "loss": 0.6705,
       "step": 16270
     },
     {
       "epoch": 3.0641821946169774,
-      "grad_norm": 17.950807571411133,
-      "learning_rate": 7.743271221532092e-06,
-      "loss": 0.615,
+      "grad_norm": 35.934181213378906,
+      "learning_rate": 1.1614906832298136e-05,
+      "loss": 0.5326,
       "step": 16280
     },
     {
       "epoch": 3.0660643704121964,
-      "grad_norm": 10.872349739074707,
-      "learning_rate": 7.735742518351215e-06,
-      "loss": 0.7945,
+      "grad_norm": 11.104010581970215,
+      "learning_rate": 1.1603613777526822e-05,
+      "loss": 0.8268,
       "step": 16290
     },
     {
       "epoch": 3.067946546207416,
-      "grad_norm": 9.358575820922852,
-      "learning_rate": 7.728213815170338e-06,
-      "loss": 0.6757,
+      "grad_norm": 6.697833061218262,
+      "learning_rate": 1.1592320722755506e-05,
+      "loss": 0.6316,
       "step": 16300
     },
     {
       "epoch": 3.0698287220026352,
-      "grad_norm": 22.87818717956543,
-      "learning_rate": 7.72068511198946e-06,
-      "loss": 0.5002,
+      "grad_norm": 22.01197052001953,
+      "learning_rate": 1.1581027667984189e-05,
+      "loss": 0.5534,
       "step": 16310
     },
     {
       "epoch": 3.0717108977978542,
-      "grad_norm": 14.869016647338867,
-      "learning_rate": 7.713156408808582e-06,
-      "loss": 0.7588,
+      "grad_norm": 5.453395843505859,
+      "learning_rate": 1.1569734613212873e-05,
+      "loss": 0.7422,
       "step": 16320
     },
     {
       "epoch": 3.0735930735930737,
-      "grad_norm": 8.90829849243164,
-      "learning_rate": 7.705627705627707e-06,
-      "loss": 1.0797,
+      "grad_norm": 11.562372207641602,
+      "learning_rate": 1.155844155844156e-05,
+      "loss": 0.9676,
       "step": 16330
     },
     {
       "epoch": 3.075475249388293,
-      "grad_norm": 27.344745635986328,
-      "learning_rate": 7.69809900244683e-06,
-      "loss": 0.5867,
+      "grad_norm": 12.74392032623291,
+      "learning_rate": 1.1547148503670244e-05,
+      "loss": 0.6064,
       "step": 16340
     },
     {
       "epoch": 3.077357425183512,
-      "grad_norm": 26.566848754882812,
-      "learning_rate": 7.690570299265953e-06,
-      "loss": 0.5187,
+      "grad_norm": 20.787565231323242,
+      "learning_rate": 1.1535855448898928e-05,
+      "loss": 0.4245,
       "step": 16350
     },
     {
       "epoch": 3.0792396009787315,
-      "grad_norm": 36.06024932861328,
-      "learning_rate": 7.683041596085074e-06,
-      "loss": 0.8335,
+      "grad_norm": 27.309429168701172,
+      "learning_rate": 1.1524562394127611e-05,
+      "loss": 0.7836,
       "step": 16360
     },
     {
       "epoch": 3.0811217767739505,
-      "grad_norm": 8.580551147460938,
-      "learning_rate": 7.675512892904199e-06,
-      "loss": 0.6761,
+      "grad_norm": 4.994350433349609,
+      "learning_rate": 1.1513269339356295e-05,
+      "loss": 0.7593,
       "step": 16370
     },
     {
       "epoch": 3.08300395256917,
-      "grad_norm": 36.79864501953125,
-      "learning_rate": 7.667984189723321e-06,
-      "loss": 0.8464,
+      "grad_norm": 21.30853271484375,
+      "learning_rate": 1.1501976284584981e-05,
+      "loss": 0.6651,
       "step": 16380
     },
     {
       "epoch": 3.0848861283643894,
-      "grad_norm": 75.77937316894531,
-      "learning_rate": 7.660455486542444e-06,
-      "loss": 0.7556,
+      "grad_norm": 36.19512939453125,
+      "learning_rate": 1.1490683229813666e-05,
+      "loss": 0.6865,
       "step": 16390
     },
     {
       "epoch": 3.0867683041596083,
-      "grad_norm": 0.26863107085227966,
-      "learning_rate": 7.652926783361566e-06,
-      "loss": 0.7862,
+      "grad_norm": 0.0781698077917099,
+      "learning_rate": 1.1479390175042348e-05,
+      "loss": 0.7204,
       "step": 16400
     },
     {
       "epoch": 3.0886504799548278,
-      "grad_norm": 19.25002670288086,
-      "learning_rate": 7.645398080180689e-06,
-      "loss": 0.6988,
+      "grad_norm": 6.799276351928711,
+      "learning_rate": 1.1468097120271033e-05,
+      "loss": 0.606,
       "step": 16410
     },
     {
       "epoch": 3.090532655750047,
-      "grad_norm": 5.37202262878418,
-      "learning_rate": 7.637869376999813e-06,
-      "loss": 0.7596,
+      "grad_norm": 3.7640774250030518,
+      "learning_rate": 1.1456804065499717e-05,
+      "loss": 0.6198,
       "step": 16420
     },
     {
       "epoch": 3.092414831545266,
-      "grad_norm": 21.078012466430664,
-      "learning_rate": 7.630340673818936e-06,
-      "loss": 0.6614,
+      "grad_norm": 16.729280471801758,
+      "learning_rate": 1.1445511010728403e-05,
+      "loss": 0.6511,
       "step": 16430
     },
     {
       "epoch": 3.0942970073404856,
-      "grad_norm": 26.502809524536133,
-      "learning_rate": 7.622811970638057e-06,
-      "loss": 0.5694,
+      "grad_norm": 24.438823699951172,
+      "learning_rate": 1.1434217955957086e-05,
+      "loss": 0.4929,
       "step": 16440
     },
     {
       "epoch": 3.096179183135705,
-      "grad_norm": 3.8008010387420654,
-      "learning_rate": 7.615283267457181e-06,
-      "loss": 0.5919,
+      "grad_norm": 5.958948612213135,
+      "learning_rate": 1.142292490118577e-05,
+      "loss": 0.5966,
       "step": 16450
     },
     {
       "epoch": 3.098061358930924,
-      "grad_norm": 2.185805320739746,
-      "learning_rate": 7.607754564276304e-06,
-      "loss": 1.0339,
+      "grad_norm": 1.2739803791046143,
+      "learning_rate": 1.1411631846414455e-05,
+      "loss": 0.9081,
       "step": 16460
     },
     {
       "epoch": 3.0999435347261435,
-      "grad_norm": 1.0301812887191772,
-      "learning_rate": 7.600225861095427e-06,
-      "loss": 0.4373,
+      "grad_norm": 0.8364112377166748,
+      "learning_rate": 1.140033879164314e-05,
+      "loss": 0.4441,
       "step": 16470
     },
     {
       "epoch": 3.1018257105213625,
-      "grad_norm": 13.050724983215332,
-      "learning_rate": 7.592697157914551e-06,
-      "loss": 0.8136,
+      "grad_norm": 11.922718048095703,
+      "learning_rate": 1.1389045736871825e-05,
+      "loss": 0.669,
       "step": 16480
     },
     {
       "epoch": 3.103707886316582,
-      "grad_norm": 17.977001190185547,
-      "learning_rate": 7.585168454733673e-06,
-      "loss": 0.5327,
+      "grad_norm": 12.48450756072998,
+      "learning_rate": 1.1377752682100508e-05,
+      "loss": 0.4577,
       "step": 16490
     },
     {
       "epoch": 3.1055900621118013,
-      "grad_norm": 14.840492248535156,
-      "learning_rate": 7.577639751552796e-06,
-      "loss": 0.5607,
+      "grad_norm": 22.12200164794922,
+      "learning_rate": 1.1366459627329192e-05,
+      "loss": 0.4632,
       "step": 16500
     },
     {
       "epoch": 3.1074722379070203,
-      "grad_norm": 9.116665840148926,
-      "learning_rate": 7.570111048371919e-06,
-      "loss": 0.7519,
+      "grad_norm": 21.450420379638672,
+      "learning_rate": 1.1355166572557877e-05,
+      "loss": 0.798,
       "step": 16510
     },
     {
       "epoch": 3.1093544137022397,
-      "grad_norm": 1.0839731693267822,
-      "learning_rate": 7.5625823451910415e-06,
-      "loss": 0.726,
+      "grad_norm": 0.42957913875579834,
+      "learning_rate": 1.1343873517786563e-05,
+      "loss": 0.5381,
       "step": 16520
     },
     {
       "epoch": 3.111236589497459,
-      "grad_norm": 11.97722339630127,
-      "learning_rate": 7.555053642010164e-06,
-      "loss": 0.681,
+      "grad_norm": 5.7693281173706055,
+      "learning_rate": 1.1332580463015245e-05,
+      "loss": 0.5663,
       "step": 16530
     },
     {
       "epoch": 3.113118765292678,
-      "grad_norm": 28.19617462158203,
-      "learning_rate": 7.547524938829287e-06,
-      "loss": 0.5871,
+      "grad_norm": 25.847362518310547,
+      "learning_rate": 1.132128740824393e-05,
+      "loss": 0.5048,
       "step": 16540
     },
     {
       "epoch": 3.1150009410878976,
-      "grad_norm": 31.05242347717285,
-      "learning_rate": 7.53999623564841e-06,
-      "loss": 0.5718,
+      "grad_norm": 3.086296319961548,
+      "learning_rate": 1.1309994353472614e-05,
+      "loss": 0.5231,
       "step": 16550
     },
     {
       "epoch": 3.116883116883117,
-      "grad_norm": 8.402994155883789,
-      "learning_rate": 7.532467532467533e-06,
-      "loss": 0.6193,
+      "grad_norm": 6.833352088928223,
+      "learning_rate": 1.12987012987013e-05,
+      "loss": 0.6046,
       "step": 16560
     },
     {
       "epoch": 3.118765292678336,
-      "grad_norm": 10.946696281433105,
-      "learning_rate": 7.524938829286656e-06,
-      "loss": 0.5893,
+      "grad_norm": 11.189995765686035,
+      "learning_rate": 1.1287408243929985e-05,
+      "loss": 0.4659,
       "step": 16570
     },
     {
       "epoch": 3.1206474684735555,
-      "grad_norm": 29.695755004882812,
-      "learning_rate": 7.517410126105778e-06,
-      "loss": 0.8272,
+      "grad_norm": 17.51889419555664,
+      "learning_rate": 1.1276115189158667e-05,
+      "loss": 0.7338,
       "step": 16580
     },
     {
       "epoch": 3.122529644268775,
-      "grad_norm": 12.49916934967041,
-      "learning_rate": 7.509881422924902e-06,
-      "loss": 0.3927,
+      "grad_norm": 18.669206619262695,
+      "learning_rate": 1.1264822134387352e-05,
+      "loss": 0.3663,
       "step": 16590
     },
     {
       "epoch": 3.124411820063994,
-      "grad_norm": 22.52407455444336,
-      "learning_rate": 7.502352719744025e-06,
-      "loss": 0.8379,
+      "grad_norm": 4.161768436431885,
+      "learning_rate": 1.1253529079616036e-05,
+      "loss": 0.7614,
       "step": 16600
     },
     {
       "epoch": 3.1262939958592133,
-      "grad_norm": 20.72262191772461,
-      "learning_rate": 7.494824016563148e-06,
-      "loss": 0.8475,
+      "grad_norm": 17.522029876708984,
+      "learning_rate": 1.1242236024844722e-05,
+      "loss": 0.7639,
       "step": 16610
     },
     {
       "epoch": 3.1281761716544327,
-      "grad_norm": 18.68262481689453,
-      "learning_rate": 7.48729531338227e-06,
-      "loss": 0.5057,
+      "grad_norm": 20.829774856567383,
+      "learning_rate": 1.1230942970073405e-05,
+      "loss": 0.5254,
       "step": 16620
     },
     {
       "epoch": 3.1300583474496517,
-      "grad_norm": 21.154441833496094,
-      "learning_rate": 7.479766610201394e-06,
-      "loss": 0.6002,
+      "grad_norm": 12.240787506103516,
+      "learning_rate": 1.121964991530209e-05,
+      "loss": 0.6138,
       "step": 16630
     },
     {
       "epoch": 3.131940523244871,
-      "grad_norm": 33.61641311645508,
-      "learning_rate": 7.472237907020517e-06,
-      "loss": 0.7546,
+      "grad_norm": 18.72231101989746,
+      "learning_rate": 1.1208356860530774e-05,
+      "loss": 0.8026,
       "step": 16640
     },
     {
       "epoch": 3.13382269904009,
-      "grad_norm": 54.20226287841797,
-      "learning_rate": 7.4647092038396395e-06,
-      "loss": 0.5867,
+      "grad_norm": 8.982686042785645,
+      "learning_rate": 1.1197063805759458e-05,
+      "loss": 0.5638,
       "step": 16650
     },
     {
       "epoch": 3.1357048748353096,
-      "grad_norm": 0.9356986284255981,
-      "learning_rate": 7.457180500658762e-06,
-      "loss": 0.6669,
+      "grad_norm": 0.6803452968597412,
+      "learning_rate": 1.1185770750988142e-05,
+      "loss": 0.6837,
       "step": 16660
     },
     {
       "epoch": 3.137587050630529,
-      "grad_norm": 2.053644895553589,
-      "learning_rate": 7.4496517974778845e-06,
-      "loss": 0.6829,
+      "grad_norm": 2.5247535705566406,
+      "learning_rate": 1.1174477696216827e-05,
+      "loss": 0.6615,
       "step": 16670
     },
     {
       "epoch": 3.139469226425748,
-      "grad_norm": 8.472742080688477,
-      "learning_rate": 7.442123094297008e-06,
-      "loss": 0.6525,
+      "grad_norm": 7.049471378326416,
+      "learning_rate": 1.1163184641445511e-05,
+      "loss": 0.5665,
       "step": 16680
     },
     {
       "epoch": 3.1413514022209674,
-      "grad_norm": 19.01861000061035,
-      "learning_rate": 7.434594391116131e-06,
-      "loss": 0.9197,
+      "grad_norm": 22.32665252685547,
+      "learning_rate": 1.1151891586674196e-05,
+      "loss": 0.878,
       "step": 16690
     },
     {
       "epoch": 3.143233578016187,
-      "grad_norm": 100.38238525390625,
-      "learning_rate": 7.427065687935254e-06,
-      "loss": 0.6449,
+      "grad_norm": 18.906843185424805,
+      "learning_rate": 1.1140598531902882e-05,
+      "loss": 0.6834,
       "step": 16700
     },
     {
       "epoch": 3.145115753811406,
-      "grad_norm": 7.82566499710083,
-      "learning_rate": 7.419536984754376e-06,
-      "loss": 0.7607,
+      "grad_norm": 0.7685168385505676,
+      "learning_rate": 1.1129305477131564e-05,
+      "loss": 0.6544,
       "step": 16710
     },
     {
       "epoch": 3.1469979296066253,
-      "grad_norm": 1.0181591510772705,
-      "learning_rate": 7.412008281573499e-06,
-      "loss": 0.7252,
+      "grad_norm": 0.6841329336166382,
+      "learning_rate": 1.1118012422360249e-05,
+      "loss": 0.5769,
       "step": 16720
     },
     {
       "epoch": 3.1488801054018447,
-      "grad_norm": 3.0104870796203613,
-      "learning_rate": 7.404479578392623e-06,
-      "loss": 0.5994,
+      "grad_norm": 9.697307586669922,
+      "learning_rate": 1.1106719367588933e-05,
+      "loss": 0.5116,
       "step": 16730
     },
     {
       "epoch": 3.1507622811970637,
-      "grad_norm": 23.66948699951172,
-      "learning_rate": 7.396950875211746e-06,
-      "loss": 0.6255,
+      "grad_norm": 16.02259635925293,
+      "learning_rate": 1.1095426312817617e-05,
+      "loss": 0.5188,
       "step": 16740
     },
     {
       "epoch": 3.152644456992283,
-      "grad_norm": 10.450725555419922,
-      "learning_rate": 7.389422172030868e-06,
-      "loss": 0.8733,
+      "grad_norm": 5.338298797607422,
+      "learning_rate": 1.1084133258046302e-05,
+      "loss": 0.8259,
       "step": 16750
     },
     {
       "epoch": 3.1545266327875026,
-      "grad_norm": 17.46381378173828,
-      "learning_rate": 7.381893468849991e-06,
-      "loss": 0.572,
+      "grad_norm": 18.724058151245117,
+      "learning_rate": 1.1072840203274986e-05,
+      "loss": 0.6036,
       "step": 16760
     },
     {
       "epoch": 3.1564088085827215,
-      "grad_norm": 5.500278949737549,
-      "learning_rate": 7.3743647656691146e-06,
-      "loss": 0.8854,
+      "grad_norm": 5.406846523284912,
+      "learning_rate": 1.106154714850367e-05,
+      "loss": 0.8246,
       "step": 16770
     },
     {
       "epoch": 3.158290984377941,
-      "grad_norm": 7.526846885681152,
-      "learning_rate": 7.3668360624882375e-06,
-      "loss": 0.9439,
+      "grad_norm": 8.084793090820312,
+      "learning_rate": 1.1050254093732355e-05,
+      "loss": 0.8828,
       "step": 16780
     },
     {
       "epoch": 3.16017316017316,
-      "grad_norm": 14.849970817565918,
-      "learning_rate": 7.3593073593073596e-06,
-      "loss": 0.9489,
+      "grad_norm": 24.73105812072754,
+      "learning_rate": 1.1038961038961038e-05,
+      "loss": 0.8427,
       "step": 16790
     },
     {
       "epoch": 3.1620553359683794,
-      "grad_norm": 23.20258140563965,
-      "learning_rate": 7.3517786561264825e-06,
-      "loss": 0.7435,
+      "grad_norm": 21.116586685180664,
+      "learning_rate": 1.1027667984189724e-05,
+      "loss": 0.6331,
       "step": 16800
     },
     {
       "epoch": 3.163937511763599,
-      "grad_norm": 11.882279396057129,
-      "learning_rate": 7.344249952945605e-06,
-      "loss": 0.69,
+      "grad_norm": 15.968790054321289,
+      "learning_rate": 1.1016374929418408e-05,
+      "loss": 0.562,
       "step": 16810
     },
     {
       "epoch": 3.165819687558818,
-      "grad_norm": 17.517498016357422,
-      "learning_rate": 7.336721249764729e-06,
-      "loss": 0.5923,
+      "grad_norm": 13.15872859954834,
+      "learning_rate": 1.1005081874647093e-05,
+      "loss": 0.4653,
       "step": 16820
     },
     {
       "epoch": 3.1677018633540373,
-      "grad_norm": 11.301204681396484,
-      "learning_rate": 7.329192546583852e-06,
-      "loss": 0.447,
+      "grad_norm": 13.242486953735352,
+      "learning_rate": 1.0993788819875777e-05,
+      "loss": 0.3991,
       "step": 16830
     },
     {
       "epoch": 3.1695840391492567,
-      "grad_norm": 21.43952178955078,
-      "learning_rate": 7.321663843402974e-06,
-      "loss": 0.9853,
+      "grad_norm": 12.819069862365723,
+      "learning_rate": 1.0982495765104461e-05,
+      "loss": 0.7846,
       "step": 16840
     },
     {
       "epoch": 3.1714662149444757,
-      "grad_norm": 25.572917938232422,
-      "learning_rate": 7.314135140222097e-06,
-      "loss": 0.9242,
+      "grad_norm": 13.536721229553223,
+      "learning_rate": 1.0971202710333146e-05,
+      "loss": 0.9832,
       "step": 16850
     },
     {
       "epoch": 3.173348390739695,
-      "grad_norm": 1.663003921508789,
-      "learning_rate": 7.30660643704122e-06,
-      "loss": 0.4109,
+      "grad_norm": 2.7697112560272217,
+      "learning_rate": 1.095990965556183e-05,
+      "loss": 0.4785,
       "step": 16860
     },
     {
       "epoch": 3.1752305665349145,
-      "grad_norm": 32.387046813964844,
-      "learning_rate": 7.299077733860344e-06,
-      "loss": 0.9121,
+      "grad_norm": 5.944636821746826,
+      "learning_rate": 1.0948616600790514e-05,
+      "loss": 0.6094,
       "step": 16870
     },
     {
       "epoch": 3.1771127423301335,
-      "grad_norm": 7.099456787109375,
-      "learning_rate": 7.291549030679466e-06,
-      "loss": 0.5254,
+      "grad_norm": 2.6697866916656494,
+      "learning_rate": 1.0937323546019197e-05,
+      "loss": 0.3562,
       "step": 16880
     },
     {
       "epoch": 3.178994918125353,
-      "grad_norm": 8.126106262207031,
-      "learning_rate": 7.284020327498589e-06,
-      "loss": 0.9133,
+      "grad_norm": 7.476838111877441,
+      "learning_rate": 1.0926030491247883e-05,
+      "loss": 0.8976,
       "step": 16890
     },
     {
       "epoch": 3.1808770939205724,
-      "grad_norm": 21.565263748168945,
-      "learning_rate": 7.276491624317712e-06,
-      "loss": 0.4913,
+      "grad_norm": 6.028730869293213,
+      "learning_rate": 1.0914737436476568e-05,
+      "loss": 0.4407,
       "step": 16900
     },
     {
       "epoch": 3.1827592697157914,
-      "grad_norm": 1.3021163940429688,
-      "learning_rate": 7.2689629211368355e-06,
-      "loss": 0.6226,
+      "grad_norm": 1.0638821125030518,
+      "learning_rate": 1.0903444381705252e-05,
+      "loss": 0.6061,
       "step": 16910
     },
     {
       "epoch": 3.184641445511011,
-      "grad_norm": 28.11388397216797,
-      "learning_rate": 7.2614342179559576e-06,
-      "loss": 0.6418,
+      "grad_norm": 39.25739669799805,
+      "learning_rate": 1.0892151326933935e-05,
+      "loss": 0.6129,
       "step": 16920
     },
     {
       "epoch": 3.18652362130623,
-      "grad_norm": 21.68681526184082,
-      "learning_rate": 7.2539055147750805e-06,
-      "loss": 0.6138,
+      "grad_norm": 26.864404678344727,
+      "learning_rate": 1.0880858272162619e-05,
+      "loss": 0.6046,
       "step": 16930
     },
     {
       "epoch": 3.1884057971014492,
-      "grad_norm": 2.91939640045166,
-      "learning_rate": 7.246376811594203e-06,
-      "loss": 0.6506,
+      "grad_norm": 1.348874568939209,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 0.6278,
       "step": 16940
     },
     {
       "epoch": 3.1902879728966687,
-      "grad_norm": 16.150680541992188,
-      "learning_rate": 7.238848108413326e-06,
-      "loss": 0.8231,
+      "grad_norm": 13.918986320495605,
+      "learning_rate": 1.085827216261999e-05,
+      "loss": 0.8359,
       "step": 16950
     },
     {
       "epoch": 3.1921701486918876,
-      "grad_norm": 5.776913642883301,
-      "learning_rate": 7.23131940523245e-06,
-      "loss": 0.4947,
+      "grad_norm": 12.762542724609375,
+      "learning_rate": 1.0846979107848674e-05,
+      "loss": 0.4523,
       "step": 16960
     },
     {
       "epoch": 3.194052324487107,
-      "grad_norm": 1.355134129524231,
-      "learning_rate": 7.223790702051572e-06,
-      "loss": 0.4341,
+      "grad_norm": 0.14461807906627655,
+      "learning_rate": 1.0835686053077357e-05,
+      "loss": 0.3795,
       "step": 16970
     },
     {
       "epoch": 3.1959345002823265,
-      "grad_norm": 8.270890235900879,
-      "learning_rate": 7.216261998870695e-06,
-      "loss": 0.6211,
+      "grad_norm": 12.9683256149292,
+      "learning_rate": 1.0824392998306043e-05,
+      "loss": 0.6801,
       "step": 16980
     },
     {
       "epoch": 3.1978166760775455,
-      "grad_norm": 9.533167839050293,
-      "learning_rate": 7.208733295689818e-06,
-      "loss": 0.6194,
+      "grad_norm": 18.373062133789062,
+      "learning_rate": 1.0813099943534727e-05,
+      "loss": 0.5675,
       "step": 16990
     },
     {
       "epoch": 3.199698851872765,
-      "grad_norm": 8.640917778015137,
-      "learning_rate": 7.201204592508941e-06,
-      "loss": 0.9036,
+      "grad_norm": 6.4098310470581055,
+      "learning_rate": 1.0801806888763411e-05,
+      "loss": 0.7209,
       "step": 17000
     },
     {
       "epoch": 3.2015810276679844,
-      "grad_norm": 19.37293243408203,
-      "learning_rate": 7.193675889328063e-06,
-      "loss": 0.7672,
+      "grad_norm": 1.8334890604019165,
+      "learning_rate": 1.0790513833992094e-05,
+      "loss": 0.7512,
       "step": 17010
     },
     {
       "epoch": 3.2034632034632033,
-      "grad_norm": 11.042953491210938,
-      "learning_rate": 7.186147186147187e-06,
-      "loss": 0.9198,
+      "grad_norm": 12.161665916442871,
+      "learning_rate": 1.0779220779220778e-05,
+      "loss": 0.8162,
       "step": 17020
     },
     {
       "epoch": 3.2053453792584228,
-      "grad_norm": 9.942525863647461,
-      "learning_rate": 7.17861848296631e-06,
-      "loss": 0.384,
+      "grad_norm": 14.931073188781738,
+      "learning_rate": 1.0767927724449465e-05,
+      "loss": 0.3628,
       "step": 17030
     },
     {
       "epoch": 3.207227555053642,
-      "grad_norm": 33.597999572753906,
-      "learning_rate": 7.171089779785433e-06,
-      "loss": 0.9105,
+      "grad_norm": 31.593656539916992,
+      "learning_rate": 1.0756634669678149e-05,
+      "loss": 0.8832,
       "step": 17040
     },
     {
       "epoch": 3.209109730848861,
-      "grad_norm": 14.744696617126465,
-      "learning_rate": 7.163561076604555e-06,
-      "loss": 0.574,
+      "grad_norm": 15.769742965698242,
+      "learning_rate": 1.0745341614906832e-05,
+      "loss": 0.5008,
       "step": 17050
     },
     {
       "epoch": 3.2109919066440806,
-      "grad_norm": 14.952895164489746,
-      "learning_rate": 7.1560323734236784e-06,
-      "loss": 0.662,
+      "grad_norm": 14.6060152053833,
+      "learning_rate": 1.0734048560135516e-05,
+      "loss": 0.6849,
       "step": 17060
     },
     {
       "epoch": 3.2128740824393,
-      "grad_norm": 7.853272914886475,
-      "learning_rate": 7.148503670242801e-06,
-      "loss": 0.9395,
+      "grad_norm": 9.041996002197266,
+      "learning_rate": 1.0722755505364202e-05,
+      "loss": 0.8141,
       "step": 17070
     },
     {
       "epoch": 3.214756258234519,
-      "grad_norm": 16.070457458496094,
-      "learning_rate": 7.140974967061924e-06,
-      "loss": 0.6052,
+      "grad_norm": 17.594465255737305,
+      "learning_rate": 1.0711462450592886e-05,
+      "loss": 0.5805,
       "step": 17080
     },
     {
       "epoch": 3.2166384340297385,
-      "grad_norm": 7.213963508605957,
-      "learning_rate": 7.133446263881047e-06,
-      "loss": 0.6311,
+      "grad_norm": 26.354915618896484,
+      "learning_rate": 1.070016939582157e-05,
+      "loss": 0.6139,
       "step": 17090
     },
     {
       "epoch": 3.2185206098249575,
-      "grad_norm": 26.420625686645508,
-      "learning_rate": 7.125917560700169e-06,
-      "loss": 0.7319,
+      "grad_norm": 20.24001693725586,
+      "learning_rate": 1.0688876341050254e-05,
+      "loss": 0.726,
       "step": 17100
     },
     {
       "epoch": 3.220402785620177,
-      "grad_norm": 12.303365707397461,
-      "learning_rate": 7.118388857519293e-06,
-      "loss": 0.8293,
+      "grad_norm": 9.388453483581543,
+      "learning_rate": 1.0677583286278938e-05,
+      "loss": 0.837,
       "step": 17110
     },
     {
       "epoch": 3.2222849614153963,
-      "grad_norm": 19.405977249145508,
-      "learning_rate": 7.110860154338416e-06,
-      "loss": 0.6416,
+      "grad_norm": 33.33871841430664,
+      "learning_rate": 1.0666290231507624e-05,
+      "loss": 0.6729,
       "step": 17120
     },
     {
       "epoch": 3.2241671372106153,
-      "grad_norm": 20.179763793945312,
-      "learning_rate": 7.103331451157539e-06,
-      "loss": 0.635,
+      "grad_norm": 14.027726173400879,
+      "learning_rate": 1.0654997176736308e-05,
+      "loss": 0.6477,
       "step": 17130
     },
     {
       "epoch": 3.2260493130058348,
-      "grad_norm": 8.78466510772705,
-      "learning_rate": 7.095802747976661e-06,
-      "loss": 0.5423,
+      "grad_norm": 10.19504451751709,
+      "learning_rate": 1.0643704121964991e-05,
+      "loss": 0.3972,
       "step": 17140
     },
     {
       "epoch": 3.227931488801054,
-      "grad_norm": 39.822975158691406,
-      "learning_rate": 7.088274044795784e-06,
-      "loss": 0.6596,
+      "grad_norm": 20.70015525817871,
+      "learning_rate": 1.0632411067193675e-05,
+      "loss": 0.6082,
       "step": 17150
     },
     {
       "epoch": 3.229813664596273,
-      "grad_norm": 15.491219520568848,
-      "learning_rate": 7.080745341614908e-06,
-      "loss": 0.8531,
+      "grad_norm": 6.828171730041504,
+      "learning_rate": 1.062111801242236e-05,
+      "loss": 0.7273,
       "step": 17160
     },
     {
       "epoch": 3.2316958403914926,
-      "grad_norm": 14.816864967346191,
-      "learning_rate": 7.073216638434031e-06,
-      "loss": 0.8523,
+      "grad_norm": 13.635649681091309,
+      "learning_rate": 1.0609824957651046e-05,
+      "loss": 0.6924,
       "step": 17170
     },
     {
       "epoch": 3.233578016186712,
-      "grad_norm": 2.6879067420959473,
-      "learning_rate": 7.0656879352531535e-06,
-      "loss": 0.4871,
+      "grad_norm": 5.508321285247803,
+      "learning_rate": 1.059853190287973e-05,
+      "loss": 0.5075,
       "step": 17180
     },
     {
       "epoch": 3.235460191981931,
-      "grad_norm": 15.283522605895996,
-      "learning_rate": 7.058159232072276e-06,
-      "loss": 0.7526,
+      "grad_norm": 20.685861587524414,
+      "learning_rate": 1.0587238848108413e-05,
+      "loss": 0.5851,
       "step": 17190
     },
     {
       "epoch": 3.2373423677771505,
-      "grad_norm": 14.024410247802734,
-      "learning_rate": 7.050630528891399e-06,
-      "loss": 0.5458,
+      "grad_norm": 14.502724647521973,
+      "learning_rate": 1.0575945793337097e-05,
+      "loss": 0.4921,
       "step": 17200
     },
     {
       "epoch": 3.2392245435723694,
-      "grad_norm": 17.729820251464844,
-      "learning_rate": 7.043101825710522e-06,
-      "loss": 0.7159,
+      "grad_norm": 8.47658634185791,
+      "learning_rate": 1.0564652738565783e-05,
+      "loss": 0.6247,
       "step": 17210
     },
     {
       "epoch": 3.241106719367589,
-      "grad_norm": 11.570318222045898,
-      "learning_rate": 7.035573122529645e-06,
-      "loss": 0.8422,
+      "grad_norm": 10.943485260009766,
+      "learning_rate": 1.0553359683794468e-05,
+      "loss": 0.8117,
       "step": 17220
     },
     {
       "epoch": 3.2429888951628083,
-      "grad_norm": 29.138668060302734,
-      "learning_rate": 7.028044419348767e-06,
-      "loss": 0.7466,
+      "grad_norm": 10.894013404846191,
+      "learning_rate": 1.054206662902315e-05,
+      "loss": 0.494,
       "step": 17230
     },
     {
       "epoch": 3.2448710709580273,
-      "grad_norm": 27.035675048828125,
-      "learning_rate": 7.02051571616789e-06,
-      "loss": 0.5832,
+      "grad_norm": 12.969403266906738,
+      "learning_rate": 1.0530773574251835e-05,
+      "loss": 0.4713,
       "step": 17240
     },
     {
       "epoch": 3.2467532467532467,
-      "grad_norm": 12.801972389221191,
-      "learning_rate": 7.012987012987014e-06,
-      "loss": 0.7841,
+      "grad_norm": 13.456480979919434,
+      "learning_rate": 1.051948051948052e-05,
+      "loss": 0.6888,
       "step": 17250
     },
     {
       "epoch": 3.248635422548466,
-      "grad_norm": 16.305395126342773,
-      "learning_rate": 7.005458309806137e-06,
-      "loss": 0.7556,
+      "grad_norm": 12.490468978881836,
+      "learning_rate": 1.0508187464709205e-05,
+      "loss": 0.6876,
       "step": 17260
     },
     {
       "epoch": 3.250517598343685,
-      "grad_norm": 23.742103576660156,
-      "learning_rate": 6.997929606625259e-06,
-      "loss": 0.6456,
+      "grad_norm": 13.233989715576172,
+      "learning_rate": 1.0496894409937888e-05,
+      "loss": 0.5344,
       "step": 17270
     },
     {
       "epoch": 3.2523997741389046,
-      "grad_norm": 14.787322044372559,
-      "learning_rate": 6.990400903444382e-06,
-      "loss": 0.7138,
+      "grad_norm": 26.581886291503906,
+      "learning_rate": 1.0485601355166572e-05,
+      "loss": 0.6548,
       "step": 17280
     },
     {
       "epoch": 3.254281949934124,
-      "grad_norm": 15.448890686035156,
-      "learning_rate": 6.982872200263505e-06,
-      "loss": 0.4583,
+      "grad_norm": 5.6854777336120605,
+      "learning_rate": 1.0474308300395257e-05,
+      "loss": 0.3306,
       "step": 17290
     },
     {
       "epoch": 3.256164125729343,
-      "grad_norm": 17.958166122436523,
-      "learning_rate": 6.9753434970826286e-06,
-      "loss": 0.5352,
+      "grad_norm": 15.864883422851562,
+      "learning_rate": 1.0463015245623941e-05,
+      "loss": 0.3117,
       "step": 17300
     },
     {
       "epoch": 3.2580463015245624,
-      "grad_norm": 29.221757888793945,
-      "learning_rate": 6.9678147939017515e-06,
-      "loss": 0.7444,
+      "grad_norm": 1.5253409147262573,
+      "learning_rate": 1.0451722190852627e-05,
+      "loss": 0.6778,
       "step": 17310
     },
     {
       "epoch": 3.259928477319782,
-      "grad_norm": 5.633476734161377,
-      "learning_rate": 6.9602860907208736e-06,
-      "loss": 0.4572,
+      "grad_norm": 13.313026428222656,
+      "learning_rate": 1.044042913608131e-05,
+      "loss": 0.4523,
       "step": 17320
     },
     {
       "epoch": 3.261810653115001,
-      "grad_norm": 11.22506046295166,
-      "learning_rate": 6.9527573875399965e-06,
-      "loss": 0.5099,
+      "grad_norm": 6.4470343589782715,
+      "learning_rate": 1.0429136081309994e-05,
+      "loss": 0.4609,
       "step": 17330
     },
     {
       "epoch": 3.2636928289102203,
-      "grad_norm": 10.039155960083008,
-      "learning_rate": 6.94522868435912e-06,
-      "loss": 0.5323,
+      "grad_norm": 6.336509704589844,
+      "learning_rate": 1.0417843026538679e-05,
+      "loss": 0.398,
       "step": 17340
     },
     {
       "epoch": 3.2655750047054397,
-      "grad_norm": 9.84754467010498,
-      "learning_rate": 6.937699981178243e-06,
-      "loss": 0.7944,
+      "grad_norm": 17.98333740234375,
+      "learning_rate": 1.0406549971767365e-05,
+      "loss": 0.8517,
       "step": 17350
     },
     {
       "epoch": 3.2674571805006587,
-      "grad_norm": 34.082881927490234,
-      "learning_rate": 6.930171277997365e-06,
-      "loss": 0.7083,
+      "grad_norm": 22.246747970581055,
+      "learning_rate": 1.0395256916996047e-05,
+      "loss": 0.5689,
       "step": 17360
     },
     {
       "epoch": 3.269339356295878,
-      "grad_norm": 16.67072105407715,
-      "learning_rate": 6.922642574816488e-06,
-      "loss": 0.8544,
+      "grad_norm": 16.68685531616211,
+      "learning_rate": 1.0383963862224732e-05,
+      "loss": 0.8071,
       "step": 17370
     },
     {
       "epoch": 3.271221532091097,
-      "grad_norm": 15.253617286682129,
-      "learning_rate": 6.915113871635611e-06,
-      "loss": 0.6721,
+      "grad_norm": 25.194631576538086,
+      "learning_rate": 1.0372670807453416e-05,
+      "loss": 0.6331,
       "step": 17380
     },
     {
       "epoch": 3.2731037078863165,
-      "grad_norm": 12.421513557434082,
-      "learning_rate": 6.907585168454735e-06,
-      "loss": 0.55,
+      "grad_norm": 10.572707176208496,
+      "learning_rate": 1.03613777526821e-05,
+      "loss": 0.4869,
       "step": 17390
     },
     {
       "epoch": 3.274985883681536,
-      "grad_norm": 7.298317909240723,
-      "learning_rate": 6.900056465273857e-06,
-      "loss": 0.6685,
+      "grad_norm": 7.161264896392822,
+      "learning_rate": 1.0350084697910785e-05,
+      "loss": 0.4585,
       "step": 17400
     },
     {
       "epoch": 3.276868059476755,
-      "grad_norm": 10.030412673950195,
-      "learning_rate": 6.89252776209298e-06,
-      "loss": 0.9101,
+      "grad_norm": 13.183558464050293,
+      "learning_rate": 1.033879164313947e-05,
+      "loss": 0.8198,
       "step": 17410
     },
     {
       "epoch": 3.2787502352719744,
-      "grad_norm": 12.07267951965332,
-      "learning_rate": 6.884999058912103e-06,
-      "loss": 0.6002,
+      "grad_norm": 21.05937385559082,
+      "learning_rate": 1.0327498588368154e-05,
+      "loss": 0.4036,
       "step": 17420
     },
     {
       "epoch": 3.280632411067194,
-      "grad_norm": 9.45308780670166,
-      "learning_rate": 6.8774703557312265e-06,
-      "loss": 0.6294,
+      "grad_norm": 5.843810558319092,
+      "learning_rate": 1.0316205533596838e-05,
+      "loss": 0.5591,
       "step": 17430
     },
     {
       "epoch": 3.282514586862413,
-      "grad_norm": 3.554718255996704,
-      "learning_rate": 6.8699416525503495e-06,
-      "loss": 0.5614,
+      "grad_norm": 3.102691888809204,
+      "learning_rate": 1.0304912478825523e-05,
+      "loss": 0.511,
       "step": 17440
     },
     {
       "epoch": 3.2843967626576323,
-      "grad_norm": 29.615537643432617,
-      "learning_rate": 6.8624129493694715e-06,
-      "loss": 0.6053,
+      "grad_norm": 22.08820915222168,
+      "learning_rate": 1.0293619424054207e-05,
+      "loss": 0.5755,
       "step": 17450
     },
     {
       "epoch": 3.2862789384528517,
-      "grad_norm": 15.930276870727539,
-      "learning_rate": 6.8548842461885945e-06,
-      "loss": 0.3862,
+      "grad_norm": 6.78153657913208,
+      "learning_rate": 1.0282326369282891e-05,
+      "loss": 0.3316,
       "step": 17460
     },
     {
       "epoch": 3.2881611142480707,
-      "grad_norm": 25.522640228271484,
-      "learning_rate": 6.847355543007717e-06,
-      "loss": 0.5614,
+      "grad_norm": 24.446062088012695,
+      "learning_rate": 1.0271033314511576e-05,
+      "loss": 0.4758,
       "step": 17470
     },
     {
       "epoch": 3.29004329004329,
-      "grad_norm": 36.311302185058594,
-      "learning_rate": 6.839826839826841e-06,
-      "loss": 0.674,
+      "grad_norm": 23.952836990356445,
+      "learning_rate": 1.025974025974026e-05,
+      "loss": 0.622,
       "step": 17480
     },
     {
       "epoch": 3.291925465838509,
-      "grad_norm": 19.43730354309082,
-      "learning_rate": 6.832298136645963e-06,
-      "loss": 0.5512,
+      "grad_norm": 39.4639892578125,
+      "learning_rate": 1.0248447204968944e-05,
+      "loss": 0.5434,
       "step": 17490
     },
     {
       "epoch": 3.2938076416337285,
-      "grad_norm": 5.826242446899414,
-      "learning_rate": 6.824769433465086e-06,
-      "loss": 0.9262,
+      "grad_norm": 17.225452423095703,
+      "learning_rate": 1.0237154150197629e-05,
+      "loss": 0.9825,
       "step": 17500
     },
     {
       "epoch": 3.295689817428948,
-      "grad_norm": 11.90711498260498,
-      "learning_rate": 6.817240730284209e-06,
-      "loss": 0.7307,
+      "grad_norm": 9.208147048950195,
+      "learning_rate": 1.0225861095426313e-05,
+      "loss": 0.6339,
       "step": 17510
     },
     {
       "epoch": 3.2975719932241674,
-      "grad_norm": 35.005157470703125,
-      "learning_rate": 6.809712027103332e-06,
-      "loss": 0.752,
+      "grad_norm": 30.452302932739258,
+      "learning_rate": 1.0214568040654998e-05,
+      "loss": 0.6128,
       "step": 17520
     },
     {
       "epoch": 3.2994541690193864,
-      "grad_norm": 12.222931861877441,
-      "learning_rate": 6.802183323922454e-06,
-      "loss": 0.6441,
+      "grad_norm": 32.60814666748047,
+      "learning_rate": 1.020327498588368e-05,
+      "loss": 0.5782,
       "step": 17530
     },
     {
       "epoch": 3.301336344814606,
-      "grad_norm": 14.466747283935547,
-      "learning_rate": 6.794654620741578e-06,
-      "loss": 0.6763,
+      "grad_norm": 16.545825958251953,
+      "learning_rate": 1.0191981931112366e-05,
+      "loss": 0.5999,
       "step": 17540
     },
     {
       "epoch": 3.303218520609825,
-      "grad_norm": 17.527795791625977,
-      "learning_rate": 6.787125917560701e-06,
-      "loss": 0.4761,
+      "grad_norm": 15.124717712402344,
+      "learning_rate": 1.018068887634105e-05,
+      "loss": 0.4035,
       "step": 17550
     },
     {
       "epoch": 3.3051006964050442,
-      "grad_norm": 11.280905723571777,
-      "learning_rate": 6.779597214379824e-06,
-      "loss": 0.939,
+      "grad_norm": 6.470951080322266,
+      "learning_rate": 1.0169395821569735e-05,
+      "loss": 0.7927,
       "step": 17560
     },
     {
       "epoch": 3.3069828722002637,
-      "grad_norm": 19.548994064331055,
-      "learning_rate": 6.7720685111989474e-06,
-      "loss": 0.747,
+      "grad_norm": 16.971155166625977,
+      "learning_rate": 1.015810276679842e-05,
+      "loss": 0.6804,
       "step": 17570
     },
     {
       "epoch": 3.3088650479954826,
-      "grad_norm": 28.56268310546875,
-      "learning_rate": 6.764539808018069e-06,
-      "loss": 0.7412,
+      "grad_norm": 29.655359268188477,
+      "learning_rate": 1.0146809712027104e-05,
+      "loss": 0.7321,
       "step": 17580
     },
     {
       "epoch": 3.310747223790702,
-      "grad_norm": 23.072845458984375,
-      "learning_rate": 6.7570111048371924e-06,
-      "loss": 0.8036,
+      "grad_norm": 28.08059310913086,
+      "learning_rate": 1.0135516657255788e-05,
+      "loss": 0.7635,
       "step": 17590
     },
     {
       "epoch": 3.3126293995859215,
-      "grad_norm": 1.0671284198760986,
-      "learning_rate": 6.749482401656315e-06,
-      "loss": 0.5617,
+      "grad_norm": 0.4805198013782501,
+      "learning_rate": 1.0124223602484473e-05,
+      "loss": 0.4791,
       "step": 17600
     },
     {
       "epoch": 3.3145115753811405,
-      "grad_norm": 17.233400344848633,
-      "learning_rate": 6.741953698475438e-06,
-      "loss": 0.6841,
+      "grad_norm": 30.67386245727539,
+      "learning_rate": 1.0112930547713157e-05,
+      "loss": 0.6166,
       "step": 17610
     },
     {
       "epoch": 3.31639375117636,
-      "grad_norm": 11.159910202026367,
-      "learning_rate": 6.73442499529456e-06,
-      "loss": 0.9642,
+      "grad_norm": 30.00206756591797,
+      "learning_rate": 1.010163749294184e-05,
+      "loss": 0.9969,
       "step": 17620
     },
     {
       "epoch": 3.3182759269715794,
-      "grad_norm": 7.40802001953125,
-      "learning_rate": 6.726896292113684e-06,
-      "loss": 0.5016,
+      "grad_norm": 11.875704765319824,
+      "learning_rate": 1.0090344438170526e-05,
+      "loss": 0.4861,
       "step": 17630
     },
     {
       "epoch": 3.3201581027667983,
-      "grad_norm": 19.014680862426758,
-      "learning_rate": 6.719367588932807e-06,
-      "loss": 0.7707,
+      "grad_norm": 59.30400848388672,
+      "learning_rate": 1.007905138339921e-05,
+      "loss": 0.6634,
       "step": 17640
     },
     {
       "epoch": 3.322040278562018,
-      "grad_norm": 13.653407096862793,
-      "learning_rate": 6.71183888575193e-06,
-      "loss": 0.3072,
+      "grad_norm": 7.319840908050537,
+      "learning_rate": 1.0067758328627895e-05,
+      "loss": 0.1843,
       "step": 17650
     },
     {
       "epoch": 3.3239224543572368,
-      "grad_norm": 2.027129650115967,
-      "learning_rate": 6.704310182571052e-06,
-      "loss": 0.8146,
+      "grad_norm": 1.659640908241272,
+      "learning_rate": 1.0056465273856577e-05,
+      "loss": 0.8501,
       "step": 17660
     },
     {
       "epoch": 3.325804630152456,
-      "grad_norm": 12.079022407531738,
-      "learning_rate": 6.696781479390175e-06,
-      "loss": 0.6395,
+      "grad_norm": 15.587262153625488,
+      "learning_rate": 1.0045172219085262e-05,
+      "loss": 0.5065,
       "step": 17670
     },
     {
       "epoch": 3.3276868059476756,
-      "grad_norm": 40.99082565307617,
-      "learning_rate": 6.689252776209299e-06,
-      "loss": 0.8666,
+      "grad_norm": 25.570293426513672,
+      "learning_rate": 1.0033879164313948e-05,
+      "loss": 0.8362,
       "step": 17680
     },
     {
       "epoch": 3.3295689817428946,
-      "grad_norm": 10.816984176635742,
-      "learning_rate": 6.681724073028422e-06,
-      "loss": 0.7107,
+      "grad_norm": 5.94968318939209,
+      "learning_rate": 1.0022586109542632e-05,
+      "loss": 0.7938,
       "step": 17690
     },
     {
       "epoch": 3.331451157538114,
-      "grad_norm": 14.31318473815918,
-      "learning_rate": 6.6741953698475446e-06,
-      "loss": 0.8113,
+      "grad_norm": 5.780099868774414,
+      "learning_rate": 1.0011293054771316e-05,
+      "loss": 0.6672,
       "step": 17700
     },
     {
       "epoch": 3.3333333333333335,
-      "grad_norm": 37.56098175048828,
-      "learning_rate": 6.666666666666667e-06,
-      "loss": 0.7322,
+      "grad_norm": 32.476585388183594,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 0.5259,
       "step": 17710
     },
     {
       "epoch": 3.3352155091285525,
-      "grad_norm": 7.864965915679932,
-      "learning_rate": 6.65913796348579e-06,
-      "loss": 0.7357,
+      "grad_norm": 6.785778999328613,
+      "learning_rate": 9.988706945228685e-06,
+      "loss": 0.719,
       "step": 17720
     },
     {
       "epoch": 3.337097684923772,
-      "grad_norm": 20.20433807373047,
-      "learning_rate": 6.651609260304913e-06,
-      "loss": 0.8787,
+      "grad_norm": 16.405250549316406,
+      "learning_rate": 9.97741389045737e-06,
+      "loss": 0.8035,
       "step": 17730
     },
     {
       "epoch": 3.3389798607189913,
-      "grad_norm": 15.400228500366211,
-      "learning_rate": 6.644080557124036e-06,
-      "loss": 0.8685,
+      "grad_norm": 8.930052757263184,
+      "learning_rate": 9.966120835686054e-06,
+      "loss": 0.7513,
       "step": 17740
     },
     {
       "epoch": 3.3408620365142103,
-      "grad_norm": 1.1104925870895386,
-      "learning_rate": 6.636551853943158e-06,
-      "loss": 0.5544,
+      "grad_norm": 0.5434146523475647,
+      "learning_rate": 9.954827780914737e-06,
+      "loss": 0.5323,
       "step": 17750
     },
     {
       "epoch": 3.3427442123094298,
-      "grad_norm": 16.083879470825195,
-      "learning_rate": 6.629023150762281e-06,
-      "loss": 0.6363,
+      "grad_norm": 12.213115692138672,
+      "learning_rate": 9.943534726143421e-06,
+      "loss": 0.5251,
       "step": 17760
     },
     {
       "epoch": 3.3446263881046487,
-      "grad_norm": 18.40869903564453,
-      "learning_rate": 6.621494447581405e-06,
-      "loss": 0.9739,
+      "grad_norm": 18.9610595703125,
+      "learning_rate": 9.932241671372107e-06,
+      "loss": 0.833,
       "step": 17770
     },
     {
       "epoch": 3.346508563899868,
-      "grad_norm": 4.225210189819336,
-      "learning_rate": 6.613965744400528e-06,
-      "loss": 0.8647,
+      "grad_norm": 8.55979061126709,
+      "learning_rate": 9.920948616600791e-06,
+      "loss": 0.9362,
       "step": 17780
     },
     {
       "epoch": 3.3483907396950876,
-      "grad_norm": 3.2670443058013916,
-      "learning_rate": 6.60643704121965e-06,
-      "loss": 0.5671,
+      "grad_norm": 1.9761569499969482,
+      "learning_rate": 9.909655561829474e-06,
+      "loss": 0.5021,
       "step": 17790
     },
     {
       "epoch": 3.350272915490307,
-      "grad_norm": 23.90607261657715,
-      "learning_rate": 6.598908338038773e-06,
-      "loss": 0.7112,
+      "grad_norm": 16.7060546875,
+      "learning_rate": 9.898362507058159e-06,
+      "loss": 0.5859,
       "step": 17800
     },
     {
       "epoch": 3.352155091285526,
-      "grad_norm": 12.343167304992676,
-      "learning_rate": 6.591379634857896e-06,
-      "loss": 0.7628,
+      "grad_norm": 16.749752044677734,
+      "learning_rate": 9.887069452286843e-06,
+      "loss": 0.4982,
       "step": 17810
     },
     {
       "epoch": 3.3540372670807455,
-      "grad_norm": 5.21704626083374,
-      "learning_rate": 6.58385093167702e-06,
-      "loss": 0.5627,
+      "grad_norm": 3.789137601852417,
+      "learning_rate": 9.875776397515529e-06,
+      "loss": 0.4616,
       "step": 17820
     },
     {
       "epoch": 3.3559194428759644,
-      "grad_norm": 15.107742309570312,
-      "learning_rate": 6.5763222284961426e-06,
-      "loss": 0.6367,
+      "grad_norm": 13.295493125915527,
+      "learning_rate": 9.864483342744213e-06,
+      "loss": 0.6548,
       "step": 17830
     },
     {
       "epoch": 3.357801618671184,
-      "grad_norm": 13.048922538757324,
-      "learning_rate": 6.568793525315265e-06,
-      "loss": 0.6175,
+      "grad_norm": 25.58221435546875,
+      "learning_rate": 9.853190287972896e-06,
+      "loss": 0.7388,
       "step": 17840
     },
     {
       "epoch": 3.3596837944664033,
-      "grad_norm": 18.35613441467285,
-      "learning_rate": 6.5612648221343875e-06,
-      "loss": 0.629,
+      "grad_norm": 29.051992416381836,
+      "learning_rate": 9.84189723320158e-06,
+      "loss": 0.5679,
       "step": 17850
     },
     {
       "epoch": 3.3615659702616223,
-      "grad_norm": 15.142684936523438,
-      "learning_rate": 6.553736118953511e-06,
-      "loss": 0.8152,
+      "grad_norm": 11.573369026184082,
+      "learning_rate": 9.830604178430267e-06,
+      "loss": 0.7047,
       "step": 17860
     },
     {
       "epoch": 3.3634481460568417,
-      "grad_norm": 27.841224670410156,
-      "learning_rate": 6.546207415772634e-06,
-      "loss": 0.7843,
+      "grad_norm": 16.204544067382812,
+      "learning_rate": 9.819311123658951e-06,
+      "loss": 0.5335,
       "step": 17870
     },
     {
       "epoch": 3.365330321852061,
-      "grad_norm": 7.560338020324707,
-      "learning_rate": 6.538678712591756e-06,
-      "loss": 0.5603,
+      "grad_norm": 18.992883682250977,
+      "learning_rate": 9.808018068887634e-06,
+      "loss": 0.5,
       "step": 17880
     },
     {
       "epoch": 3.36721249764728,
-      "grad_norm": 21.53891944885254,
-      "learning_rate": 6.531150009410879e-06,
-      "loss": 0.615,
+      "grad_norm": 17.08180809020996,
+      "learning_rate": 9.796725014116318e-06,
+      "loss": 0.6355,
       "step": 17890
     },
     {
       "epoch": 3.3690946734424996,
-      "grad_norm": 11.022046089172363,
-      "learning_rate": 6.523621306230002e-06,
-      "loss": 0.5921,
+      "grad_norm": 12.415989875793457,
+      "learning_rate": 9.785431959345002e-06,
+      "loss": 0.6227,
       "step": 17900
     },
     {
       "epoch": 3.370976849237719,
-      "grad_norm": 11.184412956237793,
-      "learning_rate": 6.516092603049126e-06,
-      "loss": 0.7016,
+      "grad_norm": 23.010984420776367,
+      "learning_rate": 9.774138904573688e-06,
+      "loss": 0.6095,
       "step": 17910
     },
     {
       "epoch": 3.372859025032938,
-      "grad_norm": 19.147296905517578,
-      "learning_rate": 6.508563899868249e-06,
-      "loss": 0.87,
+      "grad_norm": 20.306901931762695,
+      "learning_rate": 9.762845849802373e-06,
+      "loss": 0.7475,
       "step": 17920
     },
     {
       "epoch": 3.3747412008281574,
-      "grad_norm": 8.234039306640625,
-      "learning_rate": 6.501035196687371e-06,
-      "loss": 0.8574,
+      "grad_norm": 7.907451629638672,
+      "learning_rate": 9.751552795031056e-06,
+      "loss": 0.7657,
       "step": 17930
     },
     {
       "epoch": 3.3766233766233764,
-      "grad_norm": 19.271259307861328,
-      "learning_rate": 6.493506493506494e-06,
-      "loss": 0.9815,
+      "grad_norm": 10.358199119567871,
+      "learning_rate": 9.74025974025974e-06,
+      "loss": 0.8862,
       "step": 17940
     },
     {
       "epoch": 3.378505552418596,
-      "grad_norm": 9.072455406188965,
-      "learning_rate": 6.485977790325617e-06,
-      "loss": 0.5195,
+      "grad_norm": 7.511529445648193,
+      "learning_rate": 9.728966685488426e-06,
+      "loss": 0.4816,
       "step": 17950
     },
     {
       "epoch": 3.3803877282138153,
-      "grad_norm": 9.557124137878418,
-      "learning_rate": 6.4784490871447405e-06,
-      "loss": 0.6915,
+      "grad_norm": 11.36762523651123,
+      "learning_rate": 9.71767363071711e-06,
+      "loss": 0.6414,
       "step": 17960
     },
     {
       "epoch": 3.3822699040090343,
-      "grad_norm": 2.8738791942596436,
-      "learning_rate": 6.470920383963863e-06,
-      "loss": 0.396,
+      "grad_norm": 4.441726207733154,
+      "learning_rate": 9.706380575945793e-06,
+      "loss": 0.3488,
       "step": 17970
     },
     {
       "epoch": 3.3841520798042537,
-      "grad_norm": 12.238557815551758,
-      "learning_rate": 6.4633916807829855e-06,
-      "loss": 0.7431,
+      "grad_norm": 19.867027282714844,
+      "learning_rate": 9.695087521174477e-06,
+      "loss": 0.7145,
       "step": 17980
     },
     {
       "epoch": 3.386034255599473,
-      "grad_norm": 14.482816696166992,
-      "learning_rate": 6.4558629776021084e-06,
-      "loss": 0.6753,
+      "grad_norm": 10.9057035446167,
+      "learning_rate": 9.683794466403162e-06,
+      "loss": 0.6342,
       "step": 17990
     },
     {
       "epoch": 3.387916431394692,
-      "grad_norm": 7.765755653381348,
-      "learning_rate": 6.448334274421232e-06,
-      "loss": 0.6066,
+      "grad_norm": 8.741947174072266,
+      "learning_rate": 9.672501411631848e-06,
+      "loss": 0.5844,
       "step": 18000
     },
     {
       "epoch": 3.3897986071899116,
-      "grad_norm": 30.565616607666016,
-      "learning_rate": 6.440805571240354e-06,
-      "loss": 0.6136,
+      "grad_norm": 25.616764068603516,
+      "learning_rate": 9.66120835686053e-06,
+      "loss": 0.6007,
       "step": 18010
     },
     {
       "epoch": 3.391680782985131,
-      "grad_norm": 13.92880916595459,
-      "learning_rate": 6.433276868059477e-06,
-      "loss": 0.6848,
+      "grad_norm": 8.889225959777832,
+      "learning_rate": 9.649915302089215e-06,
+      "loss": 0.4933,
       "step": 18020
     },
     {
       "epoch": 3.39356295878035,
-      "grad_norm": 9.684536933898926,
-      "learning_rate": 6.4257481648786e-06,
-      "loss": 0.6168,
+      "grad_norm": 9.683721542358398,
+      "learning_rate": 9.6386222473179e-06,
+      "loss": 0.5452,
       "step": 18030
     },
     {
       "epoch": 3.3954451345755694,
-      "grad_norm": 3.2142350673675537,
-      "learning_rate": 6.418219461697723e-06,
-      "loss": 0.6285,
+      "grad_norm": 5.871509075164795,
+      "learning_rate": 9.627329192546584e-06,
+      "loss": 0.5289,
       "step": 18040
     },
     {
       "epoch": 3.3973273103707884,
-      "grad_norm": 6.0369977951049805,
-      "learning_rate": 6.410690758516847e-06,
-      "loss": 0.6715,
+      "grad_norm": 12.041633605957031,
+      "learning_rate": 9.61603613777527e-06,
+      "loss": 0.6335,
       "step": 18050
     },
     {
       "epoch": 3.399209486166008,
-      "grad_norm": 16.24129295349121,
-      "learning_rate": 6.403162055335969e-06,
-      "loss": 0.8986,
+      "grad_norm": 17.63007926940918,
+      "learning_rate": 9.604743083003952e-06,
+      "loss": 0.8369,
       "step": 18060
     },
     {
       "epoch": 3.4010916619612273,
-      "grad_norm": 26.860872268676758,
-      "learning_rate": 6.395633352155092e-06,
-      "loss": 0.6195,
+      "grad_norm": 17.726680755615234,
+      "learning_rate": 9.593450028232637e-06,
+      "loss": 0.5378,
       "step": 18070
     },
     {
       "epoch": 3.4029738377564467,
-      "grad_norm": 11.599557876586914,
-      "learning_rate": 6.388104648974215e-06,
-      "loss": 0.6924,
+      "grad_norm": 7.5816330909729,
+      "learning_rate": 9.582156973461321e-06,
+      "loss": 0.5959,
       "step": 18080
     },
     {
       "epoch": 3.4048560135516657,
-      "grad_norm": 21.338090896606445,
-      "learning_rate": 6.380575945793338e-06,
-      "loss": 0.6343,
+      "grad_norm": 12.254039764404297,
+      "learning_rate": 9.570863918690007e-06,
+      "loss": 0.5402,
       "step": 18090
     },
     {
       "epoch": 3.406738189346885,
-      "grad_norm": 17.614233016967773,
-      "learning_rate": 6.37304724261246e-06,
-      "loss": 0.5183,
+      "grad_norm": 20.039281845092773,
+      "learning_rate": 9.55957086391869e-06,
+      "loss": 0.4543,
       "step": 18100
     },
     {
       "epoch": 3.408620365142104,
-      "grad_norm": 58.38087844848633,
-      "learning_rate": 6.3655185394315835e-06,
-      "loss": 0.6562,
+      "grad_norm": 31.039440155029297,
+      "learning_rate": 9.548277809147374e-06,
+      "loss": 0.7242,
       "step": 18110
     },
     {
       "epoch": 3.4105025409373235,
-      "grad_norm": 21.725601196289062,
-      "learning_rate": 6.3579898362507064e-06,
-      "loss": 0.5947,
+      "grad_norm": 21.17696189880371,
+      "learning_rate": 9.536984754376059e-06,
+      "loss": 0.5722,
       "step": 18120
     },
     {
       "epoch": 3.412384716732543,
-      "grad_norm": 31.378341674804688,
-      "learning_rate": 6.350461133069829e-06,
-      "loss": 0.6798,
+      "grad_norm": 37.80180740356445,
+      "learning_rate": 9.525691699604743e-06,
+      "loss": 0.6266,
       "step": 18130
     },
     {
       "epoch": 3.414266892527762,
-      "grad_norm": 0.62470942735672,
-      "learning_rate": 6.342932429888951e-06,
-      "loss": 0.669,
+      "grad_norm": 0.5937038064002991,
+      "learning_rate": 9.514398644833428e-06,
+      "loss": 0.6332,
       "step": 18140
     },
     {
       "epoch": 3.4161490683229814,
-      "grad_norm": 6.63415002822876,
-      "learning_rate": 6.335403726708075e-06,
-      "loss": 0.6178,
+      "grad_norm": 18.84284782409668,
+      "learning_rate": 9.503105590062112e-06,
+      "loss": 0.5928,
       "step": 18150
     },
     {
       "epoch": 3.418031244118201,
-      "grad_norm": 5.154765605926514,
-      "learning_rate": 6.327875023527198e-06,
-      "loss": 0.666,
+      "grad_norm": 3.241466522216797,
+      "learning_rate": 9.491812535290796e-06,
+      "loss": 0.5802,
       "step": 18160
     },
     {
       "epoch": 3.41991341991342,
-      "grad_norm": 9.039029121398926,
-      "learning_rate": 6.320346320346321e-06,
-      "loss": 0.4487,
+      "grad_norm": 4.864845275878906,
+      "learning_rate": 9.48051948051948e-06,
+      "loss": 0.4499,
       "step": 18170
     },
     {
       "epoch": 3.4217955957086392,
-      "grad_norm": 4.979629039764404,
-      "learning_rate": 6.312817617165444e-06,
-      "loss": 0.5112,
+      "grad_norm": 10.21737289428711,
+      "learning_rate": 9.469226425748165e-06,
+      "loss": 0.466,
       "step": 18180
     },
     {
       "epoch": 3.4236777715038587,
-      "grad_norm": 14.24013614654541,
-      "learning_rate": 6.305288913984566e-06,
-      "loss": 0.4772,
+      "grad_norm": 21.10309410095215,
+      "learning_rate": 9.45793337097685e-06,
+      "loss": 0.4943,
       "step": 18190
     },
     {
       "epoch": 3.4255599472990776,
-      "grad_norm": 20.53621482849121,
-      "learning_rate": 6.29776021080369e-06,
-      "loss": 0.4101,
+      "grad_norm": 20.39488983154297,
+      "learning_rate": 9.446640316205534e-06,
+      "loss": 0.4178,
       "step": 18200
     },
     {
       "epoch": 3.427442123094297,
-      "grad_norm": 10.990429878234863,
-      "learning_rate": 6.290231507622813e-06,
-      "loss": 0.5397,
+      "grad_norm": 10.568327903747559,
+      "learning_rate": 9.435347261434218e-06,
+      "loss": 0.5787,
       "step": 18210
     },
     {
       "epoch": 3.429324298889516,
-      "grad_norm": 11.857673645019531,
-      "learning_rate": 6.282702804441936e-06,
-      "loss": 0.6357,
+      "grad_norm": 15.859696388244629,
+      "learning_rate": 9.424054206662903e-06,
+      "loss": 0.5257,
       "step": 18220
     },
     {
       "epoch": 3.4312064746847355,
-      "grad_norm": 6.039271354675293,
-      "learning_rate": 6.275174101261058e-06,
-      "loss": 0.5146,
+      "grad_norm": 3.5136094093322754,
+      "learning_rate": 9.412761151891587e-06,
+      "loss": 0.4556,
       "step": 18230
     },
     {
       "epoch": 3.433088650479955,
-      "grad_norm": 14.336277961730957,
-      "learning_rate": 6.267645398080181e-06,
-      "loss": 0.2947,
+      "grad_norm": 43.259742736816406,
+      "learning_rate": 9.401468097120271e-06,
+      "loss": 0.2424,
       "step": 18240
     },
     {
       "epoch": 3.434970826275174,
-      "grad_norm": 24.33298110961914,
-      "learning_rate": 6.260116694899304e-06,
-      "loss": 0.9247,
+      "grad_norm": 17.20220947265625,
+      "learning_rate": 9.390175042348956e-06,
+      "loss": 0.8755,
       "step": 18250
     },
     {
       "epoch": 3.4368530020703933,
-      "grad_norm": 12.793835639953613,
-      "learning_rate": 6.252587991718427e-06,
-      "loss": 0.4952,
+      "grad_norm": 35.17365646362305,
+      "learning_rate": 9.37888198757764e-06,
+      "loss": 0.5231,
       "step": 18260
     },
     {
       "epoch": 3.438735177865613,
-      "grad_norm": 7.1921234130859375,
-      "learning_rate": 6.245059288537549e-06,
-      "loss": 0.7946,
+      "grad_norm": 11.22391128540039,
+      "learning_rate": 9.367588932806323e-06,
+      "loss": 0.8492,
       "step": 18270
     },
     {
       "epoch": 3.4406173536608318,
-      "grad_norm": 21.25719451904297,
-      "learning_rate": 6.237530585356672e-06,
-      "loss": 0.73,
+      "grad_norm": 15.989961624145508,
+      "learning_rate": 9.356295878035009e-06,
+      "loss": 0.6255,
       "step": 18280
     },
     {
       "epoch": 3.442499529456051,
-      "grad_norm": 12.582639694213867,
-      "learning_rate": 6.230001882175796e-06,
-      "loss": 0.6567,
+      "grad_norm": 10.886322975158691,
+      "learning_rate": 9.345002823263693e-06,
+      "loss": 0.5604,
       "step": 18290
     },
     {
       "epoch": 3.4443817052512706,
-      "grad_norm": 2.455739736557007,
-      "learning_rate": 6.222473178994919e-06,
-      "loss": 0.5813,
+      "grad_norm": 0.8173845410346985,
+      "learning_rate": 9.333709768492378e-06,
+      "loss": 0.5142,
       "step": 18300
     },
     {
       "epoch": 3.4462638810464896,
-      "grad_norm": 15.017657279968262,
-      "learning_rate": 6.214944475814042e-06,
-      "loss": 0.7049,
+      "grad_norm": 12.035292625427246,
+      "learning_rate": 9.322416713721062e-06,
+      "loss": 0.7188,
       "step": 18310
     },
     {
       "epoch": 3.448146056841709,
-      "grad_norm": 10.231067657470703,
-      "learning_rate": 6.207415772633164e-06,
-      "loss": 0.4853,
+      "grad_norm": 24.942886352539062,
+      "learning_rate": 9.311123658949746e-06,
+      "loss": 0.5322,
       "step": 18320
     },
     {
       "epoch": 3.4500282326369285,
-      "grad_norm": 0.6758200526237488,
-      "learning_rate": 6.199887069452287e-06,
-      "loss": 0.6545,
+      "grad_norm": 0.916724681854248,
+      "learning_rate": 9.29983060417843e-06,
+      "loss": 0.6067,
       "step": 18330
     },
     {
       "epoch": 3.4519104084321475,
-      "grad_norm": 2.3873395919799805,
-      "learning_rate": 6.192358366271411e-06,
-      "loss": 0.6442,
+      "grad_norm": 1.716382384300232,
+      "learning_rate": 9.288537549407115e-06,
+      "loss": 0.5839,
       "step": 18340
     },
     {
       "epoch": 3.453792584227367,
-      "grad_norm": 8.824675559997559,
-      "learning_rate": 6.184829663090534e-06,
-      "loss": 0.6202,
+      "grad_norm": 5.684045314788818,
+      "learning_rate": 9.2772444946358e-06,
+      "loss": 0.6052,
       "step": 18350
     },
     {
       "epoch": 3.4556747600225863,
-      "grad_norm": 1.2755610942840576,
-      "learning_rate": 6.177300959909656e-06,
-      "loss": 0.6583,
+      "grad_norm": 3.6360864639282227,
+      "learning_rate": 9.265951439864482e-06,
+      "loss": 0.6305,
       "step": 18360
     },
     {
       "epoch": 3.4575569358178053,
-      "grad_norm": 8.281763076782227,
-      "learning_rate": 6.169772256728779e-06,
-      "loss": 0.5532,
+      "grad_norm": 7.340415954589844,
+      "learning_rate": 9.254658385093168e-06,
+      "loss": 0.4902,
       "step": 18370
     },
     {
       "epoch": 3.4594391116130248,
-      "grad_norm": 19.429561614990234,
-      "learning_rate": 6.1622435535479015e-06,
-      "loss": 0.6067,
+      "grad_norm": 12.40035343170166,
+      "learning_rate": 9.243365330321853e-06,
+      "loss": 0.6022,
       "step": 18380
     },
     {
       "epoch": 3.4613212874082437,
-      "grad_norm": 5.232605457305908,
-      "learning_rate": 6.154714850367025e-06,
-      "loss": 0.702,
+      "grad_norm": 6.647785186767578,
+      "learning_rate": 9.232072275550537e-06,
+      "loss": 0.6062,
       "step": 18390
     },
     {
       "epoch": 3.463203463203463,
-      "grad_norm": 38.510826110839844,
-      "learning_rate": 6.147186147186147e-06,
-      "loss": 0.6362,
+      "grad_norm": 16.84423828125,
+      "learning_rate": 9.22077922077922e-06,
+      "loss": 0.5901,
       "step": 18400
     },
     {
       "epoch": 3.4650856389986826,
-      "grad_norm": 5.245325088500977,
-      "learning_rate": 6.13965744400527e-06,
-      "loss": 0.7006,
+      "grad_norm": 4.819279193878174,
+      "learning_rate": 9.209486166007904e-06,
+      "loss": 0.6798,
       "step": 18410
     },
     {
       "epoch": 3.4669678147939016,
-      "grad_norm": 42.82587814331055,
-      "learning_rate": 6.132128740824393e-06,
-      "loss": 0.9136,
+      "grad_norm": 20.76621437072754,
+      "learning_rate": 9.19819311123659e-06,
+      "loss": 0.9342,
       "step": 18420
     },
     {
       "epoch": 3.468849990589121,
-      "grad_norm": 18.611064910888672,
-      "learning_rate": 6.124600037643517e-06,
-      "loss": 0.8132,
+      "grad_norm": 18.379783630371094,
+      "learning_rate": 9.186900056465275e-06,
+      "loss": 0.7618,
       "step": 18430
     },
     {
       "epoch": 3.4707321663843405,
-      "grad_norm": 24.1761417388916,
-      "learning_rate": 6.11707133446264e-06,
-      "loss": 0.7517,
+      "grad_norm": 22.449026107788086,
+      "learning_rate": 9.175607001693959e-06,
+      "loss": 0.7875,
       "step": 18440
     },
     {
       "epoch": 3.4726143421795594,
-      "grad_norm": 14.236517906188965,
-      "learning_rate": 6.109542631281762e-06,
-      "loss": 0.7883,
+      "grad_norm": 3.216062545776367,
+      "learning_rate": 9.164313946922642e-06,
+      "loss": 0.7226,
       "step": 18450
     },
     {
       "epoch": 3.474496517974779,
-      "grad_norm": 23.19389533996582,
-      "learning_rate": 6.102013928100885e-06,
-      "loss": 0.6778,
+      "grad_norm": 18.014450073242188,
+      "learning_rate": 9.153020892151328e-06,
+      "loss": 0.6888,
       "step": 18460
     },
     {
       "epoch": 3.4763786937699983,
-      "grad_norm": 17.408361434936523,
-      "learning_rate": 6.094485224920008e-06,
-      "loss": 0.425,
+      "grad_norm": 33.41580581665039,
+      "learning_rate": 9.141727837380012e-06,
+      "loss": 0.4367,
       "step": 18470
     },
     {
       "epoch": 3.4782608695652173,
-      "grad_norm": 1.1383931636810303,
-      "learning_rate": 6.086956521739132e-06,
-      "loss": 0.6786,
+      "grad_norm": 0.49927645921707153,
+      "learning_rate": 9.130434782608697e-06,
+      "loss": 0.683,
       "step": 18480
     },
     {
       "epoch": 3.4801430453604367,
-      "grad_norm": 18.74462890625,
-      "learning_rate": 6.079427818558254e-06,
-      "loss": 0.5072,
+      "grad_norm": 16.403583526611328,
+      "learning_rate": 9.11914172783738e-06,
+      "loss": 0.5037,
       "step": 18490
     },
     {
       "epoch": 3.4820252211556557,
-      "grad_norm": 12.033364295959473,
-      "learning_rate": 6.071899115377377e-06,
-      "loss": 0.8424,
+      "grad_norm": 19.10352325439453,
+      "learning_rate": 9.107848673066064e-06,
+      "loss": 0.8453,
       "step": 18500
     },
     {
       "epoch": 3.483907396950875,
-      "grad_norm": 13.617997169494629,
-      "learning_rate": 6.0643704121964995e-06,
-      "loss": 0.4691,
+      "grad_norm": 42.00209426879883,
+      "learning_rate": 9.09655561829475e-06,
+      "loss": 0.4188,
       "step": 18510
     },
     {
       "epoch": 3.4857895727460946,
-      "grad_norm": 18.998056411743164,
-      "learning_rate": 6.0568417090156224e-06,
-      "loss": 0.9883,
+      "grad_norm": 22.176076889038086,
+      "learning_rate": 9.085262563523434e-06,
+      "loss": 0.8926,
       "step": 18520
     },
     {
       "epoch": 3.4876717485413136,
-      "grad_norm": 13.122930526733398,
-      "learning_rate": 6.049313005834746e-06,
-      "loss": 0.7358,
+      "grad_norm": 13.631558418273926,
+      "learning_rate": 9.073969508752118e-06,
+      "loss": 0.7174,
       "step": 18530
     },
     {
       "epoch": 3.489553924336533,
-      "grad_norm": 57.448787689208984,
-      "learning_rate": 6.041784302653868e-06,
-      "loss": 0.4635,
+      "grad_norm": 16.817617416381836,
+      "learning_rate": 9.062676453980801e-06,
+      "loss": 0.3922,
       "step": 18540
     },
     {
       "epoch": 3.4914361001317524,
-      "grad_norm": 7.738725662231445,
-      "learning_rate": 6.034255599472991e-06,
-      "loss": 0.6152,
+      "grad_norm": 7.2339768409729,
+      "learning_rate": 9.051383399209486e-06,
+      "loss": 0.5843,
       "step": 18550
     },
     {
       "epoch": 3.4933182759269714,
-      "grad_norm": 6.369852542877197,
-      "learning_rate": 6.026726896292114e-06,
-      "loss": 0.7051,
+      "grad_norm": 7.782161712646484,
+      "learning_rate": 9.040090344438172e-06,
+      "loss": 0.6768,
       "step": 18560
     },
     {
       "epoch": 3.495200451722191,
-      "grad_norm": 9.485119819641113,
-      "learning_rate": 6.019198193111238e-06,
-      "loss": 0.5485,
+      "grad_norm": 14.207245826721191,
+      "learning_rate": 9.028797289666856e-06,
+      "loss": 0.5713,
       "step": 18570
     },
     {
       "epoch": 3.4970826275174103,
-      "grad_norm": 14.34579849243164,
-      "learning_rate": 6.01166948993036e-06,
-      "loss": 0.5845,
+      "grad_norm": 13.005362510681152,
+      "learning_rate": 9.017504234895539e-06,
+      "loss": 0.7295,
       "step": 18580
     },
     {
       "epoch": 3.4989648033126293,
-      "grad_norm": 8.484230995178223,
-      "learning_rate": 6.004140786749483e-06,
-      "loss": 0.5187,
+      "grad_norm": 5.221713066101074,
+      "learning_rate": 9.006211180124223e-06,
+      "loss": 0.5095,
       "step": 18590
     },
     {
       "epoch": 3.5008469791078487,
-      "grad_norm": 9.235902786254883,
-      "learning_rate": 5.996612083568606e-06,
-      "loss": 0.7469,
+      "grad_norm": 6.634870529174805,
+      "learning_rate": 8.994918125352909e-06,
+      "loss": 0.6905,
       "step": 18600
     },
     {
       "epoch": 3.5027291549030677,
-      "grad_norm": 19.91741371154785,
-      "learning_rate": 5.989083380387729e-06,
-      "loss": 0.5219,
+      "grad_norm": 22.765445709228516,
+      "learning_rate": 8.983625070581594e-06,
+      "loss": 0.4576,
       "step": 18610
     },
     {
       "epoch": 3.504611330698287,
-      "grad_norm": 33.60355758666992,
-      "learning_rate": 5.981554677206851e-06,
-      "loss": 0.8795,
+      "grad_norm": 12.745418548583984,
+      "learning_rate": 8.972332015810276e-06,
+      "loss": 0.7584,
       "step": 18620
     },
     {
       "epoch": 3.5064935064935066,
-      "grad_norm": 7.930181503295898,
-      "learning_rate": 5.9740259740259746e-06,
-      "loss": 0.7002,
+      "grad_norm": 7.202927112579346,
+      "learning_rate": 8.96103896103896e-06,
+      "loss": 0.6597,
       "step": 18630
     },
     {
       "epoch": 3.508375682288726,
-      "grad_norm": 25.92783546447754,
-      "learning_rate": 5.9664972708450975e-06,
-      "loss": 0.6384,
+      "grad_norm": 19.406003952026367,
+      "learning_rate": 8.949745906267645e-06,
+      "loss": 0.6717,
       "step": 18640
     },
     {
       "epoch": 3.510257858083945,
-      "grad_norm": 7.213856220245361,
-      "learning_rate": 5.95896856766422e-06,
-      "loss": 0.5609,
+      "grad_norm": 9.290009498596191,
+      "learning_rate": 8.938452851496331e-06,
+      "loss": 0.4197,
       "step": 18650
     },
     {
       "epoch": 3.5121400338791644,
-      "grad_norm": 29.71574592590332,
-      "learning_rate": 5.951439864483344e-06,
-      "loss": 1.0729,
+      "grad_norm": 29.786117553710938,
+      "learning_rate": 8.927159796725015e-06,
+      "loss": 1.0078,
       "step": 18660
     },
     {
       "epoch": 3.5140222096743834,
-      "grad_norm": 26.87563705444336,
-      "learning_rate": 5.943911161302465e-06,
-      "loss": 0.6124,
+      "grad_norm": 27.318464279174805,
+      "learning_rate": 8.915866741953698e-06,
+      "loss": 0.6775,
       "step": 18670
     },
     {
       "epoch": 3.515904385469603,
-      "grad_norm": 5.8294782638549805,
-      "learning_rate": 5.936382458121589e-06,
-      "loss": 0.4782,
+      "grad_norm": 3.237138509750366,
+      "learning_rate": 8.904573687182382e-06,
+      "loss": 0.429,
       "step": 18680
     },
     {
       "epoch": 3.5177865612648223,
-      "grad_norm": 7.82859468460083,
-      "learning_rate": 5.928853754940712e-06,
-      "loss": 0.5349,
+      "grad_norm": 5.157870292663574,
+      "learning_rate": 8.893280632411067e-06,
+      "loss": 0.5416,
       "step": 18690
     },
     {
       "epoch": 3.5196687370600412,
-      "grad_norm": 13.210909843444824,
-      "learning_rate": 5.921325051759835e-06,
-      "loss": 0.6603,
+      "grad_norm": 16.835599899291992,
+      "learning_rate": 8.881987577639753e-06,
+      "loss": 0.6814,
       "step": 18700
     },
     {
       "epoch": 3.5215509128552607,
-      "grad_norm": 9.926183700561523,
-      "learning_rate": 5.913796348578957e-06,
-      "loss": 0.5646,
+      "grad_norm": 15.013919830322266,
+      "learning_rate": 8.870694522868436e-06,
+      "loss": 0.6731,
       "step": 18710
     },
     {
       "epoch": 3.52343308865048,
-      "grad_norm": 12.920879364013672,
-      "learning_rate": 5.906267645398081e-06,
-      "loss": 0.6175,
+      "grad_norm": 20.667890548706055,
+      "learning_rate": 8.85940146809712e-06,
+      "loss": 0.5596,
       "step": 18720
     },
     {
       "epoch": 3.525315264445699,
-      "grad_norm": 20.73908233642578,
-      "learning_rate": 5.898738942217204e-06,
-      "loss": 0.7321,
+      "grad_norm": 14.67215347290039,
+      "learning_rate": 8.848108413325804e-06,
+      "loss": 0.6222,
       "step": 18730
     },
     {
       "epoch": 3.5271974402409185,
-      "grad_norm": 4.936865329742432,
-      "learning_rate": 5.891210239036327e-06,
-      "loss": 0.8867,
+      "grad_norm": 11.272614479064941,
+      "learning_rate": 8.83681535855449e-06,
+      "loss": 0.8376,
       "step": 18740
     },
     {
       "epoch": 3.529079616036138,
-      "grad_norm": 16.362173080444336,
-      "learning_rate": 5.883681535855449e-06,
-      "loss": 0.6746,
+      "grad_norm": 10.547713279724121,
+      "learning_rate": 8.825522303783173e-06,
+      "loss": 0.5504,
       "step": 18750
     },
     {
       "epoch": 3.530961791831357,
-      "grad_norm": 12.123333930969238,
-      "learning_rate": 5.876152832674572e-06,
-      "loss": 0.5928,
+      "grad_norm": 12.09305191040039,
+      "learning_rate": 8.814229249011858e-06,
+      "loss": 0.5685,
       "step": 18760
     },
     {
       "epoch": 3.5328439676265764,
-      "grad_norm": 16.486318588256836,
-      "learning_rate": 5.8686241294936955e-06,
-      "loss": 0.6531,
+      "grad_norm": 13.567402839660645,
+      "learning_rate": 8.802936194240542e-06,
+      "loss": 0.4545,
       "step": 18770
     },
     {
       "epoch": 3.5347261434217954,
-      "grad_norm": 5.473000526428223,
-      "learning_rate": 5.861095426312818e-06,
-      "loss": 0.7118,
+      "grad_norm": 8.988545417785645,
+      "learning_rate": 8.791643139469226e-06,
+      "loss": 0.7346,
       "step": 18780
     },
     {
       "epoch": 3.536608319217015,
-      "grad_norm": 12.31059741973877,
-      "learning_rate": 5.853566723131941e-06,
-      "loss": 0.613,
+      "grad_norm": 13.106191635131836,
+      "learning_rate": 8.780350084697912e-06,
+      "loss": 0.6023,
       "step": 18790
     },
     {
       "epoch": 3.5384904950122342,
-      "grad_norm": 10.703392028808594,
-      "learning_rate": 5.846038019951063e-06,
-      "loss": 0.6473,
+      "grad_norm": 28.7813777923584,
+      "learning_rate": 8.769057029926595e-06,
+      "loss": 0.6156,
       "step": 18800
     },
     {
       "epoch": 3.5403726708074537,
-      "grad_norm": 53.403602600097656,
-      "learning_rate": 5.838509316770186e-06,
-      "loss": 0.6796,
+      "grad_norm": 22.4103946685791,
+      "learning_rate": 8.75776397515528e-06,
+      "loss": 0.4925,
       "step": 18810
     },
     {
       "epoch": 3.5422548466026726,
-      "grad_norm": 10.487499237060547,
-      "learning_rate": 5.83098061358931e-06,
-      "loss": 0.5779,
+      "grad_norm": 20.627689361572266,
+      "learning_rate": 8.746470920383964e-06,
+      "loss": 0.6625,
       "step": 18820
     },
     {
       "epoch": 3.544137022397892,
-      "grad_norm": 11.738143920898438,
-      "learning_rate": 5.823451910408433e-06,
-      "loss": 0.7215,
+      "grad_norm": 8.956947326660156,
+      "learning_rate": 8.73517786561265e-06,
+      "loss": 0.5561,
       "step": 18830
     },
     {
       "epoch": 3.546019198193111,
-      "grad_norm": 13.654657363891602,
-      "learning_rate": 5.815923207227555e-06,
-      "loss": 0.6334,
+      "grad_norm": 7.5617594718933105,
+      "learning_rate": 8.723884810841333e-06,
+      "loss": 0.604,
       "step": 18840
     },
     {
       "epoch": 3.5479013739883305,
-      "grad_norm": 14.808473587036133,
-      "learning_rate": 5.808394504046678e-06,
-      "loss": 0.7413,
+      "grad_norm": 17.85495948791504,
+      "learning_rate": 8.712591756070017e-06,
+      "loss": 0.6587,
       "step": 18850
     },
     {
       "epoch": 3.54978354978355,
-      "grad_norm": 17.71009635925293,
-      "learning_rate": 5.800865800865802e-06,
-      "loss": 0.9974,
+      "grad_norm": 8.020273208618164,
+      "learning_rate": 8.701298701298701e-06,
+      "loss": 0.8588,
       "step": 18860
     },
     {
       "epoch": 3.551665725578769,
-      "grad_norm": 15.377747535705566,
-      "learning_rate": 5.793337097684925e-06,
-      "loss": 0.5763,
+      "grad_norm": 25.53630256652832,
+      "learning_rate": 8.690005646527386e-06,
+      "loss": 0.6081,
       "step": 18870
     },
     {
       "epoch": 3.5535479013739883,
-      "grad_norm": 12.32382869720459,
-      "learning_rate": 5.785808394504047e-06,
-      "loss": 0.4418,
+      "grad_norm": 5.950412273406982,
+      "learning_rate": 8.67871259175607e-06,
+      "loss": 0.3223,
       "step": 18880
     },
     {
       "epoch": 3.5554300771692073,
-      "grad_norm": 12.920076370239258,
-      "learning_rate": 5.77827969132317e-06,
-      "loss": 0.6994,
+      "grad_norm": 7.792263031005859,
+      "learning_rate": 8.667419536984755e-06,
+      "loss": 0.6444,
       "step": 18890
     },
     {
       "epoch": 3.5573122529644268,
-      "grad_norm": 15.59774112701416,
-      "learning_rate": 5.770750988142293e-06,
-      "loss": 0.5499,
+      "grad_norm": 8.66256332397461,
+      "learning_rate": 8.656126482213439e-06,
+      "loss": 0.3844,
       "step": 18900
     },
     {
       "epoch": 3.559194428759646,
-      "grad_norm": 22.626419067382812,
-      "learning_rate": 5.763222284961416e-06,
-      "loss": 0.7776,
+      "grad_norm": 22.50640296936035,
+      "learning_rate": 8.644833427442123e-06,
+      "loss": 0.6602,
       "step": 18910
     },
     {
       "epoch": 3.5610766045548656,
-      "grad_norm": 4.442861080169678,
-      "learning_rate": 5.755693581780539e-06,
-      "loss": 0.3559,
+      "grad_norm": 2.1984527111053467,
+      "learning_rate": 8.633540372670808e-06,
+      "loss": 0.288,
       "step": 18920
     },
     {
       "epoch": 3.5629587803500846,
-      "grad_norm": 11.74802017211914,
-      "learning_rate": 5.748164878599661e-06,
-      "loss": 0.714,
+      "grad_norm": 8.957244873046875,
+      "learning_rate": 8.622247317899492e-06,
+      "loss": 0.7115,
       "step": 18930
     },
     {
       "epoch": 3.564840956145304,
-      "grad_norm": 15.844473838806152,
-      "learning_rate": 5.740636175418784e-06,
-      "loss": 0.6158,
+      "grad_norm": 12.717720985412598,
+      "learning_rate": 8.610954263128176e-06,
+      "loss": 0.5397,
       "step": 18940
     },
     {
       "epoch": 3.566723131940523,
-      "grad_norm": 32.16844940185547,
-      "learning_rate": 5.733107472237908e-06,
-      "loss": 0.8462,
+      "grad_norm": 17.717315673828125,
+      "learning_rate": 8.59966120835686e-06,
+      "loss": 0.8032,
       "step": 18950
     },
     {
       "epoch": 3.5686053077357425,
-      "grad_norm": 28.286787033081055,
-      "learning_rate": 5.725578769057031e-06,
-      "loss": 0.6199,
+      "grad_norm": 35.10055160522461,
+      "learning_rate": 8.588368153585545e-06,
+      "loss": 0.6323,
       "step": 18960
     },
     {
       "epoch": 3.570487483530962,
-      "grad_norm": 38.367637634277344,
-      "learning_rate": 5.718050065876153e-06,
-      "loss": 1.119,
+      "grad_norm": 16.153400421142578,
+      "learning_rate": 8.57707509881423e-06,
+      "loss": 0.9937,
       "step": 18970
     },
     {
       "epoch": 3.5723696593261813,
-      "grad_norm": 8.818977355957031,
-      "learning_rate": 5.710521362695276e-06,
-      "loss": 0.6292,
+      "grad_norm": 7.813040733337402,
+      "learning_rate": 8.565782044042914e-06,
+      "loss": 0.5768,
       "step": 18980
     },
     {
       "epoch": 3.5742518351214003,
-      "grad_norm": 54.24068832397461,
-      "learning_rate": 5.702992659514399e-06,
-      "loss": 0.5373,
+      "grad_norm": 38.98237228393555,
+      "learning_rate": 8.554488989271598e-06,
+      "loss": 0.4722,
       "step": 18990
     },
     {
       "epoch": 3.5761340109166198,
-      "grad_norm": 17.338998794555664,
-      "learning_rate": 5.695463956333523e-06,
-      "loss": 0.6531,
+      "grad_norm": 8.980436325073242,
+      "learning_rate": 8.543195934500283e-06,
+      "loss": 0.6281,
       "step": 19000
     },
     {
       "epoch": 3.5780161867118387,
-      "grad_norm": 9.126494407653809,
-      "learning_rate": 5.687935253152645e-06,
-      "loss": 0.8443,
+      "grad_norm": 4.206551551818848,
+      "learning_rate": 8.531902879728965e-06,
+      "loss": 0.7266,
       "step": 19010
     },
     {
       "epoch": 3.579898362507058,
-      "grad_norm": 6.135142803192139,
-      "learning_rate": 5.680406549971768e-06,
-      "loss": 0.68,
+      "grad_norm": 7.171810150146484,
+      "learning_rate": 8.520609824957651e-06,
+      "loss": 0.5348,
       "step": 19020
     },
     {
       "epoch": 3.5817805383022776,
-      "grad_norm": 39.82160568237305,
-      "learning_rate": 5.672877846790891e-06,
-      "loss": 0.4722,
+      "grad_norm": 8.998128890991211,
+      "learning_rate": 8.509316770186336e-06,
+      "loss": 0.3459,
       "step": 19030
     },
     {
       "epoch": 3.5836627140974966,
-      "grad_norm": 1.3005073070526123,
-      "learning_rate": 5.6653491436100135e-06,
-      "loss": 0.5634,
+      "grad_norm": 11.471923828125,
+      "learning_rate": 8.49802371541502e-06,
+      "loss": 0.5549,
       "step": 19040
     },
     {
       "epoch": 3.585544889892716,
-      "grad_norm": 2.85640811920166,
-      "learning_rate": 5.657820440429137e-06,
-      "loss": 0.5805,
+      "grad_norm": 1.2988466024398804,
+      "learning_rate": 8.486730660643705e-06,
+      "loss": 0.4202,
       "step": 19050
     },
     {
       "epoch": 3.587427065687935,
-      "grad_norm": 29.665571212768555,
-      "learning_rate": 5.650291737248259e-06,
-      "loss": 0.9181,
+      "grad_norm": 12.022832870483398,
+      "learning_rate": 8.475437605872387e-06,
+      "loss": 0.882,
       "step": 19060
     },
     {
       "epoch": 3.5893092414831544,
-      "grad_norm": 33.12388229370117,
-      "learning_rate": 5.642763034067382e-06,
-      "loss": 0.5326,
+      "grad_norm": 18.806732177734375,
+      "learning_rate": 8.464144551101073e-06,
+      "loss": 0.4618,
       "step": 19070
     },
     {
       "epoch": 3.591191417278374,
-      "grad_norm": 4.879305362701416,
-      "learning_rate": 5.635234330886505e-06,
-      "loss": 0.6239,
+      "grad_norm": 7.265424728393555,
+      "learning_rate": 8.452851496329758e-06,
+      "loss": 0.5997,
       "step": 19080
     },
     {
       "epoch": 3.5930735930735933,
-      "grad_norm": 33.6097297668457,
-      "learning_rate": 5.627705627705629e-06,
-      "loss": 0.7447,
+      "grad_norm": 11.584159851074219,
+      "learning_rate": 8.441558441558442e-06,
+      "loss": 0.6913,
       "step": 19090
     },
     {
       "epoch": 3.5949557688688123,
-      "grad_norm": 27.47477912902832,
-      "learning_rate": 5.620176924524751e-06,
-      "loss": 0.6465,
+      "grad_norm": 39.18511199951172,
+      "learning_rate": 8.430265386787125e-06,
+      "loss": 0.6348,
       "step": 19100
     },
     {
       "epoch": 3.5968379446640317,
-      "grad_norm": 26.439849853515625,
-      "learning_rate": 5.612648221343874e-06,
-      "loss": 0.9106,
+      "grad_norm": 29.593963623046875,
+      "learning_rate": 8.418972332015811e-06,
+      "loss": 0.7804,
       "step": 19110
     },
     {
       "epoch": 3.5987201204592507,
-      "grad_norm": 23.74281883239746,
-      "learning_rate": 5.605119518162997e-06,
-      "loss": 0.7119,
+      "grad_norm": 12.343049049377441,
+      "learning_rate": 8.407679277244495e-06,
+      "loss": 0.5887,
       "step": 19120
     },
     {
       "epoch": 3.60060229625447,
-      "grad_norm": 27.2669677734375,
-      "learning_rate": 5.59759081498212e-06,
-      "loss": 0.6561,
+      "grad_norm": 27.028095245361328,
+      "learning_rate": 8.39638622247318e-06,
+      "loss": 0.5748,
       "step": 19130
     },
     {
       "epoch": 3.6024844720496896,
-      "grad_norm": 10.283041000366211,
-      "learning_rate": 5.590062111801242e-06,
-      "loss": 0.5015,
+      "grad_norm": 10.208629608154297,
+      "learning_rate": 8.385093167701862e-06,
+      "loss": 0.5336,
       "step": 19140
     },
     {
       "epoch": 3.6043666478449086,
-      "grad_norm": 5.571829795837402,
-      "learning_rate": 5.582533408620366e-06,
-      "loss": 0.3962,
+      "grad_norm": 6.941380023956299,
+      "learning_rate": 8.373800112930547e-06,
+      "loss": 0.5231,
       "step": 19150
     },
     {
       "epoch": 3.606248823640128,
-      "grad_norm": 16.38119888305664,
-      "learning_rate": 5.5750047054394886e-06,
-      "loss": 0.9378,
+      "grad_norm": 24.04265785217285,
+      "learning_rate": 8.362507058159233e-06,
+      "loss": 0.9771,
       "step": 19160
     },
     {
       "epoch": 3.608130999435347,
-      "grad_norm": 17.61245346069336,
-      "learning_rate": 5.5674760022586115e-06,
-      "loss": 0.7312,
+      "grad_norm": 7.969100475311279,
+      "learning_rate": 8.351214003387917e-06,
+      "loss": 0.6908,
       "step": 19170
     },
     {
       "epoch": 3.6100131752305664,
-      "grad_norm": 117.15702056884766,
-      "learning_rate": 5.559947299077734e-06,
-      "loss": 0.4922,
+      "grad_norm": 5.254550457000732,
+      "learning_rate": 8.339920948616602e-06,
+      "loss": 0.4686,
       "step": 19180
     },
     {
       "epoch": 3.611895351025786,
-      "grad_norm": 8.270508766174316,
-      "learning_rate": 5.5524185958968565e-06,
-      "loss": 0.8726,
+      "grad_norm": 11.416996955871582,
+      "learning_rate": 8.328627893845284e-06,
+      "loss": 0.9016,
       "step": 19190
     },
     {
       "epoch": 3.6137775268210053,
-      "grad_norm": 39.04629135131836,
-      "learning_rate": 5.54488989271598e-06,
-      "loss": 0.7457,
+      "grad_norm": 22.956119537353516,
+      "learning_rate": 8.31733483907397e-06,
+      "loss": 0.718,
       "step": 19200
     },
     {
       "epoch": 3.6156597026162243,
-      "grad_norm": 43.17094802856445,
-      "learning_rate": 5.537361189535103e-06,
-      "loss": 0.7313,
+      "grad_norm": 20.53351402282715,
+      "learning_rate": 8.306041784302655e-06,
+      "loss": 0.6639,
       "step": 19210
     },
     {
       "epoch": 3.6175418784114437,
-      "grad_norm": 22.584318161010742,
-      "learning_rate": 5.529832486354226e-06,
-      "loss": 0.4514,
+      "grad_norm": 15.131000518798828,
+      "learning_rate": 8.294748729531339e-06,
+      "loss": 0.3244,
       "step": 19220
     },
     {
       "epoch": 3.6194240542066627,
-      "grad_norm": 25.183490753173828,
-      "learning_rate": 5.522303783173348e-06,
-      "loss": 0.5244,
+      "grad_norm": 24.35951805114746,
+      "learning_rate": 8.283455674760022e-06,
+      "loss": 0.467,
       "step": 19230
     },
     {
       "epoch": 3.621306230001882,
-      "grad_norm": 22.077592849731445,
-      "learning_rate": 5.514775079992472e-06,
-      "loss": 0.8154,
+      "grad_norm": 10.79614543914795,
+      "learning_rate": 8.272162619988706e-06,
+      "loss": 0.6985,
       "step": 19240
     },
     {
       "epoch": 3.6231884057971016,
-      "grad_norm": 7.8474016189575195,
-      "learning_rate": 5.507246376811595e-06,
-      "loss": 0.7153,
+      "grad_norm": 8.274187088012695,
+      "learning_rate": 8.260869565217392e-06,
+      "loss": 0.7224,
       "step": 19250
     },
     {
       "epoch": 3.625070581592321,
-      "grad_norm": 0.6397775411605835,
-      "learning_rate": 5.499717673630718e-06,
-      "loss": 0.7141,
+      "grad_norm": 0.10666459053754807,
+      "learning_rate": 8.249576510446077e-06,
+      "loss": 0.6622,
       "step": 19260
     },
     {
       "epoch": 3.62695275738754,
-      "grad_norm": 15.466002464294434,
-      "learning_rate": 5.492188970449841e-06,
-      "loss": 0.7019,
+      "grad_norm": 24.340177536010742,
+      "learning_rate": 8.238283455674761e-06,
+      "loss": 0.7348,
       "step": 19270
     },
     {
       "epoch": 3.6288349331827594,
-      "grad_norm": 30.216781616210938,
-      "learning_rate": 5.484660267268963e-06,
-      "loss": 0.8668,
+      "grad_norm": 20.038578033447266,
+      "learning_rate": 8.226990400903444e-06,
+      "loss": 0.8442,
       "step": 19280
     },
     {
       "epoch": 3.6307171089779784,
-      "grad_norm": 11.025100708007812,
-      "learning_rate": 5.4771315640880865e-06,
-      "loss": 0.6875,
+      "grad_norm": 11.465856552124023,
+      "learning_rate": 8.215697346132128e-06,
+      "loss": 0.7398,
       "step": 19290
     },
     {
       "epoch": 3.632599284773198,
-      "grad_norm": 17.533348083496094,
-      "learning_rate": 5.4696028609072095e-06,
-      "loss": 0.4557,
+      "grad_norm": 13.356294631958008,
+      "learning_rate": 8.204404291360814e-06,
+      "loss": 0.4302,
       "step": 19300
     },
     {
       "epoch": 3.6344814605684173,
-      "grad_norm": 29.006498336791992,
-      "learning_rate": 5.462074157726332e-06,
-      "loss": 0.7939,
+      "grad_norm": 17.840681076049805,
+      "learning_rate": 8.193111236589499e-06,
+      "loss": 0.562,
       "step": 19310
     },
     {
       "epoch": 3.6363636363636362,
-      "grad_norm": 18.006269454956055,
-      "learning_rate": 5.4545454545454545e-06,
-      "loss": 0.5706,
+      "grad_norm": 0.5151638984680176,
+      "learning_rate": 8.181818181818181e-06,
+      "loss": 0.5005,
       "step": 19320
     },
     {
       "epoch": 3.6382458121588557,
-      "grad_norm": 11.842194557189941,
-      "learning_rate": 5.447016751364577e-06,
-      "loss": 0.7348,
+      "grad_norm": 8.19641399383545,
+      "learning_rate": 8.170525127046866e-06,
+      "loss": 0.6157,
       "step": 19330
     },
     {
       "epoch": 3.6401279879540747,
-      "grad_norm": 5.13842248916626,
-      "learning_rate": 5.439488048183701e-06,
-      "loss": 0.5898,
+      "grad_norm": 19.37883949279785,
+      "learning_rate": 8.159232072275552e-06,
+      "loss": 0.5518,
       "step": 19340
     },
     {
       "epoch": 3.642010163749294,
-      "grad_norm": 2.101395845413208,
-      "learning_rate": 5.431959345002824e-06,
-      "loss": 0.6409,
+      "grad_norm": 2.3880703449249268,
+      "learning_rate": 8.147939017504236e-06,
+      "loss": 0.5452,
       "step": 19350
     },
     {
       "epoch": 3.6438923395445135,
-      "grad_norm": 8.308853149414062,
-      "learning_rate": 5.424430641821946e-06,
-      "loss": 0.8893,
+      "grad_norm": 12.739452362060547,
+      "learning_rate": 8.136645962732919e-06,
+      "loss": 0.945,
       "step": 19360
     },
     {
       "epoch": 3.645774515339733,
-      "grad_norm": 6.142594814300537,
-      "learning_rate": 5.416901938641069e-06,
-      "loss": 0.9554,
+      "grad_norm": 14.899114608764648,
+      "learning_rate": 8.125352907961603e-06,
+      "loss": 0.9551,
       "step": 19370
     },
     {
       "epoch": 3.647656691134952,
-      "grad_norm": 4.47888708114624,
-      "learning_rate": 5.409373235460193e-06,
-      "loss": 0.8409,
+      "grad_norm": 24.746118545532227,
+      "learning_rate": 8.114059853190288e-06,
+      "loss": 0.7507,
       "step": 19380
     },
     {
       "epoch": 3.6495388669301714,
-      "grad_norm": 9.741399765014648,
-      "learning_rate": 5.401844532279316e-06,
-      "loss": 0.66,
+      "grad_norm": 10.281793594360352,
+      "learning_rate": 8.102766798418974e-06,
+      "loss": 0.5547,
       "step": 19390
     },
     {
       "epoch": 3.6514210427253904,
-      "grad_norm": 2.779855966567993,
-      "learning_rate": 5.394315829098439e-06,
-      "loss": 0.7436,
+      "grad_norm": 0.5525919198989868,
+      "learning_rate": 8.091473743647658e-06,
+      "loss": 0.5547,
       "step": 19400
     },
     {
       "epoch": 3.65330321852061,
-      "grad_norm": 19.320926666259766,
-      "learning_rate": 5.386787125917561e-06,
-      "loss": 0.6752,
+      "grad_norm": 51.09165573120117,
+      "learning_rate": 8.08018068887634e-06,
+      "loss": 0.6717,
       "step": 19410
     },
     {
       "epoch": 3.6551853943158292,
-      "grad_norm": 27.487380981445312,
-      "learning_rate": 5.379258422736684e-06,
-      "loss": 0.7517,
+      "grad_norm": 18.030263900756836,
+      "learning_rate": 8.068887634105025e-06,
+      "loss": 0.6471,
       "step": 19420
     },
     {
       "epoch": 3.657067570111048,
-      "grad_norm": 24.027660369873047,
-      "learning_rate": 5.3717297195558074e-06,
-      "loss": 0.6483,
+      "grad_norm": 10.15441608428955,
+      "learning_rate": 8.05759457933371e-06,
+      "loss": 0.5754,
       "step": 19430
     },
     {
       "epoch": 3.6589497459062676,
-      "grad_norm": 9.766282081604004,
-      "learning_rate": 5.36420101637493e-06,
-      "loss": 0.5122,
+      "grad_norm": 8.153667449951172,
+      "learning_rate": 8.046301524562396e-06,
+      "loss": 0.4898,
       "step": 19440
     },
     {
       "epoch": 3.660831921701487,
-      "grad_norm": 20.8446102142334,
-      "learning_rate": 5.3566723131940524e-06,
-      "loss": 0.7974,
+      "grad_norm": 22.26079750061035,
+      "learning_rate": 8.035008469791078e-06,
+      "loss": 0.6954,
       "step": 19450
     },
     {
       "epoch": 3.662714097496706,
-      "grad_norm": 15.975231170654297,
-      "learning_rate": 5.349143610013175e-06,
-      "loss": 0.6866,
+      "grad_norm": 8.062853813171387,
+      "learning_rate": 8.023715415019763e-06,
+      "loss": 0.5817,
       "step": 19460
     },
     {
       "epoch": 3.6645962732919255,
-      "grad_norm": 30.064542770385742,
-      "learning_rate": 5.341614906832298e-06,
-      "loss": 0.6704,
+      "grad_norm": 20.452653884887695,
+      "learning_rate": 8.012422360248447e-06,
+      "loss": 0.6119,
       "step": 19470
     },
     {
       "epoch": 3.666478449087145,
-      "grad_norm": 16.590923309326172,
-      "learning_rate": 5.334086203651422e-06,
-      "loss": 0.7576,
+      "grad_norm": 16.677806854248047,
+      "learning_rate": 8.001129305477133e-06,
+      "loss": 0.5955,
       "step": 19480
     },
     {
       "epoch": 3.668360624882364,
-      "grad_norm": 11.002893447875977,
-      "learning_rate": 5.326557500470544e-06,
-      "loss": 0.5608,
+      "grad_norm": 13.450298309326172,
+      "learning_rate": 7.989836250705816e-06,
+      "loss": 0.4935,
       "step": 19490
     },
     {
       "epoch": 3.6702428006775834,
-      "grad_norm": 10.331709861755371,
-      "learning_rate": 5.319028797289667e-06,
-      "loss": 0.8275,
+      "grad_norm": 12.314119338989258,
+      "learning_rate": 7.9785431959345e-06,
+      "loss": 0.7169,
       "step": 19500
     },
     {
       "epoch": 3.6721249764728023,
-      "grad_norm": 17.15074348449707,
-      "learning_rate": 5.31150009410879e-06,
-      "loss": 0.7005,
+      "grad_norm": 13.286541938781738,
+      "learning_rate": 7.967250141163185e-06,
+      "loss": 0.7006,
       "step": 19510
     },
     {
       "epoch": 3.6740071522680218,
-      "grad_norm": 11.861909866333008,
-      "learning_rate": 5.303971390927914e-06,
-      "loss": 0.4037,
+      "grad_norm": 8.442977905273438,
+      "learning_rate": 7.955957086391869e-06,
+      "loss": 0.4225,
       "step": 19520
     },
     {
       "epoch": 3.675889328063241,
-      "grad_norm": 17.04161262512207,
-      "learning_rate": 5.296442687747037e-06,
-      "loss": 0.5099,
+      "grad_norm": 18.344675064086914,
+      "learning_rate": 7.944664031620555e-06,
+      "loss": 0.5021,
       "step": 19530
     },
     {
       "epoch": 3.6777715038584606,
-      "grad_norm": 20.979482650756836,
-      "learning_rate": 5.288913984566159e-06,
-      "loss": 0.5826,
+      "grad_norm": 22.64541244506836,
+      "learning_rate": 7.933370976849238e-06,
+      "loss": 0.596,
       "step": 19540
     },
     {
       "epoch": 3.6796536796536796,
-      "grad_norm": 18.8994140625,
-      "learning_rate": 5.281385281385282e-06,
-      "loss": 0.774,
+      "grad_norm": 20.384031295776367,
+      "learning_rate": 7.922077922077922e-06,
+      "loss": 0.6877,
       "step": 19550
     },
     {
       "epoch": 3.681535855448899,
-      "grad_norm": 4.718742370605469,
-      "learning_rate": 5.2738565782044046e-06,
-      "loss": 0.7723,
+      "grad_norm": 8.203783988952637,
+      "learning_rate": 7.910784867306606e-06,
+      "loss": 0.6984,
       "step": 19560
     },
     {
       "epoch": 3.683418031244118,
-      "grad_norm": 18.240686416625977,
-      "learning_rate": 5.266327875023528e-06,
-      "loss": 0.6407,
+      "grad_norm": 8.949334144592285,
+      "learning_rate": 7.89949181253529e-06,
+      "loss": 0.642,
       "step": 19570
     },
     {
       "epoch": 3.6853002070393375,
-      "grad_norm": 6.685810565948486,
-      "learning_rate": 5.25879917184265e-06,
-      "loss": 0.7073,
+      "grad_norm": 9.0608491897583,
+      "learning_rate": 7.888198757763975e-06,
+      "loss": 0.6827,
       "step": 19580
     },
     {
       "epoch": 3.687182382834557,
-      "grad_norm": 15.091521263122559,
-      "learning_rate": 5.251270468661773e-06,
-      "loss": 0.6525,
+      "grad_norm": 11.608131408691406,
+      "learning_rate": 7.87690570299266e-06,
+      "loss": 0.5154,
       "step": 19590
     },
     {
       "epoch": 3.689064558629776,
-      "grad_norm": 26.886425018310547,
-      "learning_rate": 5.243741765480896e-06,
-      "loss": 0.847,
+      "grad_norm": 42.18769836425781,
+      "learning_rate": 7.865612648221344e-06,
+      "loss": 0.6003,
       "step": 19600
     },
     {
       "epoch": 3.6909467344249953,
-      "grad_norm": 11.26270866394043,
-      "learning_rate": 5.236213062300019e-06,
-      "loss": 0.5092,
+      "grad_norm": 10.675950050354004,
+      "learning_rate": 7.854319593450028e-06,
+      "loss": 0.4056,
       "step": 19610
     },
     {
       "epoch": 3.6928289102202143,
-      "grad_norm": 17.176101684570312,
-      "learning_rate": 5.228684359119141e-06,
-      "loss": 0.6328,
+      "grad_norm": 21.224416732788086,
+      "learning_rate": 7.843026538678713e-06,
+      "loss": 0.5917,
       "step": 19620
     },
     {
       "epoch": 3.6947110860154337,
-      "grad_norm": 15.91916275024414,
-      "learning_rate": 5.221155655938265e-06,
-      "loss": 0.6865,
+      "grad_norm": 13.734648704528809,
+      "learning_rate": 7.831733483907397e-06,
+      "loss": 0.6431,
       "step": 19630
     },
     {
       "epoch": 3.696593261810653,
-      "grad_norm": 10.485185623168945,
-      "learning_rate": 5.213626952757388e-06,
-      "loss": 0.5489,
+      "grad_norm": 5.891987323760986,
+      "learning_rate": 7.820440429136081e-06,
+      "loss": 0.4731,
       "step": 19640
     },
     {
       "epoch": 3.6984754376058726,
-      "grad_norm": 14.304201126098633,
-      "learning_rate": 5.206098249576511e-06,
-      "loss": 0.6536,
+      "grad_norm": 6.475367546081543,
+      "learning_rate": 7.809147374364766e-06,
+      "loss": 0.5094,
       "step": 19650
     },
     {
       "epoch": 3.7003576134010916,
-      "grad_norm": 10.666767120361328,
-      "learning_rate": 5.198569546395635e-06,
-      "loss": 0.5502,
+      "grad_norm": 5.192591667175293,
+      "learning_rate": 7.79785431959345e-06,
+      "loss": 0.508,
       "step": 19660
     },
     {
       "epoch": 3.702239789196311,
-      "grad_norm": 13.534510612487793,
-      "learning_rate": 5.191040843214757e-06,
-      "loss": 0.5719,
+      "grad_norm": 5.926048755645752,
+      "learning_rate": 7.786561264822135e-06,
+      "loss": 0.5057,
       "step": 19670
     },
     {
       "epoch": 3.70412196499153,
-      "grad_norm": 16.280838012695312,
-      "learning_rate": 5.18351214003388e-06,
-      "loss": 0.8457,
+      "grad_norm": 41.563819885253906,
+      "learning_rate": 7.775268210050819e-06,
+      "loss": 0.7474,
       "step": 19680
     },
     {
       "epoch": 3.7060041407867494,
-      "grad_norm": 21.206565856933594,
-      "learning_rate": 5.1759834368530025e-06,
-      "loss": 0.4464,
+      "grad_norm": 22.899145126342773,
+      "learning_rate": 7.763975155279503e-06,
+      "loss": 0.4772,
       "step": 19690
     },
     {
       "epoch": 3.707886316581969,
-      "grad_norm": 2.042757987976074,
-      "learning_rate": 5.1684547336721255e-06,
-      "loss": 0.4024,
+      "grad_norm": 1.0630342960357666,
+      "learning_rate": 7.752682100508188e-06,
+      "loss": 0.3682,
       "step": 19700
     },
     {
       "epoch": 3.709768492377188,
-      "grad_norm": 13.518824577331543,
-      "learning_rate": 5.1609260304912475e-06,
-      "loss": 0.723,
+      "grad_norm": 10.798871994018555,
+      "learning_rate": 7.741389045736872e-06,
+      "loss": 0.7488,
       "step": 19710
     },
     {
       "epoch": 3.7116506681724073,
-      "grad_norm": 23.17169761657715,
-      "learning_rate": 5.153397327310371e-06,
-      "loss": 0.4086,
+      "grad_norm": 9.957785606384277,
+      "learning_rate": 7.730095990965557e-06,
+      "loss": 0.4364,
       "step": 19720
     },
     {
       "epoch": 3.7135328439676267,
-      "grad_norm": 10.826898574829102,
-      "learning_rate": 5.145868624129494e-06,
-      "loss": 0.4519,
+      "grad_norm": 3.4311161041259766,
+      "learning_rate": 7.718802936194241e-06,
+      "loss": 0.4192,
       "step": 19730
     },
     {
       "epoch": 3.7154150197628457,
-      "grad_norm": 10.846261024475098,
-      "learning_rate": 5.138339920948617e-06,
-      "loss": 0.6457,
+      "grad_norm": 10.924822807312012,
+      "learning_rate": 7.707509881422925e-06,
+      "loss": 0.5174,
       "step": 19740
     },
     {
       "epoch": 3.717297195558065,
-      "grad_norm": 6.996768474578857,
-      "learning_rate": 5.130811217767739e-06,
-      "loss": 0.5838,
+      "grad_norm": 11.127129554748535,
+      "learning_rate": 7.696216826651608e-06,
+      "loss": 0.4632,
       "step": 19750
     },
     {
       "epoch": 3.7191793713532846,
-      "grad_norm": 2.6749439239501953,
-      "learning_rate": 5.123282514586862e-06,
-      "loss": 0.5056,
+      "grad_norm": 2.7011196613311768,
+      "learning_rate": 7.684923771880294e-06,
+      "loss": 0.499,
       "step": 19760
     },
     {
       "epoch": 3.7210615471485036,
-      "grad_norm": 2.4390110969543457,
-      "learning_rate": 5.115753811405986e-06,
-      "loss": 0.6283,
+      "grad_norm": 3.9069223403930664,
+      "learning_rate": 7.673630717108978e-06,
+      "loss": 0.6029,
       "step": 19770
     },
     {
       "epoch": 3.722943722943723,
-      "grad_norm": 6.546153545379639,
-      "learning_rate": 5.108225108225109e-06,
-      "loss": 0.8119,
+      "grad_norm": 4.461427211761475,
+      "learning_rate": 7.662337662337663e-06,
+      "loss": 0.7491,
       "step": 19780
     },
     {
       "epoch": 3.724825898738942,
-      "grad_norm": 22.23106575012207,
-      "learning_rate": 5.100696405044232e-06,
-      "loss": 0.7151,
+      "grad_norm": 19.533382415771484,
+      "learning_rate": 7.651044607566347e-06,
+      "loss": 0.6521,
       "step": 19790
     },
     {
       "epoch": 3.7267080745341614,
-      "grad_norm": 15.080765724182129,
-      "learning_rate": 5.093167701863354e-06,
-      "loss": 0.629,
+      "grad_norm": 2.6040291786193848,
+      "learning_rate": 7.63975155279503e-06,
+      "loss": 0.4989,
       "step": 19800
     },
     {
       "epoch": 3.728590250329381,
-      "grad_norm": 13.616122245788574,
-      "learning_rate": 5.085638998682478e-06,
-      "loss": 0.6677,
+      "grad_norm": 11.266732215881348,
+      "learning_rate": 7.628458498023715e-06,
+      "loss": 0.5427,
       "step": 19810
     },
     {
       "epoch": 3.7304724261246003,
-      "grad_norm": 3.014514684677124,
-      "learning_rate": 5.0781102955016005e-06,
-      "loss": 0.5376,
+      "grad_norm": 0.7041659951210022,
+      "learning_rate": 7.6171654432524e-06,
+      "loss": 0.482,
       "step": 19820
     },
     {
       "epoch": 3.7323546019198193,
-      "grad_norm": 5.815634250640869,
-      "learning_rate": 5.0705815923207234e-06,
-      "loss": 0.6015,
+      "grad_norm": 7.669864654541016,
+      "learning_rate": 7.605872388481085e-06,
+      "loss": 0.4862,
       "step": 19830
     },
     {
       "epoch": 3.7342367777150387,
-      "grad_norm": 9.241657257080078,
-      "learning_rate": 5.0630528891398455e-06,
-      "loss": 0.8149,
+      "grad_norm": 10.8816499710083,
+      "learning_rate": 7.594579333709768e-06,
+      "loss": 0.7602,
       "step": 19840
     },
     {
       "epoch": 3.7361189535102577,
-      "grad_norm": 24.852537155151367,
-      "learning_rate": 5.0555241859589684e-06,
-      "loss": 0.4347,
+      "grad_norm": 12.468655586242676,
+      "learning_rate": 7.583286278938453e-06,
+      "loss": 0.4088,
       "step": 19850
     },
     {
       "epoch": 3.738001129305477,
-      "grad_norm": 7.402453422546387,
-      "learning_rate": 5.047995482778092e-06,
-      "loss": 0.3041,
+      "grad_norm": 0.618268609046936,
+      "learning_rate": 7.571993224167137e-06,
+      "loss": 0.2964,
       "step": 19860
     },
     {
       "epoch": 3.7398833051006966,
-      "grad_norm": 14.335990905761719,
-      "learning_rate": 5.040466779597215e-06,
-      "loss": 0.821,
+      "grad_norm": 19.64562225341797,
+      "learning_rate": 7.560700169395822e-06,
+      "loss": 0.7775,
       "step": 19870
     },
     {
       "epoch": 3.7417654808959155,
-      "grad_norm": 16.409059524536133,
-      "learning_rate": 5.032938076416338e-06,
-      "loss": 0.9245,
+      "grad_norm": 18.206167221069336,
+      "learning_rate": 7.549407114624507e-06,
+      "loss": 0.7153,
       "step": 19880
     },
     {
       "epoch": 3.743647656691135,
-      "grad_norm": 13.059428215026855,
-      "learning_rate": 5.02540937323546e-06,
-      "loss": 0.5391,
+      "grad_norm": 7.192927837371826,
+      "learning_rate": 7.53811405985319e-06,
+      "loss": 0.4181,
       "step": 19890
     },
     {
       "epoch": 3.745529832486354,
-      "grad_norm": 14.962140083312988,
-      "learning_rate": 5.017880670054583e-06,
-      "loss": 0.4168,
+      "grad_norm": 29.97498893737793,
+      "learning_rate": 7.5268210050818746e-06,
+      "loss": 0.3835,
       "step": 19900
     },
     {
       "epoch": 3.7474120082815734,
-      "grad_norm": 8.028316497802734,
-      "learning_rate": 5.010351966873707e-06,
-      "loss": 0.5744,
+      "grad_norm": 5.730542182922363,
+      "learning_rate": 7.51552795031056e-06,
+      "loss": 0.5346,
       "step": 19910
     },
     {
       "epoch": 3.749294184076793,
-      "grad_norm": 16.884626388549805,
-      "learning_rate": 5.00282326369283e-06,
-      "loss": 0.5241,
+      "grad_norm": 15.303305625915527,
+      "learning_rate": 7.504234895539244e-06,
+      "loss": 0.5668,
       "step": 19920
     },
     {
       "epoch": 3.7511763598720123,
-      "grad_norm": 17.39742660522461,
-      "learning_rate": 4.995294560511953e-06,
-      "loss": 0.542,
+      "grad_norm": 7.135068893432617,
+      "learning_rate": 7.4929418407679286e-06,
+      "loss": 0.5419,
       "step": 19930
     },
     {
       "epoch": 3.7530585356672312,
-      "grad_norm": 13.098750114440918,
-      "learning_rate": 4.987765857331075e-06,
-      "loss": 0.7032,
+      "grad_norm": 18.504470825195312,
+      "learning_rate": 7.481648785996612e-06,
+      "loss": 0.6651,
       "step": 19940
     },
     {
       "epoch": 3.7549407114624507,
-      "grad_norm": 10.768354415893555,
-      "learning_rate": 4.9802371541501985e-06,
-      "loss": 0.7465,
+      "grad_norm": 4.049873352050781,
+      "learning_rate": 7.4703557312252965e-06,
+      "loss": 0.6697,
       "step": 19950
     },
     {
       "epoch": 3.7568228872576697,
-      "grad_norm": 0.5966710448265076,
-      "learning_rate": 4.9727084509693206e-06,
-      "loss": 0.4171,
+      "grad_norm": 0.3904266953468323,
+      "learning_rate": 7.459062676453981e-06,
+      "loss": 0.5437,
       "step": 19960
     },
     {
       "epoch": 3.758705063052889,
-      "grad_norm": 3.4089717864990234,
-      "learning_rate": 4.9651797477884435e-06,
-      "loss": 0.538,
+      "grad_norm": 2.0129761695861816,
+      "learning_rate": 7.447769621682665e-06,
+      "loss": 0.4477,
       "step": 19970
     },
     {
       "epoch": 3.7605872388481085,
-      "grad_norm": 35.42552185058594,
-      "learning_rate": 4.957651044607566e-06,
-      "loss": 0.6669,
+      "grad_norm": 25.443471908569336,
+      "learning_rate": 7.43647656691135e-06,
+      "loss": 0.6205,
       "step": 19980
     },
     {
       "epoch": 3.762469414643328,
-      "grad_norm": 24.851398468017578,
-      "learning_rate": 4.950122341426689e-06,
-      "loss": 0.4661,
+      "grad_norm": 27.52094841003418,
+      "learning_rate": 7.425183512140034e-06,
+      "loss": 0.4978,
       "step": 19990
     },
     {
       "epoch": 3.764351590438547,
-      "grad_norm": 0.36463797092437744,
-      "learning_rate": 4.942593638245812e-06,
-      "loss": 0.4581,
+      "grad_norm": 0.09147056192159653,
+      "learning_rate": 7.413890457368718e-06,
+      "loss": 0.398,
       "step": 20000
     },
     {
       "epoch": 3.7662337662337664,
-      "grad_norm": 3.793344736099243,
-      "learning_rate": 4.935064935064935e-06,
-      "loss": 0.4361,
+      "grad_norm": 0.8058493733406067,
+      "learning_rate": 7.402597402597403e-06,
+      "loss": 0.4703,
       "step": 20010
     },
     {
       "epoch": 3.7681159420289854,
-      "grad_norm": 12.976563453674316,
-      "learning_rate": 4.927536231884059e-06,
-      "loss": 0.6536,
+      "grad_norm": 8.946433067321777,
+      "learning_rate": 7.391304347826087e-06,
+      "loss": 0.6018,
       "step": 20020
     },
     {
       "epoch": 3.769998117824205,
-      "grad_norm": 33.84760284423828,
-      "learning_rate": 4.920007528703181e-06,
-      "loss": 0.6451,
+      "grad_norm": 12.180794715881348,
+      "learning_rate": 7.3800112930547715e-06,
+      "loss": 0.5685,
       "step": 20030
     },
     {
       "epoch": 3.7718802936194242,
-      "grad_norm": 9.5062255859375,
-      "learning_rate": 4.912478825522305e-06,
-      "loss": 0.6208,
+      "grad_norm": 11.476252555847168,
+      "learning_rate": 7.368718238283456e-06,
+      "loss": 0.5504,
       "step": 20040
     },
     {
       "epoch": 3.773762469414643,
-      "grad_norm": 16.10189437866211,
-      "learning_rate": 4.904950122341427e-06,
-      "loss": 0.6699,
+      "grad_norm": 14.593816757202148,
+      "learning_rate": 7.35742518351214e-06,
+      "loss": 0.5439,
       "step": 20050
     },
     {
       "epoch": 3.7756446452098626,
-      "grad_norm": 29.097278594970703,
-      "learning_rate": 4.89742141916055e-06,
-      "loss": 0.7841,
+      "grad_norm": 42.94921112060547,
+      "learning_rate": 7.346132128740825e-06,
+      "loss": 0.7765,
       "step": 20060
     },
     {
       "epoch": 3.7775268210050816,
-      "grad_norm": 12.631217002868652,
-      "learning_rate": 4.889892715979673e-06,
-      "loss": 0.5276,
+      "grad_norm": 8.45206069946289,
+      "learning_rate": 7.334839073969509e-06,
+      "loss": 0.4416,
       "step": 20070
     },
     {
       "epoch": 3.779408996800301,
-      "grad_norm": 10.902958869934082,
-      "learning_rate": 4.882364012798796e-06,
-      "loss": 0.7378,
+      "grad_norm": 11.394601821899414,
+      "learning_rate": 7.3235460191981934e-06,
+      "loss": 0.5376,
       "step": 20080
     },
     {
       "epoch": 3.7812911725955205,
-      "grad_norm": 1.129839301109314,
-      "learning_rate": 4.8748353096179186e-06,
-      "loss": 0.5461,
+      "grad_norm": 0.5047250986099243,
+      "learning_rate": 7.312252964426877e-06,
+      "loss": 0.5465,
       "step": 20090
     },
     {
       "epoch": 3.78317334839074,
-      "grad_norm": 8.831612586975098,
-      "learning_rate": 4.8673066064370415e-06,
-      "loss": 0.5889,
+      "grad_norm": 8.610776901245117,
+      "learning_rate": 7.300959909655562e-06,
+      "loss": 0.4937,
       "step": 20100
     },
     {
       "epoch": 3.785055524185959,
-      "grad_norm": 42.60455322265625,
-      "learning_rate": 4.859777903256164e-06,
-      "loss": 0.4435,
+      "grad_norm": 16.135894775390625,
+      "learning_rate": 7.289666854884246e-06,
+      "loss": 0.3992,
       "step": 20110
     },
     {
       "epoch": 3.7869376999811784,
-      "grad_norm": 24.456605911254883,
-      "learning_rate": 4.852249200075287e-06,
-      "loss": 0.7794,
+      "grad_norm": 10.323850631713867,
+      "learning_rate": 7.278373800112931e-06,
+      "loss": 0.6229,
       "step": 20120
     },
     {
       "epoch": 3.7888198757763973,
-      "grad_norm": 3.61232328414917,
-      "learning_rate": 4.84472049689441e-06,
-      "loss": 0.7502,
+      "grad_norm": 24.070560455322266,
+      "learning_rate": 7.267080745341615e-06,
+      "loss": 0.6724,
       "step": 20130
     },
     {
       "epoch": 3.7907020515716168,
-      "grad_norm": 16.118135452270508,
-      "learning_rate": 4.837191793713533e-06,
-      "loss": 0.5583,
+      "grad_norm": 15.11103343963623,
+      "learning_rate": 7.2557876905703e-06,
+      "loss": 0.4975,
       "step": 20140
     },
     {
       "epoch": 3.792584227366836,
-      "grad_norm": 25.525503158569336,
-      "learning_rate": 4.829663090532656e-06,
-      "loss": 0.8087,
+      "grad_norm": 14.725980758666992,
+      "learning_rate": 7.244494635798984e-06,
+      "loss": 0.65,
       "step": 20150
     },
     {
       "epoch": 3.794466403162055,
-      "grad_norm": 22.253944396972656,
-      "learning_rate": 4.822134387351779e-06,
-      "loss": 0.3545,
+      "grad_norm": 13.276191711425781,
+      "learning_rate": 7.233201581027668e-06,
+      "loss": 0.2699,
       "step": 20160
     },
     {
       "epoch": 3.7963485789572746,
-      "grad_norm": 21.64002799987793,
-      "learning_rate": 4.814605684170902e-06,
-      "loss": 0.7379,
+      "grad_norm": 16.452234268188477,
+      "learning_rate": 7.221908526256353e-06,
+      "loss": 0.7108,
       "step": 20170
     },
     {
       "epoch": 3.7982307547524936,
-      "grad_norm": 19.44178581237793,
-      "learning_rate": 4.807076980990025e-06,
-      "loss": 0.4395,
+      "grad_norm": 18.935544967651367,
+      "learning_rate": 7.210615471485036e-06,
+      "loss": 0.4207,
       "step": 20180
     },
     {
       "epoch": 3.800112930547713,
-      "grad_norm": 13.115379333496094,
-      "learning_rate": 4.799548277809148e-06,
-      "loss": 0.6557,
+      "grad_norm": 14.351372718811035,
+      "learning_rate": 7.199322416713722e-06,
+      "loss": 0.554,
       "step": 20190
     },
     {
       "epoch": 3.8019951063429325,
-      "grad_norm": 4.139683246612549,
-      "learning_rate": 4.792019574628271e-06,
-      "loss": 0.5407,
+      "grad_norm": 2.5707204341888428,
+      "learning_rate": 7.188029361942405e-06,
+      "loss": 0.392,
       "step": 20200
     },
     {
       "epoch": 3.803877282138152,
-      "grad_norm": 0.6734669208526611,
-      "learning_rate": 4.784490871447394e-06,
-      "loss": 0.4562,
+      "grad_norm": 1.579572081565857,
+      "learning_rate": 7.17673630717109e-06,
+      "loss": 0.3267,
       "step": 20210
     },
     {
       "epoch": 3.805759457933371,
-      "grad_norm": 18.29108238220215,
-      "learning_rate": 4.7769621682665165e-06,
-      "loss": 0.6391,
+      "grad_norm": 4.0918660163879395,
+      "learning_rate": 7.165443252399774e-06,
+      "loss": 0.6002,
       "step": 20220
     },
     {
       "epoch": 3.8076416337285903,
-      "grad_norm": 20.505464553833008,
-      "learning_rate": 4.7694334650856395e-06,
-      "loss": 0.4091,
+      "grad_norm": 16.16851043701172,
+      "learning_rate": 7.154150197628458e-06,
+      "loss": 0.3712,
       "step": 20230
     },
     {
       "epoch": 3.8095238095238093,
-      "grad_norm": 15.422745704650879,
-      "learning_rate": 4.761904761904762e-06,
-      "loss": 0.9266,
+      "grad_norm": 11.626598358154297,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.8826,
       "step": 20240
     },
     {
       "epoch": 3.8114059853190287,
-      "grad_norm": 21.413530349731445,
-      "learning_rate": 4.754376058723885e-06,
-      "loss": 0.9884,
+      "grad_norm": 17.99329948425293,
+      "learning_rate": 7.131564088085827e-06,
+      "loss": 0.8901,
       "step": 20250
     },
     {
       "epoch": 3.813288161114248,
-      "grad_norm": 6.468101501464844,
-      "learning_rate": 4.746847355543008e-06,
-      "loss": 0.8135,
+      "grad_norm": 0.3546718955039978,
+      "learning_rate": 7.120271033314512e-06,
+      "loss": 0.7524,
       "step": 20260
     },
     {
       "epoch": 3.8151703369094676,
-      "grad_norm": 9.80642032623291,
-      "learning_rate": 4.739318652362131e-06,
-      "loss": 0.3934,
+      "grad_norm": 6.967835426330566,
+      "learning_rate": 7.108977978543196e-06,
+      "loss": 0.3611,
       "step": 20270
     },
     {
       "epoch": 3.8170525127046866,
-      "grad_norm": 8.44393253326416,
-      "learning_rate": 4.731789949181254e-06,
-      "loss": 0.643,
+      "grad_norm": 0.7506970763206482,
+      "learning_rate": 7.097684923771881e-06,
+      "loss": 0.6239,
       "step": 20280
     },
     {
       "epoch": 3.818934688499906,
-      "grad_norm": 4.562558174133301,
-      "learning_rate": 4.724261246000377e-06,
-      "loss": 0.6693,
+      "grad_norm": 7.021599769592285,
+      "learning_rate": 7.086391869000565e-06,
+      "loss": 0.8331,
       "step": 20290
     },
     {
       "epoch": 3.820816864295125,
-      "grad_norm": 5.316037654876709,
-      "learning_rate": 4.7167325428195e-06,
-      "loss": 0.6138,
+      "grad_norm": 0.8781418800354004,
+      "learning_rate": 7.075098814229249e-06,
+      "loss": 0.5364,
       "step": 20300
     },
     {
       "epoch": 3.8226990400903444,
-      "grad_norm": 10.8194580078125,
-      "learning_rate": 4.709203839638623e-06,
-      "loss": 0.3942,
+      "grad_norm": 10.943838119506836,
+      "learning_rate": 7.063805759457933e-06,
+      "loss": 0.3288,
       "step": 20310
     },
     {
       "epoch": 3.824581215885564,
-      "grad_norm": 22.41046142578125,
-      "learning_rate": 4.701675136457746e-06,
-      "loss": 0.299,
+      "grad_norm": 6.258754730224609,
+      "learning_rate": 7.052512704686618e-06,
+      "loss": 0.2297,
       "step": 20320
     },
     {
       "epoch": 3.826463391680783,
-      "grad_norm": 6.464774131774902,
-      "learning_rate": 4.694146433276869e-06,
-      "loss": 0.4668,
+      "grad_norm": 6.189356803894043,
+      "learning_rate": 7.041219649915302e-06,
+      "loss": 0.4229,
       "step": 20330
     },
     {
       "epoch": 3.8283455674760023,
-      "grad_norm": 0.8023673892021179,
-      "learning_rate": 4.686617730095992e-06,
-      "loss": 0.8259,
+      "grad_norm": 0.3370305299758911,
+      "learning_rate": 7.0299265951439865e-06,
+      "loss": 0.8312,
       "step": 20340
     },
     {
       "epoch": 3.8302277432712213,
-      "grad_norm": 0.310139536857605,
-      "learning_rate": 4.679089026915114e-06,
-      "loss": 0.6507,
+      "grad_norm": 0.39705947041511536,
+      "learning_rate": 7.018633540372671e-06,
+      "loss": 0.6544,
       "step": 20350
     },
     {
       "epoch": 3.8321099190664407,
-      "grad_norm": 21.757549285888672,
-      "learning_rate": 4.6715603237342374e-06,
-      "loss": 0.6305,
+      "grad_norm": 5.138480186462402,
+      "learning_rate": 7.007340485601355e-06,
+      "loss": 0.5875,
       "step": 20360
     },
     {
       "epoch": 3.83399209486166,
-      "grad_norm": 23.21905517578125,
-      "learning_rate": 4.66403162055336e-06,
-      "loss": 0.7793,
+      "grad_norm": 16.162595748901367,
+      "learning_rate": 6.9960474308300405e-06,
+      "loss": 0.694,
       "step": 20370
     },
     {
       "epoch": 3.8358742706568796,
-      "grad_norm": 16.708805084228516,
-      "learning_rate": 4.656502917372483e-06,
-      "loss": 0.7145,
+      "grad_norm": 10.226083755493164,
+      "learning_rate": 6.984754376058724e-06,
+      "loss": 0.6075,
       "step": 20380
     },
     {
       "epoch": 3.8377564464520986,
-      "grad_norm": 2.589465856552124,
-      "learning_rate": 4.648974214191606e-06,
-      "loss": 0.5476,
+      "grad_norm": 11.182906150817871,
+      "learning_rate": 6.9734613212874084e-06,
+      "loss": 0.5489,
       "step": 20390
     },
     {
       "epoch": 3.839638622247318,
-      "grad_norm": 21.747697830200195,
-      "learning_rate": 4.641445511010729e-06,
-      "loss": 0.5945,
+      "grad_norm": 16.98402214050293,
+      "learning_rate": 6.962168266516093e-06,
+      "loss": 0.5389,
       "step": 20400
     },
     {
       "epoch": 3.841520798042537,
-      "grad_norm": 10.856985092163086,
-      "learning_rate": 4.633916807829852e-06,
-      "loss": 0.524,
+      "grad_norm": 7.632681846618652,
+      "learning_rate": 6.950875211744777e-06,
+      "loss": 0.4979,
       "step": 20410
     },
     {
       "epoch": 3.8434029738377564,
-      "grad_norm": 19.395648956298828,
-      "learning_rate": 4.626388104648974e-06,
-      "loss": 0.743,
+      "grad_norm": 11.018415451049805,
+      "learning_rate": 6.939582156973462e-06,
+      "loss": 0.792,
       "step": 20420
     },
     {
       "epoch": 3.845285149632976,
-      "grad_norm": 26.709550857543945,
-      "learning_rate": 4.618859401468098e-06,
-      "loss": 0.4844,
+      "grad_norm": 18.204879760742188,
+      "learning_rate": 6.928289102202146e-06,
+      "loss": 0.4916,
       "step": 20430
     },
     {
       "epoch": 3.847167325428195,
-      "grad_norm": 19.377134323120117,
-      "learning_rate": 4.61133069828722e-06,
-      "loss": 0.3497,
+      "grad_norm": 17.605222702026367,
+      "learning_rate": 6.91699604743083e-06,
+      "loss": 0.2598,
       "step": 20440
     },
     {
       "epoch": 3.8490495012234143,
-      "grad_norm": 14.577184677124023,
-      "learning_rate": 4.603801995106344e-06,
-      "loss": 0.7667,
+      "grad_norm": 31.94579315185547,
+      "learning_rate": 6.905702992659515e-06,
+      "loss": 0.843,
       "step": 20450
     },
     {
       "epoch": 3.8509316770186337,
-      "grad_norm": 11.72651481628418,
-      "learning_rate": 4.596273291925466e-06,
-      "loss": 0.7216,
+      "grad_norm": 16.11570167541504,
+      "learning_rate": 6.894409937888198e-06,
+      "loss": 0.5118,
       "step": 20460
     },
     {
       "epoch": 3.8528138528138527,
-      "grad_norm": 9.256969451904297,
-      "learning_rate": 4.5887445887445896e-06,
-      "loss": 0.6854,
+      "grad_norm": 9.326797485351562,
+      "learning_rate": 6.8831168831168835e-06,
+      "loss": 0.5509,
       "step": 20470
     },
     {
       "epoch": 3.854696028609072,
-      "grad_norm": 13.852824211120605,
-      "learning_rate": 4.581215885563712e-06,
-      "loss": 0.6446,
+      "grad_norm": 12.51727294921875,
+      "learning_rate": 6.871823828345567e-06,
+      "loss": 0.5935,
       "step": 20480
     },
     {
       "epoch": 3.8565782044042916,
-      "grad_norm": 13.739072799682617,
-      "learning_rate": 4.5736871823828346e-06,
-      "loss": 0.7871,
+      "grad_norm": 18.62541389465332,
+      "learning_rate": 6.860530773574252e-06,
+      "loss": 0.7294,
       "step": 20490
     },
     {
       "epoch": 3.8584603801995105,
-      "grad_norm": 22.21795654296875,
-      "learning_rate": 4.566158479201958e-06,
-      "loss": 0.5034,
+      "grad_norm": 18.53264045715332,
+      "learning_rate": 6.849237718802937e-06,
+      "loss": 0.4357,
       "step": 20500
     },
     {
       "epoch": 3.86034255599473,
-      "grad_norm": 18.042375564575195,
-      "learning_rate": 4.55862977602108e-06,
-      "loss": 0.5784,
+      "grad_norm": 22.299819946289062,
+      "learning_rate": 6.837944664031621e-06,
+      "loss": 0.5943,
       "step": 20510
     },
     {
       "epoch": 3.862224731789949,
-      "grad_norm": 16.496652603149414,
-      "learning_rate": 4.551101072840204e-06,
-      "loss": 0.7036,
+      "grad_norm": 23.825881958007812,
+      "learning_rate": 6.826651609260305e-06,
+      "loss": 0.5521,
       "step": 20520
     },
     {
       "epoch": 3.8641069075851684,
-      "grad_norm": 23.149200439453125,
-      "learning_rate": 4.543572369659326e-06,
-      "loss": 0.7427,
+      "grad_norm": 20.267616271972656,
+      "learning_rate": 6.815358554488989e-06,
+      "loss": 0.8127,
       "step": 20530
     },
     {
       "epoch": 3.865989083380388,
-      "grad_norm": 25.870445251464844,
-      "learning_rate": 4.53604366647845e-06,
-      "loss": 0.6495,
+      "grad_norm": 29.187458038330078,
+      "learning_rate": 6.804065499717674e-06,
+      "loss": 0.5979,
       "step": 20540
     },
     {
       "epoch": 3.8678712591756073,
-      "grad_norm": 5.6248674392700195,
-      "learning_rate": 4.528514963297572e-06,
-      "loss": 0.6211,
+      "grad_norm": 15.423782348632812,
+      "learning_rate": 6.792772444946358e-06,
+      "loss": 0.6075,
       "step": 20550
     },
     {
       "epoch": 3.8697534349708262,
-      "grad_norm": 3.3882083892822266,
-      "learning_rate": 4.520986260116695e-06,
-      "loss": 0.9215,
+      "grad_norm": 8.876800537109375,
+      "learning_rate": 6.781479390175043e-06,
+      "loss": 0.8203,
       "step": 20560
     },
     {
       "epoch": 3.8716356107660457,
-      "grad_norm": 2.2890217304229736,
-      "learning_rate": 4.513457556935818e-06,
-      "loss": 0.6994,
+      "grad_norm": 0.373295396566391,
+      "learning_rate": 6.7701863354037265e-06,
+      "loss": 0.7442,
       "step": 20570
     },
     {
       "epoch": 3.8735177865612647,
-      "grad_norm": 1.9337209463119507,
-      "learning_rate": 4.505928853754941e-06,
-      "loss": 0.3197,
+      "grad_norm": 3.6619198322296143,
+      "learning_rate": 6.758893280632412e-06,
+      "loss": 0.2888,
       "step": 20580
     },
     {
       "epoch": 3.875399962356484,
-      "grad_norm": 12.150304794311523,
-      "learning_rate": 4.498400150574064e-06,
-      "loss": 0.616,
+      "grad_norm": 7.333616256713867,
+      "learning_rate": 6.747600225861095e-06,
+      "loss": 0.5505,
       "step": 20590
     },
     {
       "epoch": 3.8772821381517035,
-      "grad_norm": 13.6411771774292,
-      "learning_rate": 4.490871447393187e-06,
-      "loss": 0.5819,
+      "grad_norm": 11.97646427154541,
+      "learning_rate": 6.73630717108978e-06,
+      "loss": 0.5715,
       "step": 20600
     },
     {
       "epoch": 3.8791643139469225,
-      "grad_norm": 18.936782836914062,
-      "learning_rate": 4.48334274421231e-06,
-      "loss": 0.7718,
+      "grad_norm": 17.42437744140625,
+      "learning_rate": 6.725014116318464e-06,
+      "loss": 0.7322,
       "step": 20610
     },
     {
       "epoch": 3.881046489742142,
-      "grad_norm": 29.339204788208008,
-      "learning_rate": 4.4758140410314325e-06,
-      "loss": 0.6563,
+      "grad_norm": 4.155119895935059,
+      "learning_rate": 6.713721061547148e-06,
+      "loss": 0.5882,
       "step": 20620
     },
     {
       "epoch": 3.882928665537361,
-      "grad_norm": 19.04207420349121,
-      "learning_rate": 4.4682853378505555e-06,
-      "loss": 0.8119,
+      "grad_norm": 20.934206008911133,
+      "learning_rate": 6.702428006775834e-06,
+      "loss": 0.7397,
       "step": 20630
     },
     {
       "epoch": 3.8848108413325804,
-      "grad_norm": 11.899054527282715,
-      "learning_rate": 4.460756634669678e-06,
-      "loss": 0.47,
+      "grad_norm": 12.857183456420898,
+      "learning_rate": 6.691134952004517e-06,
+      "loss": 0.4704,
       "step": 20640
     },
     {
       "epoch": 3.8866930171278,
-      "grad_norm": 5.735147953033447,
-      "learning_rate": 4.453227931488801e-06,
-      "loss": 0.4137,
+      "grad_norm": 8.88806438446045,
+      "learning_rate": 6.679841897233202e-06,
+      "loss": 0.3095,
       "step": 20650
     },
     {
       "epoch": 3.8885751929230192,
-      "grad_norm": 7.638981342315674,
-      "learning_rate": 4.445699228307924e-06,
-      "loss": 0.5822,
+      "grad_norm": 11.002625465393066,
+      "learning_rate": 6.668548842461886e-06,
+      "loss": 0.5045,
       "step": 20660
     },
     {
       "epoch": 3.890457368718238,
-      "grad_norm": 22.939064025878906,
-      "learning_rate": 4.438170525127047e-06,
-      "loss": 0.8084,
+      "grad_norm": 7.77680778503418,
+      "learning_rate": 6.65725578769057e-06,
+      "loss": 0.8244,
       "step": 20670
     },
     {
       "epoch": 3.8923395445134576,
-      "grad_norm": 8.393478393554688,
-      "learning_rate": 4.43064182194617e-06,
-      "loss": 0.5258,
+      "grad_norm": 0.3390113115310669,
+      "learning_rate": 6.645962732919255e-06,
+      "loss": 0.5562,
       "step": 20680
     },
     {
       "epoch": 3.8942217203086766,
-      "grad_norm": 17.132465362548828,
-      "learning_rate": 4.423113118765293e-06,
-      "loss": 0.5562,
+      "grad_norm": 19.375041961669922,
+      "learning_rate": 6.634669678147939e-06,
+      "loss": 0.4,
       "step": 20690
     },
     {
       "epoch": 3.896103896103896,
-      "grad_norm": 28.829774856567383,
-      "learning_rate": 4.415584415584416e-06,
-      "loss": 0.5178,
+      "grad_norm": 47.01906967163086,
+      "learning_rate": 6.6233766233766234e-06,
+      "loss": 0.4626,
       "step": 20700
     },
     {
       "epoch": 3.8979860718991155,
-      "grad_norm": 20.98097038269043,
-      "learning_rate": 4.408055712403539e-06,
-      "loss": 0.7207,
+      "grad_norm": 12.001082420349121,
+      "learning_rate": 6.612083568605308e-06,
+      "loss": 0.7142,
       "step": 20710
     },
     {
       "epoch": 3.8998682476943345,
-      "grad_norm": 18.7142333984375,
-      "learning_rate": 4.400527009222662e-06,
-      "loss": 0.4549,
+      "grad_norm": 12.084139823913574,
+      "learning_rate": 6.600790513833992e-06,
+      "loss": 0.4786,
       "step": 20720
     },
     {
       "epoch": 3.901750423489554,
-      "grad_norm": 6.946071147918701,
-      "learning_rate": 4.392998306041785e-06,
-      "loss": 0.4556,
+      "grad_norm": 10.616601943969727,
+      "learning_rate": 6.589497459062677e-06,
+      "loss": 0.4461,
       "step": 20730
     },
     {
       "epoch": 3.9036325992847734,
-      "grad_norm": 21.66033363342285,
-      "learning_rate": 4.385469602860908e-06,
-      "loss": 0.8115,
+      "grad_norm": 15.530953407287598,
+      "learning_rate": 6.578204404291361e-06,
+      "loss": 0.6108,
       "step": 20740
     },
     {
       "epoch": 3.9055147750799923,
-      "grad_norm": 9.17212963104248,
-      "learning_rate": 4.3779408996800305e-06,
-      "loss": 0.5632,
+      "grad_norm": 0.3299574851989746,
+      "learning_rate": 6.566911349520045e-06,
+      "loss": 0.5048,
       "step": 20750
     },
     {
       "epoch": 3.9073969508752118,
-      "grad_norm": 1.5829265117645264,
-      "learning_rate": 4.3704121964991534e-06,
-      "loss": 0.4644,
+      "grad_norm": 0.755125880241394,
+      "learning_rate": 6.55561829474873e-06,
+      "loss": 0.3229,
       "step": 20760
     },
     {
       "epoch": 3.909279126670431,
-      "grad_norm": 26.870147705078125,
-      "learning_rate": 4.362883493318276e-06,
-      "loss": 0.4721,
+      "grad_norm": 0.48004603385925293,
+      "learning_rate": 6.544325239977414e-06,
+      "loss": 0.4146,
       "step": 20770
     },
     {
       "epoch": 3.91116130246565,
-      "grad_norm": 14.611478805541992,
-      "learning_rate": 4.355354790137399e-06,
-      "loss": 0.4564,
+      "grad_norm": 33.05055236816406,
+      "learning_rate": 6.5330321852060985e-06,
+      "loss": 0.4969,
       "step": 20780
     },
     {
       "epoch": 3.9130434782608696,
-      "grad_norm": 24.822168350219727,
-      "learning_rate": 4.347826086956522e-06,
-      "loss": 1.0165,
+      "grad_norm": 2.9246139526367188,
+      "learning_rate": 6.521739130434783e-06,
+      "loss": 1.0341,
       "step": 20790
     },
     {
       "epoch": 3.9149256540560886,
-      "grad_norm": 10.384610176086426,
-      "learning_rate": 4.340297383775645e-06,
-      "loss": 0.5887,
+      "grad_norm": 8.560576438903809,
+      "learning_rate": 6.510446075663467e-06,
+      "loss": 0.4784,
       "step": 20800
     },
     {
       "epoch": 3.916807829851308,
-      "grad_norm": 6.775899887084961,
-      "learning_rate": 4.332768680594768e-06,
-      "loss": 0.4609,
+      "grad_norm": 14.475516319274902,
+      "learning_rate": 6.499153020892152e-06,
+      "loss": 0.422,
       "step": 20810
     },
     {
       "epoch": 3.9186900056465275,
-      "grad_norm": 22.89283561706543,
-      "learning_rate": 4.325239977413891e-06,
-      "loss": 0.8258,
+      "grad_norm": 23.599973678588867,
+      "learning_rate": 6.487859966120836e-06,
+      "loss": 0.7579,
       "step": 20820
     },
     {
       "epoch": 3.920572181441747,
-      "grad_norm": 17.177522659301758,
-      "learning_rate": 4.317711274233014e-06,
-      "loss": 0.5948,
+      "grad_norm": 21.075658798217773,
+      "learning_rate": 6.4765669113495196e-06,
+      "loss": 0.5062,
       "step": 20830
     },
     {
       "epoch": 3.922454357236966,
-      "grad_norm": 9.44758415222168,
-      "learning_rate": 4.310182571052137e-06,
-      "loss": 0.7599,
+      "grad_norm": 11.33735466003418,
+      "learning_rate": 6.465273856578205e-06,
+      "loss": 0.7134,
       "step": 20840
     },
     {
       "epoch": 3.9243365330321853,
-      "grad_norm": 9.103228569030762,
-      "learning_rate": 4.302653867871259e-06,
-      "loss": 0.5208,
+      "grad_norm": 13.300880432128906,
+      "learning_rate": 6.453980801806888e-06,
+      "loss": 0.4887,
       "step": 20850
     },
     {
       "epoch": 3.9262187088274043,
-      "grad_norm": 15.446986198425293,
-      "learning_rate": 4.295125164690383e-06,
-      "loss": 0.4422,
+      "grad_norm": 16.807415008544922,
+      "learning_rate": 6.4426877470355736e-06,
+      "loss": 0.3613,
       "step": 20860
     },
     {
       "epoch": 3.9281008846226237,
-      "grad_norm": 15.556666374206543,
-      "learning_rate": 4.287596461509506e-06,
-      "loss": 0.5821,
+      "grad_norm": 10.84011459350586,
+      "learning_rate": 6.431394692264258e-06,
+      "loss": 0.5069,
       "step": 20870
     },
     {
       "epoch": 3.929983060417843,
-      "grad_norm": 25.277069091796875,
-      "learning_rate": 4.2800677583286285e-06,
-      "loss": 0.6588,
+      "grad_norm": 21.17901611328125,
+      "learning_rate": 6.420101637492942e-06,
+      "loss": 0.6276,
       "step": 20880
     },
     {
       "epoch": 3.931865236213062,
-      "grad_norm": 18.088552474975586,
-      "learning_rate": 4.272539055147751e-06,
-      "loss": 0.5822,
+      "grad_norm": 12.811206817626953,
+      "learning_rate": 6.408808582721627e-06,
+      "loss": 0.5739,
       "step": 20890
     },
     {
       "epoch": 3.9337474120082816,
-      "grad_norm": 3.4164326190948486,
-      "learning_rate": 4.265010351966874e-06,
-      "loss": 0.2988,
+      "grad_norm": 2.8694605827331543,
+      "learning_rate": 6.39751552795031e-06,
+      "loss": 0.3009,
       "step": 20900
     },
     {
       "epoch": 3.9356295878035006,
-      "grad_norm": 24.451601028442383,
-      "learning_rate": 4.257481648785997e-06,
-      "loss": 0.5764,
+      "grad_norm": 10.879847526550293,
+      "learning_rate": 6.3862224731789955e-06,
+      "loss": 0.5047,
       "step": 20910
     },
     {
       "epoch": 3.93751176359872,
-      "grad_norm": 10.956971168518066,
-      "learning_rate": 4.249952945605119e-06,
-      "loss": 0.7125,
+      "grad_norm": 12.21065616607666,
+      "learning_rate": 6.374929418407679e-06,
+      "loss": 0.6257,
       "step": 20920
     },
     {
       "epoch": 3.9393939393939394,
-      "grad_norm": 11.39134407043457,
-      "learning_rate": 4.242424242424243e-06,
-      "loss": 0.7606,
+      "grad_norm": 12.078612327575684,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 0.7487,
       "step": 20930
     },
     {
       "epoch": 3.941276115189159,
-      "grad_norm": 7.814131736755371,
-      "learning_rate": 4.234895539243365e-06,
-      "loss": 0.4197,
+      "grad_norm": 7.435342311859131,
+      "learning_rate": 6.352343308865048e-06,
+      "loss": 0.4637,
       "step": 20940
     },
     {
       "epoch": 3.943158290984378,
-      "grad_norm": 12.360709190368652,
-      "learning_rate": 4.227366836062489e-06,
-      "loss": 0.6645,
+      "grad_norm": 16.511627197265625,
+      "learning_rate": 6.341050254093733e-06,
+      "loss": 0.6262,
       "step": 20950
     },
     {
       "epoch": 3.9450404667795973,
-      "grad_norm": 2.4350345134735107,
-      "learning_rate": 4.219838132881611e-06,
-      "loss": 0.6532,
+      "grad_norm": 0.3107624053955078,
+      "learning_rate": 6.3297571993224165e-06,
+      "loss": 0.5074,
       "step": 20960
     },
     {
       "epoch": 3.9469226425748163,
-      "grad_norm": 0.28753194212913513,
-      "learning_rate": 4.212309429700735e-06,
-      "loss": 0.9288,
+      "grad_norm": 0.1293262541294098,
+      "learning_rate": 6.318464144551101e-06,
+      "loss": 0.7009,
       "step": 20970
     },
     {
       "epoch": 3.9488048183700357,
-      "grad_norm": 19.667470932006836,
-      "learning_rate": 4.204780726519857e-06,
-      "loss": 1.0559,
+      "grad_norm": 16.236751556396484,
+      "learning_rate": 6.307171089779785e-06,
+      "loss": 0.9354,
       "step": 20980
     },
     {
       "epoch": 3.950686994165255,
-      "grad_norm": 4.4203081130981445,
-      "learning_rate": 4.19725202333898e-06,
-      "loss": 0.7958,
+      "grad_norm": 4.352895259857178,
+      "learning_rate": 6.29587803500847e-06,
+      "loss": 0.7928,
       "step": 20990
     },
     {
       "epoch": 3.9525691699604746,
-      "grad_norm": 12.80313491821289,
-      "learning_rate": 4.1897233201581036e-06,
-      "loss": 0.4913,
+      "grad_norm": 7.811060428619385,
+      "learning_rate": 6.284584980237155e-06,
+      "loss": 0.384,
       "step": 21000
     },
     {
       "epoch": 3.9544513457556936,
-      "grad_norm": 27.15594482421875,
-      "learning_rate": 4.182194616977226e-06,
-      "loss": 0.5069,
+      "grad_norm": 30.867467880249023,
+      "learning_rate": 6.2732919254658384e-06,
+      "loss": 0.4535,
       "step": 21010
     },
     {
       "epoch": 3.956333521550913,
-      "grad_norm": 18.540098190307617,
-      "learning_rate": 4.174665913796349e-06,
-      "loss": 0.6035,
+      "grad_norm": 9.880651473999023,
+      "learning_rate": 6.261998870694524e-06,
+      "loss": 0.4897,
       "step": 21020
     },
     {
       "epoch": 3.958215697346132,
-      "grad_norm": 22.570690155029297,
-      "learning_rate": 4.1671372106154715e-06,
-      "loss": 0.675,
+      "grad_norm": 26.8701114654541,
+      "learning_rate": 6.250705815923207e-06,
+      "loss": 0.5893,
       "step": 21030
     },
     {
       "epoch": 3.9600978731413514,
-      "grad_norm": 23.07613182067871,
-      "learning_rate": 4.159608507434595e-06,
-      "loss": 0.5909,
+      "grad_norm": 21.494901657104492,
+      "learning_rate": 6.239412761151892e-06,
+      "loss": 0.5116,
       "step": 21040
     },
     {
       "epoch": 3.961980048936571,
-      "grad_norm": 12.728482246398926,
-      "learning_rate": 4.152079804253717e-06,
-      "loss": 0.7034,
+      "grad_norm": 13.416314125061035,
+      "learning_rate": 6.228119706380576e-06,
+      "loss": 0.6689,
       "step": 21050
     },
     {
       "epoch": 3.96386222473179,
-      "grad_norm": 30.579265594482422,
-      "learning_rate": 4.14455110107284e-06,
-      "loss": 0.5167,
+      "grad_norm": 7.51889705657959,
+      "learning_rate": 6.21682665160926e-06,
+      "loss": 0.4839,
       "step": 21060
     },
     {
       "epoch": 3.9657444005270093,
-      "grad_norm": 17.927875518798828,
-      "learning_rate": 4.137022397891963e-06,
-      "loss": 0.6327,
+      "grad_norm": 12.946698188781738,
+      "learning_rate": 6.205533596837945e-06,
+      "loss": 0.6613,
       "step": 21070
     },
     {
       "epoch": 3.9676265763222283,
-      "grad_norm": 14.641374588012695,
-      "learning_rate": 4.129493694711086e-06,
-      "loss": 0.7122,
+      "grad_norm": 18.94002342224121,
+      "learning_rate": 6.194240542066629e-06,
+      "loss": 0.607,
       "step": 21080
     },
     {
       "epoch": 3.9695087521174477,
-      "grad_norm": 16.535053253173828,
-      "learning_rate": 4.121964991530209e-06,
-      "loss": 0.8122,
+      "grad_norm": 13.655399322509766,
+      "learning_rate": 6.1829474872953135e-06,
+      "loss": 0.6801,
       "step": 21090
     },
     {
       "epoch": 3.971390927912667,
-      "grad_norm": 21.376726150512695,
-      "learning_rate": 4.114436288349332e-06,
-      "loss": 0.5726,
+      "grad_norm": 12.325647354125977,
+      "learning_rate": 6.171654432523998e-06,
+      "loss": 0.6112,
       "step": 21100
     },
     {
       "epoch": 3.9732731037078866,
-      "grad_norm": 14.665657997131348,
-      "learning_rate": 4.106907585168456e-06,
-      "loss": 0.7047,
+      "grad_norm": 6.132079601287842,
+      "learning_rate": 6.160361377752682e-06,
+      "loss": 0.6071,
       "step": 21110
     },
     {
       "epoch": 3.9751552795031055,
-      "grad_norm": 0.8747822046279907,
-      "learning_rate": 4.099378881987578e-06,
-      "loss": 0.5733,
+      "grad_norm": 0.7968364357948303,
+      "learning_rate": 6.149068322981367e-06,
+      "loss": 0.5114,
       "step": 21120
     },
     {
       "epoch": 3.977037455298325,
-      "grad_norm": 7.762773513793945,
-      "learning_rate": 4.091850178806701e-06,
-      "loss": 0.7884,
+      "grad_norm": 5.779326438903809,
+      "learning_rate": 6.137775268210051e-06,
+      "loss": 0.6626,
       "step": 21130
     },
     {
       "epoch": 3.978919631093544,
-      "grad_norm": 15.392909049987793,
-      "learning_rate": 4.084321475625824e-06,
-      "loss": 0.6718,
+      "grad_norm": 15.42678451538086,
+      "learning_rate": 6.126482213438735e-06,
+      "loss": 0.6293,
       "step": 21140
     },
     {
       "epoch": 3.9808018068887634,
-      "grad_norm": 34.64730453491211,
-      "learning_rate": 4.0767927724449465e-06,
-      "loss": 0.9371,
+      "grad_norm": 23.818042755126953,
+      "learning_rate": 6.11518915866742e-06,
+      "loss": 0.7765,
       "step": 21150
     },
     {
       "epoch": 3.982683982683983,
-      "grad_norm": 5.796759605407715,
-      "learning_rate": 4.0692640692640695e-06,
-      "loss": 0.7274,
+      "grad_norm": 15.091184616088867,
+      "learning_rate": 6.103896103896104e-06,
+      "loss": 0.6724,
       "step": 21160
     },
     {
       "epoch": 3.984566158479202,
-      "grad_norm": 17.814022064208984,
-      "learning_rate": 4.061735366083192e-06,
-      "loss": 0.6687,
+      "grad_norm": 13.259024620056152,
+      "learning_rate": 6.0926030491247886e-06,
+      "loss": 0.6242,
       "step": 21170
     },
     {
       "epoch": 3.9864483342744212,
-      "grad_norm": 9.838661193847656,
-      "learning_rate": 4.054206662902315e-06,
-      "loss": 0.8333,
+      "grad_norm": 10.13891887664795,
+      "learning_rate": 6.081309994353472e-06,
+      "loss": 0.807,
       "step": 21180
     },
     {
       "epoch": 3.9883305100696402,
-      "grad_norm": 9.50422191619873,
-      "learning_rate": 4.046677959721438e-06,
-      "loss": 0.498,
+      "grad_norm": 10.566964149475098,
+      "learning_rate": 6.070016939582157e-06,
+      "loss": 0.4579,
       "step": 21190
     },
     {
       "epoch": 3.9902126858648597,
-      "grad_norm": 24.345027923583984,
-      "learning_rate": 4.039149256540561e-06,
-      "loss": 0.5806,
+      "grad_norm": 9.14899730682373,
+      "learning_rate": 6.058723884810841e-06,
+      "loss": 0.5235,
       "step": 21200
     },
     {
       "epoch": 3.992094861660079,
-      "grad_norm": 17.970151901245117,
-      "learning_rate": 4.031620553359684e-06,
-      "loss": 0.4809,
+      "grad_norm": 2.9169421195983887,
+      "learning_rate": 6.047430830039526e-06,
+      "loss": 0.3648,
       "step": 21210
     },
     {
       "epoch": 3.9939770374552985,
-      "grad_norm": 6.761532783508301,
-      "learning_rate": 4.024091850178807e-06,
-      "loss": 0.546,
+      "grad_norm": 14.589804649353027,
+      "learning_rate": 6.03613777526821e-06,
+      "loss": 0.4533,
       "step": 21220
     },
     {
       "epoch": 3.9958592132505175,
-      "grad_norm": 25.05278778076172,
-      "learning_rate": 4.01656314699793e-06,
-      "loss": 0.5847,
+      "grad_norm": 24.296232223510742,
+      "learning_rate": 6.024844720496895e-06,
+      "loss": 0.5101,
       "step": 21230
     },
     {
       "epoch": 3.997741389045737,
-      "grad_norm": 25.217716217041016,
-      "learning_rate": 4.009034443817053e-06,
-      "loss": 0.7152,
+      "grad_norm": 33.149715423583984,
+      "learning_rate": 6.013551665725579e-06,
+      "loss": 0.6524,
       "step": 21240
     },
     {
       "epoch": 3.999623564840956,
-      "grad_norm": 13.365795135498047,
-      "learning_rate": 4.001505740636176e-06,
-      "loss": 0.3327,
+      "grad_norm": 15.509395599365234,
+      "learning_rate": 6.002258610954264e-06,
+      "loss": 0.3432,
       "step": 21250
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.9198666666666667,
-      "eval_loss": 0.3090076744556427,
-      "eval_runtime": 123.9046,
-      "eval_samples_per_second": 60.53,
-      "eval_steps_per_second": 7.57,
+      "eval_accuracy": 0.9209333333333334,
+      "eval_loss": 0.2990027368068695,
+      "eval_runtime": 253.833,
+      "eval_samples_per_second": 29.547,
+      "eval_steps_per_second": 3.695,
       "step": 21252
     },
     {
       "epoch": 4.001505740636175,
-      "grad_norm": 3.5540566444396973,
-      "learning_rate": 3.993977037455299e-06,
-      "loss": 0.726,
+      "grad_norm": 10.212028503417969,
+      "learning_rate": 5.990965556182948e-06,
+      "loss": 0.751,
       "step": 21260
     },
     {
       "epoch": 4.003387916431395,
-      "grad_norm": 33.1975212097168,
-      "learning_rate": 3.986448334274422e-06,
-      "loss": 0.4819,
+      "grad_norm": 12.337179183959961,
+      "learning_rate": 5.9796725014116315e-06,
+      "loss": 0.4736,
       "step": 21270
     },
     {
       "epoch": 4.005270092226614,
-      "grad_norm": 6.789219379425049,
-      "learning_rate": 3.9789196310935445e-06,
-      "loss": 0.3697,
+      "grad_norm": 1.4147861003875732,
+      "learning_rate": 5.968379446640317e-06,
+      "loss": 0.3634,
       "step": 21280
     },
     {
       "epoch": 4.007152268021834,
-      "grad_norm": 20.950092315673828,
-      "learning_rate": 3.9713909279126674e-06,
-      "loss": 0.6637,
+      "grad_norm": 21.909902572631836,
+      "learning_rate": 5.957086391869e-06,
+      "loss": 0.5764,
       "step": 21290
     },
     {
       "epoch": 4.009034443817052,
-      "grad_norm": 7.9257707595825195,
-      "learning_rate": 3.96386222473179e-06,
-      "loss": 0.645,
+      "grad_norm": 8.32170295715332,
+      "learning_rate": 5.9457933370976855e-06,
+      "loss": 0.5818,
       "step": 21300
     },
     {
       "epoch": 4.010916619612272,
-      "grad_norm": 2.989037275314331,
-      "learning_rate": 3.956333521550913e-06,
-      "loss": 0.7831,
+      "grad_norm": 4.794271945953369,
+      "learning_rate": 5.934500282326369e-06,
+      "loss": 0.7417,
       "step": 21310
     },
     {
       "epoch": 4.012798795407491,
-      "grad_norm": 23.511072158813477,
-      "learning_rate": 3.948804818370036e-06,
-      "loss": 0.4667,
+      "grad_norm": 25.876705169677734,
+      "learning_rate": 5.923207227555054e-06,
+      "loss": 0.4546,
       "step": 21320
     },
     {
       "epoch": 4.0146809712027105,
-      "grad_norm": 20.884122848510742,
-      "learning_rate": 3.941276115189159e-06,
-      "loss": 0.8412,
+      "grad_norm": 26.065282821655273,
+      "learning_rate": 5.911914172783738e-06,
+      "loss": 0.7006,
       "step": 21330
     },
     {
       "epoch": 4.01656314699793,
-      "grad_norm": 31.798341751098633,
-      "learning_rate": 3.933747412008282e-06,
-      "loss": 0.7123,
+      "grad_norm": 34.19929122924805,
+      "learning_rate": 5.900621118012422e-06,
+      "loss": 0.6222,
       "step": 21340
     },
     {
       "epoch": 4.0184453227931485,
-      "grad_norm": 10.063611030578613,
-      "learning_rate": 3.926218708827404e-06,
-      "loss": 0.5005,
+      "grad_norm": 1.1643229722976685,
+      "learning_rate": 5.889328063241107e-06,
+      "loss": 0.4165,
       "step": 21350
     },
     {
       "epoch": 4.020327498588368,
-      "grad_norm": 13.424291610717773,
-      "learning_rate": 3.918690005646528e-06,
-      "loss": 0.5276,
+      "grad_norm": 10.392925262451172,
+      "learning_rate": 5.878035008469791e-06,
+      "loss": 0.5281,
       "step": 21360
     },
     {
       "epoch": 4.022209674383587,
-      "grad_norm": 2.927164316177368,
-      "learning_rate": 3.911161302465651e-06,
-      "loss": 0.4943,
+      "grad_norm": 25.98088264465332,
+      "learning_rate": 5.866741953698476e-06,
+      "loss": 0.4653,
       "step": 21370
     },
     {
       "epoch": 4.024091850178807,
-      "grad_norm": 21.97260856628418,
-      "learning_rate": 3.903632599284774e-06,
-      "loss": 0.773,
+      "grad_norm": 19.15206527709961,
+      "learning_rate": 5.85544889892716e-06,
+      "loss": 0.6925,
       "step": 21380
     },
     {
       "epoch": 4.025974025974026,
-      "grad_norm": 14.4227876663208,
-      "learning_rate": 3.896103896103897e-06,
-      "loss": 0.8997,
+      "grad_norm": 8.375141143798828,
+      "learning_rate": 5.844155844155845e-06,
+      "loss": 0.975,
       "step": 21390
     },
     {
       "epoch": 4.027856201769246,
-      "grad_norm": 19.949077606201172,
-      "learning_rate": 3.8885751929230196e-06,
-      "loss": 0.4273,
+      "grad_norm": 11.092159271240234,
+      "learning_rate": 5.8328627893845285e-06,
+      "loss": 0.4277,
       "step": 21400
     },
     {
       "epoch": 4.029738377564464,
-      "grad_norm": 12.47314167022705,
-      "learning_rate": 3.8810464897421425e-06,
-      "loss": 0.8176,
+      "grad_norm": 17.455846786499023,
+      "learning_rate": 5.821569734613213e-06,
+      "loss": 0.7711,
       "step": 21410
     },
     {
       "epoch": 4.031620553359684,
-      "grad_norm": 8.580794334411621,
-      "learning_rate": 3.8735177865612646e-06,
-      "loss": 0.7012,
+      "grad_norm": 5.300505638122559,
+      "learning_rate": 5.810276679841897e-06,
+      "loss": 0.6583,
       "step": 21420
     },
     {
       "epoch": 4.033502729154903,
-      "grad_norm": 0.19655872881412506,
-      "learning_rate": 3.865989083380388e-06,
-      "loss": 0.6125,
+      "grad_norm": 0.19853296875953674,
+      "learning_rate": 5.798983625070582e-06,
+      "loss": 0.5738,
       "step": 21430
     },
     {
       "epoch": 4.0353849049501225,
-      "grad_norm": 39.343509674072266,
-      "learning_rate": 3.85846038019951e-06,
-      "loss": 0.3181,
+      "grad_norm": 19.96097183227539,
+      "learning_rate": 5.787690570299266e-06,
+      "loss": 0.3168,
       "step": 21440
     },
     {
       "epoch": 4.037267080745342,
-      "grad_norm": 24.081876754760742,
-      "learning_rate": 3.850931677018634e-06,
-      "loss": 0.3806,
+      "grad_norm": 10.578010559082031,
+      "learning_rate": 5.77639751552795e-06,
+      "loss": 0.3206,
       "step": 21450
     },
     {
       "epoch": 4.0391492565405605,
-      "grad_norm": 10.21368408203125,
-      "learning_rate": 3.843402973837756e-06,
-      "loss": 0.5367,
+      "grad_norm": 7.319991111755371,
+      "learning_rate": 5.765104460756635e-06,
+      "loss": 0.6817,
       "step": 21460
     },
     {
       "epoch": 4.04103143233578,
-      "grad_norm": 3.1306040287017822,
-      "learning_rate": 3.83587427065688e-06,
-      "loss": 0.5962,
+      "grad_norm": 1.0113133192062378,
+      "learning_rate": 5.753811405985319e-06,
+      "loss": 0.4977,
       "step": 21470
     },
     {
       "epoch": 4.042913608130999,
-      "grad_norm": 0.1672428995370865,
-      "learning_rate": 3.828345567476003e-06,
-      "loss": 0.3698,
+      "grad_norm": 0.2078145146369934,
+      "learning_rate": 5.7425183512140036e-06,
+      "loss": 0.3547,
       "step": 21480
     },
     {
       "epoch": 4.044795783926219,
-      "grad_norm": 22.689733505249023,
-      "learning_rate": 3.820816864295126e-06,
-      "loss": 0.4011,
+      "grad_norm": 19.170316696166992,
+      "learning_rate": 5.731225296442688e-06,
+      "loss": 0.3471,
       "step": 21490
     },
     {
       "epoch": 4.046677959721438,
-      "grad_norm": 21.669288635253906,
-      "learning_rate": 3.8132881611142488e-06,
-      "loss": 0.945,
+      "grad_norm": 20.73012924194336,
+      "learning_rate": 5.719932241671372e-06,
+      "loss": 0.8461,
       "step": 21500
     },
     {
       "epoch": 4.048560135516658,
-      "grad_norm": 27.549644470214844,
-      "learning_rate": 3.8057594579333713e-06,
-      "loss": 0.4268,
+      "grad_norm": 33.84339141845703,
+      "learning_rate": 5.708639186900057e-06,
+      "loss": 0.406,
       "step": 21510
     },
     {
       "epoch": 4.050442311311876,
-      "grad_norm": 1.7836955785751343,
-      "learning_rate": 3.798230754752494e-06,
-      "loss": 0.2654,
+      "grad_norm": 0.295187771320343,
+      "learning_rate": 5.697346132128741e-06,
+      "loss": 0.2405,
       "step": 21520
     },
     {
       "epoch": 4.052324487107096,
-      "grad_norm": 21.68084716796875,
-      "learning_rate": 3.790702051571617e-06,
-      "loss": 0.4231,
+      "grad_norm": 8.998740196228027,
+      "learning_rate": 5.6860530773574255e-06,
+      "loss": 0.4896,
       "step": 21530
     },
     {
       "epoch": 4.054206662902315,
-      "grad_norm": 15.864786148071289,
-      "learning_rate": 3.78317334839074e-06,
-      "loss": 0.4331,
+      "grad_norm": 2.618791103363037,
+      "learning_rate": 5.67476002258611e-06,
+      "loss": 0.3719,
       "step": 21540
     },
     {
       "epoch": 4.0560888386975344,
-      "grad_norm": 8.979389190673828,
-      "learning_rate": 3.7756446452098625e-06,
-      "loss": 0.8534,
+      "grad_norm": 12.432368278503418,
+      "learning_rate": 5.663466967814793e-06,
+      "loss": 0.9172,
       "step": 21550
     },
     {
       "epoch": 4.057971014492754,
-      "grad_norm": 13.514737129211426,
-      "learning_rate": 3.768115942028986e-06,
-      "loss": 0.5054,
+      "grad_norm": 22.641036987304688,
+      "learning_rate": 5.652173913043479e-06,
+      "loss": 0.498,
       "step": 21560
     },
     {
       "epoch": 4.059853190287973,
-      "grad_norm": 6.7341766357421875,
-      "learning_rate": 3.7605872388481084e-06,
-      "loss": 0.495,
+      "grad_norm": 1.2676461935043335,
+      "learning_rate": 5.640880858272162e-06,
+      "loss": 0.4709,
       "step": 21570
     },
     {
       "epoch": 4.061735366083192,
-      "grad_norm": 8.057145118713379,
-      "learning_rate": 3.7530585356672317e-06,
-      "loss": 0.4183,
+      "grad_norm": 8.317973136901855,
+      "learning_rate": 5.629587803500847e-06,
+      "loss": 0.463,
       "step": 21580
     },
     {
       "epoch": 4.063617541878411,
-      "grad_norm": 17.93224334716797,
-      "learning_rate": 3.7455298324863542e-06,
-      "loss": 0.4697,
+      "grad_norm": 12.644018173217773,
+      "learning_rate": 5.618294748729531e-06,
+      "loss": 0.4324,
       "step": 21590
     },
     {
       "epoch": 4.065499717673631,
-      "grad_norm": 12.62182331085205,
-      "learning_rate": 3.7380011293054776e-06,
-      "loss": 0.2779,
+      "grad_norm": 6.407628536224365,
+      "learning_rate": 5.607001693958216e-06,
+      "loss": 0.256,
       "step": 21600
     },
     {
       "epoch": 4.06738189346885,
-      "grad_norm": 1.0005717277526855,
-      "learning_rate": 3.7304724261246005e-06,
-      "loss": 0.6458,
+      "grad_norm": 0.3914599120616913,
+      "learning_rate": 5.5957086391869005e-06,
+      "loss": 0.6861,
       "step": 21610
     },
     {
       "epoch": 4.06926406926407,
-      "grad_norm": 9.269535064697266,
-      "learning_rate": 3.722943722943723e-06,
-      "loss": 0.4826,
+      "grad_norm": 9.922120094299316,
+      "learning_rate": 5.584415584415584e-06,
+      "loss": 0.4863,
       "step": 21620
     },
     {
       "epoch": 4.071146245059288,
-      "grad_norm": 0.3140579164028168,
-      "learning_rate": 3.7154150197628463e-06,
-      "loss": 0.6507,
+      "grad_norm": 0.7256487011909485,
+      "learning_rate": 5.573122529644269e-06,
+      "loss": 0.585,
       "step": 21630
     },
     {
       "epoch": 4.073028420854508,
-      "grad_norm": 20.74129295349121,
-      "learning_rate": 3.707886316581969e-06,
-      "loss": 0.4982,
+      "grad_norm": 10.62320613861084,
+      "learning_rate": 5.561829474872953e-06,
+      "loss": 0.43,
       "step": 21640
     },
     {
       "epoch": 4.074910596649727,
-      "grad_norm": 2.1303975582122803,
-      "learning_rate": 3.700357613401092e-06,
-      "loss": 0.5702,
+      "grad_norm": 2.3755974769592285,
+      "learning_rate": 5.550536420101638e-06,
+      "loss": 0.4797,
       "step": 21650
     },
     {
       "epoch": 4.076792772444946,
-      "grad_norm": 0.7507879734039307,
-      "learning_rate": 3.6928289102202147e-06,
-      "loss": 0.4595,
+      "grad_norm": 0.20227846503257751,
+      "learning_rate": 5.539243365330322e-06,
+      "loss": 0.4626,
       "step": 21660
     },
     {
       "epoch": 4.078674948240166,
-      "grad_norm": 12.625158309936523,
-      "learning_rate": 3.685300207039338e-06,
-      "loss": 0.3126,
+      "grad_norm": 12.040499687194824,
+      "learning_rate": 5.527950310559007e-06,
+      "loss": 0.3619,
       "step": 21670
     },
     {
       "epoch": 4.080557124035385,
-      "grad_norm": 4.407412052154541,
-      "learning_rate": 3.6777715038584605e-06,
-      "loss": 0.4585,
+      "grad_norm": 1.3578740358352661,
+      "learning_rate": 5.51665725578769e-06,
+      "loss": 0.4352,
       "step": 21680
     },
     {
       "epoch": 4.082439299830604,
-      "grad_norm": 33.84129333496094,
-      "learning_rate": 3.670242800677584e-06,
-      "loss": 0.7722,
+      "grad_norm": 7.650432109832764,
+      "learning_rate": 5.505364201016376e-06,
+      "loss": 0.673,
       "step": 21690
     },
     {
       "epoch": 4.084321475625823,
-      "grad_norm": 9.634953498840332,
-      "learning_rate": 3.6627140974967064e-06,
-      "loss": 0.5845,
+      "grad_norm": 7.1012983322143555,
+      "learning_rate": 5.494071146245059e-06,
+      "loss": 0.5103,
       "step": 21700
     },
     {
       "epoch": 4.086203651421043,
-      "grad_norm": 32.898746490478516,
-      "learning_rate": 3.6551853943158293e-06,
-      "loss": 0.4641,
+      "grad_norm": 28.039716720581055,
+      "learning_rate": 5.4827780914737435e-06,
+      "loss": 0.4092,
       "step": 21710
     },
     {
       "epoch": 4.088085827216262,
-      "grad_norm": 10.510771751403809,
-      "learning_rate": 3.6476566911349526e-06,
-      "loss": 0.7289,
+      "grad_norm": 7.539644718170166,
+      "learning_rate": 5.471485036702429e-06,
+      "loss": 0.6337,
       "step": 21720
     },
     {
       "epoch": 4.089968003011482,
-      "grad_norm": 24.86658477783203,
-      "learning_rate": 3.640127987954075e-06,
-      "loss": 0.6348,
+      "grad_norm": 28.933773040771484,
+      "learning_rate": 5.460191981931112e-06,
+      "loss": 0.4827,
       "step": 21730
     },
     {
       "epoch": 4.0918501788067,
-      "grad_norm": 14.314352989196777,
-      "learning_rate": 3.6325992847731985e-06,
-      "loss": 0.6093,
+      "grad_norm": 25.65105628967285,
+      "learning_rate": 5.4488989271597975e-06,
+      "loss": 0.5597,
       "step": 21740
     },
     {
       "epoch": 4.0937323546019195,
-      "grad_norm": 27.23653793334961,
-      "learning_rate": 3.625070581592321e-06,
-      "loss": 0.5205,
+      "grad_norm": 27.194011688232422,
+      "learning_rate": 5.437605872388481e-06,
+      "loss": 0.5449,
       "step": 21750
     },
     {
       "epoch": 4.095614530397139,
-      "grad_norm": 19.099502563476562,
-      "learning_rate": 3.6175418784114443e-06,
-      "loss": 0.6852,
+      "grad_norm": 11.716024398803711,
+      "learning_rate": 5.426312817617166e-06,
+      "loss": 0.6295,
       "step": 21760
     },
     {
       "epoch": 4.097496706192358,
-      "grad_norm": 20.99348258972168,
-      "learning_rate": 3.610013175230567e-06,
-      "loss": 0.6112,
+      "grad_norm": 17.159563064575195,
+      "learning_rate": 5.41501976284585e-06,
+      "loss": 0.4375,
       "step": 21770
     },
     {
       "epoch": 4.099378881987578,
-      "grad_norm": 8.206368446350098,
-      "learning_rate": 3.6024844720496897e-06,
-      "loss": 0.688,
+      "grad_norm": 6.972395420074463,
+      "learning_rate": 5.403726708074534e-06,
+      "loss": 0.6678,
       "step": 21780
     },
     {
       "epoch": 4.101261057782797,
-      "grad_norm": 7.729517459869385,
-      "learning_rate": 3.5949557688688127e-06,
-      "loss": 0.3646,
+      "grad_norm": 2.577281951904297,
+      "learning_rate": 5.3924336533032186e-06,
+      "loss": 0.3085,
       "step": 21790
     },
     {
       "epoch": 4.103143233578016,
-      "grad_norm": 142.78952026367188,
-      "learning_rate": 3.5874270656879356e-06,
-      "loss": 0.6408,
+      "grad_norm": 8.173493385314941,
+      "learning_rate": 5.381140598531903e-06,
+      "loss": 0.4501,
       "step": 21800
     },
     {
       "epoch": 4.105025409373235,
-      "grad_norm": 7.343946933746338,
-      "learning_rate": 3.579898362507058e-06,
-      "loss": 0.6473,
+      "grad_norm": 0.4988132417201996,
+      "learning_rate": 5.369847543760587e-06,
+      "loss": 0.4765,
       "step": 21810
     },
     {
       "epoch": 4.106907585168455,
-      "grad_norm": 9.92300796508789,
-      "learning_rate": 3.5723696593261814e-06,
-      "loss": 0.436,
+      "grad_norm": 0.59678715467453,
+      "learning_rate": 5.358554488989272e-06,
+      "loss": 0.3523,
       "step": 21820
     },
     {
       "epoch": 4.108789760963674,
-      "grad_norm": 22.44314193725586,
-      "learning_rate": 3.564840956145304e-06,
-      "loss": 0.6023,
+      "grad_norm": 22.480344772338867,
+      "learning_rate": 5.347261434217956e-06,
+      "loss": 0.5372,
       "step": 21830
     },
     {
       "epoch": 4.1106719367588935,
-      "grad_norm": 14.589250564575195,
-      "learning_rate": 3.5573122529644273e-06,
-      "loss": 0.7282,
+      "grad_norm": 6.5898284912109375,
+      "learning_rate": 5.3359683794466405e-06,
+      "loss": 0.6108,
       "step": 21840
     },
     {
       "epoch": 4.112554112554113,
-      "grad_norm": 9.727843284606934,
-      "learning_rate": 3.54978354978355e-06,
-      "loss": 0.3819,
+      "grad_norm": 11.716509819030762,
+      "learning_rate": 5.324675324675325e-06,
+      "loss": 0.3293,
       "step": 21850
     },
     {
       "epoch": 4.1144362883493315,
-      "grad_norm": 6.11102819442749,
-      "learning_rate": 3.542254846602673e-06,
-      "loss": 0.6332,
+      "grad_norm": 9.849950790405273,
+      "learning_rate": 5.313382269904009e-06,
+      "loss": 0.5861,
       "step": 21860
     },
     {
       "epoch": 4.116318464144551,
-      "grad_norm": 15.100151062011719,
-      "learning_rate": 3.534726143421796e-06,
-      "loss": 0.4667,
+      "grad_norm": 16.31063461303711,
+      "learning_rate": 5.302089215132694e-06,
+      "loss": 0.3767,
       "step": 21870
     },
     {
       "epoch": 4.11820063993977,
-      "grad_norm": 22.583946228027344,
-      "learning_rate": 3.5271974402409185e-06,
-      "loss": 0.484,
+      "grad_norm": 18.616395950317383,
+      "learning_rate": 5.290796160361378e-06,
+      "loss": 0.5778,
       "step": 21880
     },
     {
       "epoch": 4.12008281573499,
-      "grad_norm": 7.776749610900879,
-      "learning_rate": 3.519668737060042e-06,
-      "loss": 0.7223,
+      "grad_norm": 12.037795066833496,
+      "learning_rate": 5.279503105590062e-06,
+      "loss": 0.5588,
       "step": 21890
     },
     {
       "epoch": 4.121964991530209,
-      "grad_norm": 17.078176498413086,
-      "learning_rate": 3.5121400338791644e-06,
-      "loss": 0.2882,
+      "grad_norm": 5.515803337097168,
+      "learning_rate": 5.268210050818747e-06,
+      "loss": 0.2835,
       "step": 21900
     },
     {
       "epoch": 4.123847167325428,
-      "grad_norm": 20.212486267089844,
-      "learning_rate": 3.5046113306982877e-06,
-      "loss": 0.4445,
+      "grad_norm": 8.707112312316895,
+      "learning_rate": 5.256916996047431e-06,
+      "loss": 0.3438,
       "step": 21910
     },
     {
       "epoch": 4.125729343120647,
-      "grad_norm": 1.8471107482910156,
-      "learning_rate": 3.49708262751741e-06,
-      "loss": 0.4017,
+      "grad_norm": 0.553016185760498,
+      "learning_rate": 5.245623941276115e-06,
+      "loss": 0.3274,
       "step": 21920
     },
     {
       "epoch": 4.127611518915867,
-      "grad_norm": 7.003720283508301,
-      "learning_rate": 3.4895539243365336e-06,
-      "loss": 0.582,
+      "grad_norm": 14.9821195602417,
+      "learning_rate": 5.2343308865048e-06,
+      "loss": 0.5166,
       "step": 21930
     },
     {
       "epoch": 4.129493694711086,
-      "grad_norm": 29.45206069946289,
-      "learning_rate": 3.482025221155656e-06,
-      "loss": 0.6031,
+      "grad_norm": 34.491355895996094,
+      "learning_rate": 5.2230378317334834e-06,
+      "loss": 0.4849,
       "step": 21940
     },
     {
       "epoch": 4.1313758705063055,
-      "grad_norm": 14.213915824890137,
-      "learning_rate": 3.474496517974779e-06,
-      "loss": 0.8854,
+      "grad_norm": 14.449700355529785,
+      "learning_rate": 5.211744776962169e-06,
+      "loss": 0.768,
       "step": 21950
     },
     {
       "epoch": 4.133258046301525,
-      "grad_norm": 9.18936538696289,
-      "learning_rate": 3.466967814793902e-06,
-      "loss": 0.4027,
+      "grad_norm": 17.19442367553711,
+      "learning_rate": 5.200451722190852e-06,
+      "loss": 0.3727,
       "step": 21960
     },
     {
       "epoch": 4.1351402220967435,
-      "grad_norm": 2.583569049835205,
-      "learning_rate": 3.459439111613025e-06,
-      "loss": 0.6699,
+      "grad_norm": 2.9677646160125732,
+      "learning_rate": 5.1891586674195374e-06,
+      "loss": 0.6443,
       "step": 21970
     },
     {
       "epoch": 4.137022397891963,
-      "grad_norm": 9.588484764099121,
-      "learning_rate": 3.451910408432148e-06,
-      "loss": 0.7421,
+      "grad_norm": 8.971100807189941,
+      "learning_rate": 5.177865612648222e-06,
+      "loss": 0.744,
       "step": 21980
     },
     {
       "epoch": 4.138904573687182,
-      "grad_norm": 9.309282302856445,
-      "learning_rate": 3.4443817052512707e-06,
-      "loss": 0.4973,
+      "grad_norm": 14.347002029418945,
+      "learning_rate": 5.166572557876905e-06,
+      "loss": 0.4241,
       "step": 21990
     },
     {
       "epoch": 4.140786749482402,
-      "grad_norm": 18.542089462280273,
-      "learning_rate": 3.436853002070394e-06,
-      "loss": 0.6043,
+      "grad_norm": 3.654014825820923,
+      "learning_rate": 5.155279503105591e-06,
+      "loss": 0.6624,
       "step": 22000
     },
     {
       "epoch": 4.142668925277621,
-      "grad_norm": 24.662412643432617,
-      "learning_rate": 3.4293242988895165e-06,
-      "loss": 0.6221,
+      "grad_norm": 18.717609405517578,
+      "learning_rate": 5.143986448334274e-06,
+      "loss": 0.6139,
       "step": 22010
     },
     {
       "epoch": 4.144551101072841,
-      "grad_norm": 20.624713897705078,
-      "learning_rate": 3.4217955957086394e-06,
-      "loss": 0.7082,
+      "grad_norm": 25.252634048461914,
+      "learning_rate": 5.132693393562959e-06,
+      "loss": 0.5933,
       "step": 22020
     },
     {
       "epoch": 4.146433276868059,
-      "grad_norm": 5.393800258636475,
-      "learning_rate": 3.4142668925277623e-06,
-      "loss": 0.7466,
+      "grad_norm": 8.735464096069336,
+      "learning_rate": 5.121400338791643e-06,
+      "loss": 0.782,
       "step": 22030
     },
     {
       "epoch": 4.148315452663279,
-      "grad_norm": 8.6702880859375,
-      "learning_rate": 3.4067381893468853e-06,
-      "loss": 0.4577,
+      "grad_norm": 2.7719712257385254,
+      "learning_rate": 5.110107284020328e-06,
+      "loss": 0.4075,
       "step": 22040
     },
     {
       "epoch": 4.150197628458498,
-      "grad_norm": 8.493791580200195,
-      "learning_rate": 3.399209486166008e-06,
-      "loss": 0.5303,
+      "grad_norm": 14.126623153686523,
+      "learning_rate": 5.098814229249012e-06,
+      "loss": 0.4613,
       "step": 22050
     },
     {
       "epoch": 4.1520798042537175,
-      "grad_norm": 3.720416784286499,
-      "learning_rate": 3.391680782985131e-06,
-      "loss": 0.8215,
+      "grad_norm": 1.2759391069412231,
+      "learning_rate": 5.087521174477696e-06,
+      "loss": 0.7308,
       "step": 22060
     },
     {
       "epoch": 4.153961980048937,
-      "grad_norm": 3.568179130554199,
-      "learning_rate": 3.3841520798042536e-06,
-      "loss": 0.5173,
+      "grad_norm": 2.6802024841308594,
+      "learning_rate": 5.07622811970638e-06,
+      "loss": 0.3868,
       "step": 22070
     },
     {
       "epoch": 4.1558441558441555,
-      "grad_norm": 11.129983901977539,
-      "learning_rate": 3.376623376623377e-06,
-      "loss": 0.6122,
+      "grad_norm": 20.169902801513672,
+      "learning_rate": 5.064935064935065e-06,
+      "loss": 0.6009,
       "step": 22080
     },
     {
       "epoch": 4.157726331639375,
-      "grad_norm": 13.840152740478516,
-      "learning_rate": 3.3690946734425e-06,
-      "loss": 0.624,
+      "grad_norm": 8.015284538269043,
+      "learning_rate": 5.05364201016375e-06,
+      "loss": 0.5751,
       "step": 22090
     },
     {
       "epoch": 4.159608507434594,
-      "grad_norm": 28.01687240600586,
-      "learning_rate": 3.361565970261623e-06,
-      "loss": 0.6082,
+      "grad_norm": 9.399229049682617,
+      "learning_rate": 5.0423489553924336e-06,
+      "loss": 0.472,
       "step": 22100
     },
     {
       "epoch": 4.161490683229814,
-      "grad_norm": 18.05573081970215,
-      "learning_rate": 3.3540372670807457e-06,
-      "loss": 0.319,
+      "grad_norm": 16.64990997314453,
+      "learning_rate": 5.031055900621119e-06,
+      "loss": 0.2494,
       "step": 22110
     },
     {
       "epoch": 4.163372859025033,
-      "grad_norm": 27.826292037963867,
-      "learning_rate": 3.3465085638998686e-06,
-      "loss": 0.6604,
+      "grad_norm": 34.64622116088867,
+      "learning_rate": 5.019762845849802e-06,
+      "loss": 0.5575,
       "step": 22120
     },
     {
       "epoch": 4.165255034820253,
-      "grad_norm": 25.494606018066406,
-      "learning_rate": 3.3389798607189916e-06,
-      "loss": 0.5413,
+      "grad_norm": 43.623146057128906,
+      "learning_rate": 5.0084697910784875e-06,
+      "loss": 0.4865,
       "step": 22130
     },
     {
       "epoch": 4.167137210615471,
-      "grad_norm": 34.23874282836914,
-      "learning_rate": 3.331451157538114e-06,
-      "loss": 0.7515,
+      "grad_norm": 11.624351501464844,
+      "learning_rate": 4.997176736307171e-06,
+      "loss": 0.6536,
       "step": 22140
     },
     {
       "epoch": 4.169019386410691,
-      "grad_norm": 20.008041381835938,
-      "learning_rate": 3.3239224543572374e-06,
-      "loss": 0.7427,
+      "grad_norm": 6.294399261474609,
+      "learning_rate": 4.9858836815358555e-06,
+      "loss": 0.6794,
       "step": 22150
     },
     {
       "epoch": 4.17090156220591,
-      "grad_norm": 21.009740829467773,
-      "learning_rate": 3.31639375117636e-06,
-      "loss": 0.3218,
+      "grad_norm": 16.353845596313477,
+      "learning_rate": 4.97459062676454e-06,
+      "loss": 0.2492,
       "step": 22160
     },
     {
       "epoch": 4.1727837380011294,
-      "grad_norm": 31.26310920715332,
-      "learning_rate": 3.3088650479954832e-06,
-      "loss": 0.593,
+      "grad_norm": 6.7267608642578125,
+      "learning_rate": 4.963297571993224e-06,
+      "loss": 0.5912,
       "step": 22170
     },
     {
       "epoch": 4.174665913796349,
-      "grad_norm": 0.45123204588890076,
-      "learning_rate": 3.3013363448146057e-06,
-      "loss": 0.4274,
+      "grad_norm": 0.19940780103206635,
+      "learning_rate": 4.952004517221909e-06,
+      "loss": 0.3615,
       "step": 22180
     },
     {
       "epoch": 4.176548089591567,
-      "grad_norm": 16.512317657470703,
-      "learning_rate": 3.293807641633729e-06,
-      "loss": 0.7006,
+      "grad_norm": 13.068621635437012,
+      "learning_rate": 4.940711462450593e-06,
+      "loss": 0.7576,
       "step": 22190
     },
     {
       "epoch": 4.178430265386787,
-      "grad_norm": 7.592436790466309,
-      "learning_rate": 3.2862789384528516e-06,
-      "loss": 0.5583,
+      "grad_norm": 4.3102240562438965,
+      "learning_rate": 4.929418407679277e-06,
+      "loss": 0.5176,
       "step": 22200
     },
     {
       "epoch": 4.180312441182006,
-      "grad_norm": 19.937170028686523,
-      "learning_rate": 3.2787502352719745e-06,
-      "loss": 0.8125,
+      "grad_norm": 17.297819137573242,
+      "learning_rate": 4.918125352907962e-06,
+      "loss": 0.7968,
       "step": 22210
     },
     {
       "epoch": 4.182194616977226,
-      "grad_norm": 24.68047332763672,
-      "learning_rate": 3.271221532091098e-06,
-      "loss": 0.6438,
+      "grad_norm": 29.88180923461914,
+      "learning_rate": 4.906832298136646e-06,
+      "loss": 0.5499,
       "step": 22220
     },
     {
       "epoch": 4.184076792772445,
-      "grad_norm": 16.7028865814209,
-      "learning_rate": 3.2636928289102203e-06,
-      "loss": 0.3172,
+      "grad_norm": 35.76202392578125,
+      "learning_rate": 4.8955392433653305e-06,
+      "loss": 0.2527,
       "step": 22230
     },
     {
       "epoch": 4.185958968567665,
-      "grad_norm": 31.273239135742188,
-      "learning_rate": 3.2561641257293437e-06,
-      "loss": 0.413,
+      "grad_norm": 23.927745819091797,
+      "learning_rate": 4.884246188594015e-06,
+      "loss": 0.4063,
       "step": 22240
     },
     {
       "epoch": 4.187841144362883,
-      "grad_norm": 7.516437530517578,
-      "learning_rate": 3.248635422548466e-06,
-      "loss": 0.5438,
+      "grad_norm": 7.700712203979492,
+      "learning_rate": 4.872953133822699e-06,
+      "loss": 0.4893,
       "step": 22250
     },
     {
       "epoch": 4.189723320158103,
-      "grad_norm": 12.695963859558105,
-      "learning_rate": 3.2411067193675895e-06,
-      "loss": 0.7824,
+      "grad_norm": 14.746334075927734,
+      "learning_rate": 4.861660079051384e-06,
+      "loss": 0.7473,
       "step": 22260
     },
     {
       "epoch": 4.191605495953322,
-      "grad_norm": 0.8344699144363403,
-      "learning_rate": 3.233578016186712e-06,
+      "grad_norm": 0.5177063941955566,
+      "learning_rate": 4.850367024280068e-06,
       "loss": 0.5172,
       "step": 22270
     },
     {
       "epoch": 4.193487671748541,
-      "grad_norm": 7.483529090881348,
-      "learning_rate": 3.226049313005835e-06,
-      "loss": 0.5373,
+      "grad_norm": 5.81657075881958,
+      "learning_rate": 4.8390739695087524e-06,
+      "loss": 0.5498,
       "step": 22280
     },
     {
       "epoch": 4.195369847543761,
-      "grad_norm": 0.5889711380004883,
-      "learning_rate": 3.218520609824958e-06,
-      "loss": 0.5973,
+      "grad_norm": 0.22678916156291962,
+      "learning_rate": 4.827780914737436e-06,
+      "loss": 0.4705,
       "step": 22290
     },
     {
       "epoch": 4.19725202333898,
-      "grad_norm": 13.00317096710205,
-      "learning_rate": 3.210991906644081e-06,
-      "loss": 0.5395,
+      "grad_norm": 18.123903274536133,
+      "learning_rate": 4.816487859966121e-06,
+      "loss": 0.4834,
       "step": 22300
     },
     {
       "epoch": 4.199134199134199,
-      "grad_norm": 22.1431827545166,
-      "learning_rate": 3.2034632034632033e-06,
-      "loss": 0.7073,
+      "grad_norm": 22.923141479492188,
+      "learning_rate": 4.805194805194805e-06,
+      "loss": 0.6182,
       "step": 22310
     },
     {
       "epoch": 4.201016374929418,
-      "grad_norm": 24.43523406982422,
-      "learning_rate": 3.1959345002823266e-06,
-      "loss": 0.9535,
+      "grad_norm": 30.303974151611328,
+      "learning_rate": 4.79390175042349e-06,
+      "loss": 0.844,
       "step": 22320
     },
     {
       "epoch": 4.202898550724638,
-      "grad_norm": 0.73313969373703,
-      "learning_rate": 3.188405797101449e-06,
-      "loss": 0.6103,
+      "grad_norm": 0.405670702457428,
+      "learning_rate": 4.7826086956521735e-06,
+      "loss": 0.5439,
       "step": 22330
     },
     {
       "epoch": 4.204780726519857,
-      "grad_norm": 17.1301212310791,
-      "learning_rate": 3.1808770939205725e-06,
-      "loss": 0.5049,
+      "grad_norm": 19.21993064880371,
+      "learning_rate": 4.771315640880859e-06,
+      "loss": 0.4976,
       "step": 22340
     },
     {
       "epoch": 4.206662902315077,
-      "grad_norm": 19.849651336669922,
-      "learning_rate": 3.1733483907396954e-06,
-      "loss": 0.8419,
+      "grad_norm": 18.2979679107666,
+      "learning_rate": 4.760022586109543e-06,
+      "loss": 0.7726,
       "step": 22350
     },
     {
       "epoch": 4.208545078110295,
-      "grad_norm": 7.085902690887451,
-      "learning_rate": 3.1658196875588183e-06,
-      "loss": 0.7864,
+      "grad_norm": 4.73086404800415,
+      "learning_rate": 4.748729531338227e-06,
+      "loss": 0.6009,
       "step": 22360
     },
     {
       "epoch": 4.2104272539055145,
-      "grad_norm": 0.4933435916900635,
-      "learning_rate": 3.1582909843779412e-06,
-      "loss": 0.5251,
+      "grad_norm": 0.11267448216676712,
+      "learning_rate": 4.737436476566912e-06,
+      "loss": 0.4873,
       "step": 22370
     },
     {
       "epoch": 4.212309429700734,
-      "grad_norm": 16.553647994995117,
-      "learning_rate": 3.1507622811970637e-06,
-      "loss": 0.6869,
+      "grad_norm": 13.262358665466309,
+      "learning_rate": 4.726143421795595e-06,
+      "loss": 0.588,
       "step": 22380
     },
     {
       "epoch": 4.214191605495953,
-      "grad_norm": 5.6313018798828125,
-      "learning_rate": 3.143233578016187e-06,
-      "loss": 0.4803,
+      "grad_norm": 5.213744163513184,
+      "learning_rate": 4.714850367024281e-06,
+      "loss": 0.4623,
       "step": 22390
     },
     {
       "epoch": 4.216073781291173,
-      "grad_norm": 20.206897735595703,
-      "learning_rate": 3.1357048748353096e-06,
-      "loss": 0.6774,
+      "grad_norm": 16.470924377441406,
+      "learning_rate": 4.703557312252964e-06,
+      "loss": 0.6373,
       "step": 22400
     },
     {
       "epoch": 4.217955957086392,
-      "grad_norm": 12.4583158493042,
-      "learning_rate": 3.128176171654433e-06,
-      "loss": 0.6524,
+      "grad_norm": 10.145905494689941,
+      "learning_rate": 4.692264257481649e-06,
+      "loss": 0.6058,
       "step": 22410
     },
     {
       "epoch": 4.219838132881611,
-      "grad_norm": 11.112521171569824,
-      "learning_rate": 3.1206474684735554e-06,
-      "loss": 0.5888,
+      "grad_norm": 1.2421616315841675,
+      "learning_rate": 4.680971202710333e-06,
+      "loss": 0.4736,
       "step": 22420
     },
     {
       "epoch": 4.22172030867683,
-      "grad_norm": 25.20332145690918,
-      "learning_rate": 3.1131187652926788e-06,
-      "loss": 0.4639,
+      "grad_norm": 19.461740493774414,
+      "learning_rate": 4.669678147939017e-06,
+      "loss": 0.4471,
       "step": 22430
     },
     {
       "epoch": 4.22360248447205,
-      "grad_norm": 13.431328773498535,
-      "learning_rate": 3.1055900621118013e-06,
-      "loss": 0.6233,
+      "grad_norm": 11.891735076904297,
+      "learning_rate": 4.658385093167702e-06,
+      "loss": 0.5111,
       "step": 22440
     },
     {
       "epoch": 4.225484660267269,
-      "grad_norm": 25.92032241821289,
-      "learning_rate": 3.0980613589309246e-06,
-      "loss": 1.0161,
+      "grad_norm": 42.17231750488281,
+      "learning_rate": 4.647092038396386e-06,
+      "loss": 1.0541,
       "step": 22450
     },
     {
       "epoch": 4.2273668360624885,
-      "grad_norm": 8.30104923248291,
-      "learning_rate": 3.0905326557500475e-06,
-      "loss": 0.7447,
+      "grad_norm": 8.65245246887207,
+      "learning_rate": 4.635798983625071e-06,
+      "loss": 0.6871,
       "step": 22460
     },
     {
       "epoch": 4.229249011857707,
-      "grad_norm": 2.044919967651367,
-      "learning_rate": 3.08300395256917e-06,
-      "loss": 0.5459,
+      "grad_norm": 0.8841702938079834,
+      "learning_rate": 4.624505928853755e-06,
+      "loss": 0.478,
       "step": 22470
     },
     {
       "epoch": 4.2311311876529265,
-      "grad_norm": 6.667474746704102,
-      "learning_rate": 3.0754752493882934e-06,
-      "loss": 0.8596,
+      "grad_norm": 7.112485408782959,
+      "learning_rate": 4.61321287408244e-06,
+      "loss": 0.7968,
       "step": 22480
     },
     {
       "epoch": 4.233013363448146,
-      "grad_norm": 4.499909400939941,
-      "learning_rate": 3.067946546207416e-06,
-      "loss": 0.3851,
+      "grad_norm": 11.136316299438477,
+      "learning_rate": 4.601919819311124e-06,
+      "loss": 0.3291,
       "step": 22490
     },
     {
       "epoch": 4.234895539243365,
-      "grad_norm": 12.6906099319458,
-      "learning_rate": 3.0604178430265392e-06,
-      "loss": 0.5335,
+      "grad_norm": 11.090184211730957,
+      "learning_rate": 4.590626764539808e-06,
+      "loss": 0.5061,
       "step": 22500
     },
     {
       "epoch": 4.236777715038585,
-      "grad_norm": 6.795680046081543,
-      "learning_rate": 3.0528891398456617e-06,
-      "loss": 0.4769,
+      "grad_norm": 3.2487759590148926,
+      "learning_rate": 4.579333709768492e-06,
+      "loss": 0.4647,
       "step": 22510
     },
     {
       "epoch": 4.238659890833804,
-      "grad_norm": 22.774063110351562,
-      "learning_rate": 3.045360436664785e-06,
-      "loss": 0.453,
+      "grad_norm": 30.191455841064453,
+      "learning_rate": 4.568040654997177e-06,
+      "loss": 0.3747,
       "step": 22520
     },
     {
       "epoch": 4.240542066629023,
-      "grad_norm": 14.406441688537598,
-      "learning_rate": 3.0378317334839076e-06,
-      "loss": 0.67,
+      "grad_norm": 12.943588256835938,
+      "learning_rate": 4.556747600225861e-06,
+      "loss": 0.5714,
       "step": 22530
     },
     {
       "epoch": 4.242424242424242,
-      "grad_norm": 13.296455383300781,
-      "learning_rate": 3.0303030303030305e-06,
-      "loss": 0.4151,
+      "grad_norm": 6.223907947540283,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.37,
       "step": 22540
     },
     {
       "epoch": 4.244306418219462,
-      "grad_norm": 15.775333404541016,
-      "learning_rate": 3.0227743271221534e-06,
-      "loss": 0.5242,
+      "grad_norm": 23.05088996887207,
+      "learning_rate": 4.53416149068323e-06,
+      "loss": 0.456,
       "step": 22550
     },
     {
       "epoch": 4.246188594014681,
-      "grad_norm": 14.018509864807129,
-      "learning_rate": 3.0152456239412763e-06,
-      "loss": 0.3969,
+      "grad_norm": 21.883832931518555,
+      "learning_rate": 4.522868435911914e-06,
+      "loss": 0.3142,
       "step": 22560
     },
     {
       "epoch": 4.2480707698099005,
-      "grad_norm": 13.857187271118164,
-      "learning_rate": 3.007716920760399e-06,
-      "loss": 0.5807,
+      "grad_norm": 10.83736515045166,
+      "learning_rate": 4.511575381140599e-06,
+      "loss": 0.5158,
       "step": 22570
     },
     {
       "epoch": 4.24995294560512,
-      "grad_norm": 4.365860462188721,
-      "learning_rate": 3.000188217579522e-06,
-      "loss": 0.7116,
+      "grad_norm": 2.516878366470337,
+      "learning_rate": 4.500282326369283e-06,
+      "loss": 0.623,
       "step": 22580
     },
     {
       "epoch": 4.2518351214003385,
-      "grad_norm": 1.263258934020996,
-      "learning_rate": 2.9926595143986455e-06,
-      "loss": 0.4453,
+      "grad_norm": 0.16205434501171112,
+      "learning_rate": 4.4889892715979674e-06,
+      "loss": 0.5343,
       "step": 22590
     },
     {
       "epoch": 4.253717297195558,
-      "grad_norm": 11.76640796661377,
-      "learning_rate": 2.985130811217768e-06,
-      "loss": 0.6742,
+      "grad_norm": 8.219768524169922,
+      "learning_rate": 4.477696216826652e-06,
+      "loss": 0.612,
       "step": 22600
     },
     {
       "epoch": 4.255599472990777,
-      "grad_norm": 30.159893035888672,
-      "learning_rate": 2.977602108036891e-06,
-      "loss": 0.7244,
+      "grad_norm": 13.677404403686523,
+      "learning_rate": 4.466403162055336e-06,
+      "loss": 0.6012,
       "step": 22610
     },
     {
       "epoch": 4.257481648785997,
-      "grad_norm": 14.70153522491455,
-      "learning_rate": 2.970073404856014e-06,
-      "loss": 0.4377,
+      "grad_norm": 10.702381134033203,
+      "learning_rate": 4.4551101072840206e-06,
+      "loss": 0.3911,
       "step": 22620
     },
     {
       "epoch": 4.259363824581216,
-      "grad_norm": 0.953132688999176,
-      "learning_rate": 2.9625447016751368e-06,
-      "loss": 0.7653,
+      "grad_norm": 0.8724668622016907,
+      "learning_rate": 4.443817052512705e-06,
+      "loss": 0.7228,
       "step": 22630
     },
     {
       "epoch": 4.261246000376435,
-      "grad_norm": 32.616031646728516,
-      "learning_rate": 2.9550159984942593e-06,
-      "loss": 0.8886,
+      "grad_norm": 40.3377799987793,
+      "learning_rate": 4.432523997741389e-06,
+      "loss": 0.7456,
       "step": 22640
     },
     {
       "epoch": 4.263128176171654,
-      "grad_norm": 16.00680160522461,
-      "learning_rate": 2.9474872953133826e-06,
-      "loss": 0.6501,
+      "grad_norm": 15.633112907409668,
+      "learning_rate": 4.421230942970074e-06,
+      "loss": 0.5719,
       "step": 22650
     },
     {
       "epoch": 4.265010351966874,
-      "grad_norm": 18.027944564819336,
-      "learning_rate": 2.939958592132505e-06,
-      "loss": 0.748,
+      "grad_norm": 12.120109558105469,
+      "learning_rate": 4.409937888198757e-06,
+      "loss": 0.6745,
       "step": 22660
     },
     {
       "epoch": 4.266892527762093,
-      "grad_norm": 15.885736465454102,
-      "learning_rate": 2.9324298889516285e-06,
-      "loss": 0.5954,
+      "grad_norm": 17.709228515625,
+      "learning_rate": 4.3986448334274425e-06,
+      "loss": 0.5516,
       "step": 22670
     },
     {
       "epoch": 4.2687747035573125,
-      "grad_norm": 15.595780372619629,
-      "learning_rate": 2.924901185770751e-06,
-      "loss": 0.4439,
+      "grad_norm": 8.81289005279541,
+      "learning_rate": 4.387351778656126e-06,
+      "loss": 0.4194,
       "step": 22680
     },
     {
       "epoch": 4.270656879352532,
-      "grad_norm": 14.558534622192383,
-      "learning_rate": 2.9173724825898743e-06,
-      "loss": 0.5055,
+      "grad_norm": 14.428284645080566,
+      "learning_rate": 4.376058723884811e-06,
+      "loss": 0.4524,
       "step": 22690
     },
     {
       "epoch": 4.2725390551477505,
-      "grad_norm": 1.8835563659667969,
-      "learning_rate": 2.9098437794089972e-06,
-      "loss": 0.8219,
+      "grad_norm": 15.037586212158203,
+      "learning_rate": 4.364765669113496e-06,
+      "loss": 0.6796,
       "step": 22700
     },
     {
       "epoch": 4.27442123094297,
-      "grad_norm": 11.67325210571289,
-      "learning_rate": 2.9023150762281197e-06,
-      "loss": 0.4953,
+      "grad_norm": 7.700161933898926,
+      "learning_rate": 4.35347261434218e-06,
+      "loss": 0.4788,
       "step": 22710
     },
     {
       "epoch": 4.276303406738189,
-      "grad_norm": 5.809265613555908,
-      "learning_rate": 2.894786373047243e-06,
-      "loss": 0.5423,
+      "grad_norm": 3.1909353733062744,
+      "learning_rate": 4.342179559570864e-06,
+      "loss": 0.4486,
       "step": 22720
     },
     {
       "epoch": 4.278185582533409,
-      "grad_norm": 13.89829158782959,
-      "learning_rate": 2.8872576698663656e-06,
-      "loss": 0.6693,
+      "grad_norm": 17.010690689086914,
+      "learning_rate": 4.330886504799548e-06,
+      "loss": 0.6104,
       "step": 22730
     },
     {
       "epoch": 4.280067758328628,
-      "grad_norm": 8.091310501098633,
-      "learning_rate": 2.879728966685489e-06,
-      "loss": 0.6786,
+      "grad_norm": 8.390015602111816,
+      "learning_rate": 4.319593450028233e-06,
+      "loss": 0.5245,
       "step": 22740
     },
     {
       "epoch": 4.281949934123848,
-      "grad_norm": 18.016300201416016,
-      "learning_rate": 2.8722002635046114e-06,
-      "loss": 0.4909,
+      "grad_norm": 14.018758773803711,
+      "learning_rate": 4.308300395256917e-06,
+      "loss": 0.4835,
       "step": 22750
     },
     {
       "epoch": 4.283832109919066,
-      "grad_norm": 0.6892550587654114,
-      "learning_rate": 2.8646715603237348e-06,
-      "loss": 0.5059,
+      "grad_norm": 0.4540165662765503,
+      "learning_rate": 4.297007340485602e-06,
+      "loss": 0.5203,
       "step": 22760
     },
     {
       "epoch": 4.285714285714286,
-      "grad_norm": 0.654849648475647,
-      "learning_rate": 2.8571428571428573e-06,
-      "loss": 0.3984,
+      "grad_norm": 2.0470516681671143,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.4774,
       "step": 22770
     },
     {
       "epoch": 4.287596461509505,
-      "grad_norm": 19.24492835998535,
-      "learning_rate": 2.84961415396198e-06,
-      "loss": 0.4691,
+      "grad_norm": 9.436823844909668,
+      "learning_rate": 4.274421230942971e-06,
+      "loss": 0.3557,
       "step": 22780
     },
     {
       "epoch": 4.2894786373047245,
-      "grad_norm": 20.14637565612793,
-      "learning_rate": 2.842085450781103e-06,
-      "loss": 0.5974,
+      "grad_norm": 14.441892623901367,
+      "learning_rate": 4.263128176171654e-06,
+      "loss": 0.4613,
       "step": 22790
     },
     {
       "epoch": 4.291360813099944,
-      "grad_norm": 12.756900787353516,
-      "learning_rate": 2.834556747600226e-06,
-      "loss": 0.4315,
+      "grad_norm": 21.169797897338867,
+      "learning_rate": 4.251835121400339e-06,
+      "loss": 0.4245,
       "step": 22800
     },
     {
       "epoch": 4.293242988895162,
-      "grad_norm": 14.236254692077637,
-      "learning_rate": 2.827028044419349e-06,
-      "loss": 0.7167,
+      "grad_norm": 5.513450622558594,
+      "learning_rate": 4.240542066629023e-06,
+      "loss": 0.5082,
       "step": 22810
     },
     {
       "epoch": 4.295125164690382,
-      "grad_norm": 17.23895263671875,
-      "learning_rate": 2.819499341238472e-06,
-      "loss": 0.6919,
+      "grad_norm": 9.460198402404785,
+      "learning_rate": 4.229249011857707e-06,
+      "loss": 0.6396,
       "step": 22820
     },
     {
       "epoch": 4.297007340485601,
-      "grad_norm": 28.651317596435547,
-      "learning_rate": 2.811970638057595e-06,
-      "loss": 0.5164,
+      "grad_norm": 16.516008377075195,
+      "learning_rate": 4.217955957086393e-06,
+      "loss": 0.4007,
       "step": 22830
     },
     {
       "epoch": 4.298889516280821,
-      "grad_norm": 1.3199594020843506,
-      "learning_rate": 2.8044419348767177e-06,
-      "loss": 0.5391,
+      "grad_norm": 12.221315383911133,
+      "learning_rate": 4.206662902315076e-06,
+      "loss": 0.541,
       "step": 22840
     },
     {
       "epoch": 4.30077169207604,
-      "grad_norm": 7.875750541687012,
-      "learning_rate": 2.7969132316958406e-06,
-      "loss": 0.5423,
+      "grad_norm": 24.764720916748047,
+      "learning_rate": 4.195369847543761e-06,
+      "loss": 0.6108,
       "step": 22850
     },
     {
       "epoch": 4.30265386787126,
-      "grad_norm": 7.99800968170166,
-      "learning_rate": 2.7893845285149635e-06,
-      "loss": 0.67,
+      "grad_norm": 16.752914428710938,
+      "learning_rate": 4.184076792772445e-06,
+      "loss": 0.6381,
       "step": 22860
     },
     {
       "epoch": 4.304536043666478,
-      "grad_norm": 28.95355224609375,
-      "learning_rate": 2.7818558253340865e-06,
-      "loss": 0.692,
+      "grad_norm": 30.31788444519043,
+      "learning_rate": 4.172783738001129e-06,
+      "loss": 0.5743,
       "step": 22870
     },
     {
       "epoch": 4.306418219461698,
-      "grad_norm": 37.334197998046875,
-      "learning_rate": 2.7743271221532094e-06,
-      "loss": 0.485,
+      "grad_norm": 14.529326438903809,
+      "learning_rate": 4.161490683229814e-06,
+      "loss": 0.4165,
       "step": 22880
     },
     {
       "epoch": 4.308300395256917,
-      "grad_norm": 9.309508323669434,
-      "learning_rate": 2.7667984189723323e-06,
-      "loss": 0.7451,
+      "grad_norm": 5.0346808433532715,
+      "learning_rate": 4.150197628458498e-06,
+      "loss": 0.6804,
       "step": 22890
     },
     {
       "epoch": 4.310182571052136,
-      "grad_norm": 21.458972930908203,
-      "learning_rate": 2.759269715791455e-06,
-      "loss": 0.4967,
+      "grad_norm": 12.99887466430664,
+      "learning_rate": 4.1389045736871824e-06,
+      "loss": 0.3671,
       "step": 22900
     },
     {
       "epoch": 4.312064746847356,
-      "grad_norm": 13.24349308013916,
-      "learning_rate": 2.751741012610578e-06,
-      "loss": 0.5979,
+      "grad_norm": 26.515596389770508,
+      "learning_rate": 4.127611518915867e-06,
+      "loss": 0.479,
       "step": 22910
     },
     {
       "epoch": 4.313946922642575,
-      "grad_norm": 43.8453254699707,
-      "learning_rate": 2.7442123094297007e-06,
-      "loss": 0.6058,
+      "grad_norm": 21.167415618896484,
+      "learning_rate": 4.116318464144551e-06,
+      "loss": 0.5802,
       "step": 22920
     },
     {
       "epoch": 4.315829098437794,
-      "grad_norm": 8.151017189025879,
-      "learning_rate": 2.736683606248824e-06,
-      "loss": 0.6961,
+      "grad_norm": 16.790386199951172,
+      "learning_rate": 4.1050254093732356e-06,
+      "loss": 0.6888,
       "step": 22930
     },
     {
       "epoch": 4.317711274233013,
-      "grad_norm": 10.600540161132812,
-      "learning_rate": 2.7291549030679465e-06,
-      "loss": 0.7623,
+      "grad_norm": 10.245500564575195,
+      "learning_rate": 4.093732354601919e-06,
+      "loss": 0.7874,
       "step": 22940
     },
     {
       "epoch": 4.319593450028233,
-      "grad_norm": 13.357274055480957,
-      "learning_rate": 2.72162619988707e-06,
-      "loss": 0.4591,
+      "grad_norm": 21.216289520263672,
+      "learning_rate": 4.082439299830604e-06,
+      "loss": 0.3578,
       "step": 22950
     },
     {
       "epoch": 4.321475625823452,
-      "grad_norm": 11.977415084838867,
-      "learning_rate": 2.7140974967061928e-06,
-      "loss": 0.8256,
+      "grad_norm": 18.367511749267578,
+      "learning_rate": 4.071146245059289e-06,
+      "loss": 0.6691,
       "step": 22960
     },
     {
       "epoch": 4.323357801618672,
-      "grad_norm": 5.431375026702881,
-      "learning_rate": 2.7065687935253153e-06,
-      "loss": 0.5577,
+      "grad_norm": 24.98716163635254,
+      "learning_rate": 4.059853190287973e-06,
+      "loss": 0.5337,
       "step": 22970
     },
     {
       "epoch": 4.32523997741389,
-      "grad_norm": 0.1479095071554184,
-      "learning_rate": 2.6990400903444386e-06,
-      "loss": 0.6352,
+      "grad_norm": 0.125027135014534,
+      "learning_rate": 4.0485601355166575e-06,
+      "loss": 0.6233,
       "step": 22980
     },
     {
       "epoch": 4.3271221532091095,
-      "grad_norm": 29.114826202392578,
-      "learning_rate": 2.691511387163561e-06,
-      "loss": 0.6353,
+      "grad_norm": 19.9777889251709,
+      "learning_rate": 4.037267080745342e-06,
+      "loss": 0.5926,
       "step": 22990
     },
     {
       "epoch": 4.329004329004329,
-      "grad_norm": 45.33976364135742,
-      "learning_rate": 2.6839826839826844e-06,
-      "loss": 0.6538,
+      "grad_norm": 20.854936599731445,
+      "learning_rate": 4.025974025974026e-06,
+      "loss": 0.5644,
       "step": 23000
     },
     {
       "epoch": 4.330886504799548,
-      "grad_norm": 19.11794662475586,
-      "learning_rate": 2.676453980801807e-06,
-      "loss": 0.5803,
+      "grad_norm": 7.7878546714782715,
+      "learning_rate": 4.014680971202711e-06,
+      "loss": 0.5073,
       "step": 23010
     },
     {
       "epoch": 4.332768680594768,
-      "grad_norm": 30.116313934326172,
-      "learning_rate": 2.6689252776209303e-06,
-      "loss": 0.2622,
+      "grad_norm": 7.36667013168335,
+      "learning_rate": 4.003387916431395e-06,
+      "loss": 0.1648,
       "step": 23020
     },
     {
       "epoch": 4.334650856389986,
-      "grad_norm": 19.199289321899414,
-      "learning_rate": 2.661396574440053e-06,
-      "loss": 0.506,
+      "grad_norm": 9.753989219665527,
+      "learning_rate": 3.9920948616600785e-06,
+      "loss": 0.4856,
       "step": 23030
     },
     {
       "epoch": 4.336533032185206,
-      "grad_norm": 13.003825187683105,
-      "learning_rate": 2.6538678712591757e-06,
-      "loss": 0.5173,
+      "grad_norm": 14.510024070739746,
+      "learning_rate": 3.980801806888764e-06,
+      "loss": 0.4652,
       "step": 23040
     },
     {
       "epoch": 4.338415207980425,
-      "grad_norm": 18.61155891418457,
-      "learning_rate": 2.6463391680782986e-06,
-      "loss": 0.4607,
+      "grad_norm": 15.549747467041016,
+      "learning_rate": 3.969508752117447e-06,
+      "loss": 0.4127,
       "step": 23050
     },
     {
       "epoch": 4.340297383775645,
-      "grad_norm": 25.670454025268555,
-      "learning_rate": 2.6388104648974216e-06,
-      "loss": 0.6261,
+      "grad_norm": 8.210662841796875,
+      "learning_rate": 3.9582156973461325e-06,
+      "loss": 0.5347,
       "step": 23060
     },
     {
       "epoch": 4.342179559570864,
-      "grad_norm": 10.272873878479004,
-      "learning_rate": 2.631281761716545e-06,
-      "loss": 0.6437,
+      "grad_norm": 7.637692451477051,
+      "learning_rate": 3.946922642574817e-06,
+      "loss": 0.5743,
       "step": 23070
     },
     {
       "epoch": 4.3440617353660835,
-      "grad_norm": 0.25827351212501526,
-      "learning_rate": 2.6237530585356674e-06,
-      "loss": 0.7672,
+      "grad_norm": 0.12690700590610504,
+      "learning_rate": 3.935629587803501e-06,
+      "loss": 0.6929,
       "step": 23080
     },
     {
       "epoch": 4.345943911161302,
-      "grad_norm": 3.8775696754455566,
-      "learning_rate": 2.6162243553547907e-06,
-      "loss": 0.3559,
+      "grad_norm": 4.598480701446533,
+      "learning_rate": 3.924336533032186e-06,
+      "loss": 0.3635,
       "step": 23090
     },
     {
       "epoch": 4.3478260869565215,
-      "grad_norm": 0.6631990075111389,
-      "learning_rate": 2.6086956521739132e-06,
-      "loss": 0.6067,
+      "grad_norm": 2.793907403945923,
+      "learning_rate": 3.913043478260869e-06,
+      "loss": 0.5651,
       "step": 23100
     },
     {
       "epoch": 4.349708262751741,
-      "grad_norm": 7.079066276550293,
-      "learning_rate": 2.601166948993036e-06,
-      "loss": 0.6171,
+      "grad_norm": 32.544918060302734,
+      "learning_rate": 3.9017504234895545e-06,
+      "loss": 0.5491,
       "step": 23110
     },
     {
       "epoch": 4.35159043854696,
-      "grad_norm": 21.724056243896484,
-      "learning_rate": 2.593638245812159e-06,
-      "loss": 0.9693,
+      "grad_norm": 15.509510040283203,
+      "learning_rate": 3.890457368718238e-06,
+      "loss": 1.0166,
       "step": 23120
     },
     {
       "epoch": 4.35347261434218,
-      "grad_norm": 3.3402185440063477,
-      "learning_rate": 2.586109542631282e-06,
-      "loss": 0.6827,
+      "grad_norm": 3.5904457569122314,
+      "learning_rate": 3.879164313946923e-06,
+      "loss": 0.6177,
       "step": 23130
     },
     {
       "epoch": 4.355354790137399,
-      "grad_norm": 13.261208534240723,
-      "learning_rate": 2.578580839450405e-06,
-      "loss": 0.4732,
+      "grad_norm": 7.414836406707764,
+      "learning_rate": 3.867871259175607e-06,
+      "loss": 0.4744,
       "step": 23140
     },
     {
       "epoch": 4.357236965932618,
-      "grad_norm": 48.314605712890625,
-      "learning_rate": 2.571052136269528e-06,
-      "loss": 0.5729,
+      "grad_norm": 6.72654390335083,
+      "learning_rate": 3.856578204404292e-06,
+      "loss": 0.4433,
       "step": 23150
     },
     {
       "epoch": 4.359119141727837,
-      "grad_norm": 2.508441925048828,
-      "learning_rate": 2.5635234330886503e-06,
-      "loss": 0.5245,
+      "grad_norm": 1.880739450454712,
+      "learning_rate": 3.8452851496329755e-06,
+      "loss": 0.4546,
       "step": 23160
     },
     {
       "epoch": 4.361001317523057,
-      "grad_norm": 18.48830223083496,
-      "learning_rate": 2.5559947299077737e-06,
-      "loss": 0.6878,
+      "grad_norm": 15.60661792755127,
+      "learning_rate": 3.83399209486166e-06,
+      "loss": 0.6959,
       "step": 23170
     },
     {
       "epoch": 4.362883493318276,
-      "grad_norm": 0.715785801410675,
-      "learning_rate": 2.548466026726896e-06,
-      "loss": 0.6787,
+      "grad_norm": 0.70307856798172,
+      "learning_rate": 3.822699040090344e-06,
+      "loss": 0.5917,
       "step": 23180
     },
     {
       "epoch": 4.3647656691134955,
-      "grad_norm": 30.262195587158203,
-      "learning_rate": 2.5409373235460195e-06,
-      "loss": 0.6445,
+      "grad_norm": 16.895275115966797,
+      "learning_rate": 3.811405985319029e-06,
+      "loss": 0.6483,
       "step": 23190
     },
     {
       "epoch": 4.366647844908714,
-      "grad_norm": 5.811047554016113,
-      "learning_rate": 2.5334086203651425e-06,
-      "loss": 0.4486,
+      "grad_norm": 14.281275749206543,
+      "learning_rate": 3.8001129305477135e-06,
+      "loss": 0.4527,
       "step": 23200
     },
     {
       "epoch": 4.3685300207039335,
-      "grad_norm": 10.137606620788574,
-      "learning_rate": 2.5258799171842654e-06,
-      "loss": 0.8768,
+      "grad_norm": 16.391109466552734,
+      "learning_rate": 3.7888198757763974e-06,
+      "loss": 0.8592,
       "step": 23210
     },
     {
       "epoch": 4.370412196499153,
-      "grad_norm": 11.782840728759766,
-      "learning_rate": 2.5183512140033883e-06,
-      "loss": 0.8652,
+      "grad_norm": 10.493399620056152,
+      "learning_rate": 3.7775268210050822e-06,
+      "loss": 0.8362,
       "step": 23220
     },
     {
       "epoch": 4.372294372294372,
-      "grad_norm": 2.407595634460449,
-      "learning_rate": 2.510822510822511e-06,
-      "loss": 0.4058,
+      "grad_norm": 4.833158493041992,
+      "learning_rate": 3.766233766233766e-06,
+      "loss": 0.4836,
       "step": 23230
     },
     {
       "epoch": 4.374176548089592,
-      "grad_norm": 14.99952220916748,
-      "learning_rate": 2.503293807641634e-06,
-      "loss": 0.5289,
+      "grad_norm": 19.595304489135742,
+      "learning_rate": 3.754940711462451e-06,
+      "loss": 0.3935,
       "step": 23240
     },
     {
       "epoch": 4.376058723884811,
-      "grad_norm": 15.092422485351562,
-      "learning_rate": 2.4957651044607566e-06,
-      "loss": 0.6724,
+      "grad_norm": 9.765375137329102,
+      "learning_rate": 3.743647656691135e-06,
+      "loss": 0.642,
       "step": 23250
     },
     {
       "epoch": 4.37794089968003,
-      "grad_norm": 56.52970886230469,
-      "learning_rate": 2.48823640127988e-06,
-      "loss": 0.53,
+      "grad_norm": 33.9497184753418,
+      "learning_rate": 3.7323546019198198e-06,
+      "loss": 0.5137,
       "step": 23260
     },
     {
       "epoch": 4.379823075475249,
-      "grad_norm": 16.378936767578125,
-      "learning_rate": 2.480707698099003e-06,
-      "loss": 0.3907,
+      "grad_norm": 13.267662048339844,
+      "learning_rate": 3.7210615471485037e-06,
+      "loss": 0.3194,
       "step": 23270
     },
     {
       "epoch": 4.381705251270469,
-      "grad_norm": 13.952378273010254,
-      "learning_rate": 2.473178994918126e-06,
-      "loss": 0.401,
+      "grad_norm": 25.924053192138672,
+      "learning_rate": 3.709768492377188e-06,
+      "loss": 0.3042,
       "step": 23280
     },
     {
       "epoch": 4.383587427065688,
-      "grad_norm": 5.490935802459717,
-      "learning_rate": 2.4656502917372483e-06,
-      "loss": 0.6424,
+      "grad_norm": 0.388808935880661,
+      "learning_rate": 3.6984754376058725e-06,
+      "loss": 0.5223,
       "step": 23290
     },
     {
       "epoch": 4.3854696028609075,
-      "grad_norm": 37.093711853027344,
-      "learning_rate": 2.4581215885563712e-06,
-      "loss": 0.5544,
+      "grad_norm": 28.93108558654785,
+      "learning_rate": 3.687182382834557e-06,
+      "loss": 0.6084,
       "step": 23300
     },
     {
       "epoch": 4.387351778656127,
-      "grad_norm": 18.15984535217285,
-      "learning_rate": 2.450592885375494e-06,
-      "loss": 0.6462,
+      "grad_norm": 14.811662673950195,
+      "learning_rate": 3.6758893280632412e-06,
+      "loss": 0.522,
       "step": 23310
     },
     {
       "epoch": 4.3892339544513455,
-      "grad_norm": 13.298816680908203,
-      "learning_rate": 2.443064182194617e-06,
-      "loss": 0.6277,
+      "grad_norm": 11.806975364685059,
+      "learning_rate": 3.6645962732919256e-06,
+      "loss": 0.4666,
       "step": 23320
     },
     {
       "epoch": 4.391116130246565,
-      "grad_norm": 11.613605499267578,
-      "learning_rate": 2.43553547901374e-06,
-      "loss": 0.5823,
+      "grad_norm": 76.58401489257812,
+      "learning_rate": 3.65330321852061e-06,
+      "loss": 0.4884,
       "step": 23330
     },
     {
       "epoch": 4.392998306041784,
-      "grad_norm": 1.215378999710083,
-      "learning_rate": 2.428006775832863e-06,
-      "loss": 0.6063,
+      "grad_norm": 0.8018186688423157,
+      "learning_rate": 3.6420101637492944e-06,
+      "loss": 0.5054,
       "step": 23340
     },
     {
       "epoch": 4.394880481837004,
-      "grad_norm": 6.986719608306885,
-      "learning_rate": 2.420478072651986e-06,
-      "loss": 0.4439,
+      "grad_norm": 6.801021099090576,
+      "learning_rate": 3.6307171089779784e-06,
+      "loss": 0.339,
       "step": 23350
     },
     {
       "epoch": 4.396762657632223,
-      "grad_norm": 11.11677360534668,
-      "learning_rate": 2.4129493694711088e-06,
-      "loss": 0.4192,
+      "grad_norm": 12.988116264343262,
+      "learning_rate": 3.6194240542066627e-06,
+      "loss": 0.2921,
       "step": 23360
     },
     {
       "epoch": 4.398644833427442,
-      "grad_norm": 36.418296813964844,
-      "learning_rate": 2.4054206662902317e-06,
-      "loss": 0.5583,
+      "grad_norm": 19.021095275878906,
+      "learning_rate": 3.608130999435347e-06,
+      "loss": 0.4502,
       "step": 23370
     },
     {
       "epoch": 4.400527009222661,
-      "grad_norm": 9.455841064453125,
-      "learning_rate": 2.3978919631093546e-06,
-      "loss": 0.592,
+      "grad_norm": 8.15141487121582,
+      "learning_rate": 3.596837944664032e-06,
+      "loss": 0.6093,
       "step": 23380
     },
     {
       "epoch": 4.402409185017881,
-      "grad_norm": 0.7446449398994446,
-      "learning_rate": 2.3903632599284775e-06,
-      "loss": 0.8016,
+      "grad_norm": 0.6214210987091064,
+      "learning_rate": 3.5855448898927163e-06,
+      "loss": 0.6574,
       "step": 23390
     },
     {
       "epoch": 4.4042913608131,
-      "grad_norm": 19.491920471191406,
-      "learning_rate": 2.3828345567476005e-06,
-      "loss": 0.4662,
+      "grad_norm": 14.339752197265625,
+      "learning_rate": 3.5742518351214007e-06,
+      "loss": 0.4536,
       "step": 23400
     },
     {
       "epoch": 4.4061735366083195,
-      "grad_norm": 22.67009735107422,
-      "learning_rate": 2.3753058535667234e-06,
-      "loss": 0.3276,
+      "grad_norm": 16.84188461303711,
+      "learning_rate": 3.562958780350085e-06,
+      "loss": 0.2564,
       "step": 23410
     },
     {
       "epoch": 4.408055712403539,
-      "grad_norm": 22.89016342163086,
-      "learning_rate": 2.3677771503858463e-06,
-      "loss": 0.5626,
+      "grad_norm": 18.416526794433594,
+      "learning_rate": 3.551665725578769e-06,
+      "loss": 0.4888,
       "step": 23420
     },
     {
       "epoch": 4.409937888198757,
-      "grad_norm": 0.8171205520629883,
-      "learning_rate": 2.3602484472049692e-06,
-      "loss": 0.5954,
+      "grad_norm": 0.2957004904747009,
+      "learning_rate": 3.5403726708074534e-06,
+      "loss": 0.4994,
       "step": 23430
     },
     {
       "epoch": 4.411820063993977,
-      "grad_norm": 15.351773262023926,
-      "learning_rate": 2.352719744024092e-06,
-      "loss": 0.5399,
+      "grad_norm": 14.493985176086426,
+      "learning_rate": 3.529079616036138e-06,
+      "loss": 0.5243,
       "step": 23440
     },
     {
       "epoch": 4.413702239789196,
-      "grad_norm": 30.926944732666016,
-      "learning_rate": 2.345191040843215e-06,
-      "loss": 0.8775,
+      "grad_norm": 22.472604751586914,
+      "learning_rate": 3.517786561264822e-06,
+      "loss": 0.8668,
       "step": 23450
     },
     {
       "epoch": 4.415584415584416,
-      "grad_norm": 1.1383413076400757,
-      "learning_rate": 2.337662337662338e-06,
-      "loss": 0.6176,
+      "grad_norm": 1.9218357801437378,
+      "learning_rate": 3.5064935064935066e-06,
+      "loss": 0.55,
       "step": 23460
     },
     {
       "epoch": 4.417466591379635,
-      "grad_norm": 13.357398986816406,
-      "learning_rate": 2.3301336344814605e-06,
-      "loss": 0.6766,
+      "grad_norm": 15.064146995544434,
+      "learning_rate": 3.495200451722191e-06,
+      "loss": 0.6122,
       "step": 23470
     },
     {
       "epoch": 4.419348767174855,
-      "grad_norm": 32.312835693359375,
-      "learning_rate": 2.3226049313005834e-06,
-      "loss": 0.4389,
+      "grad_norm": 29.05467987060547,
+      "learning_rate": 3.4839073969508753e-06,
+      "loss": 0.2962,
       "step": 23480
     },
     {
       "epoch": 4.421230942970073,
-      "grad_norm": 11.716819763183594,
-      "learning_rate": 2.3150762281197063e-06,
-      "loss": 0.4396,
+      "grad_norm": 10.816146850585938,
+      "learning_rate": 3.4726143421795593e-06,
+      "loss": 0.4214,
       "step": 23490
     },
     {
       "epoch": 4.423113118765293,
-      "grad_norm": 5.9410576820373535,
-      "learning_rate": 2.3075475249388297e-06,
-      "loss": 0.4957,
+      "grad_norm": 8.355408668518066,
+      "learning_rate": 3.461321287408244e-06,
+      "loss": 0.5314,
       "step": 23500
     },
     {
       "epoch": 4.424995294560512,
-      "grad_norm": 13.58919906616211,
-      "learning_rate": 2.3000188217579526e-06,
-      "loss": 0.7802,
+      "grad_norm": 13.588001251220703,
+      "learning_rate": 3.4500282326369285e-06,
+      "loss": 0.8294,
       "step": 23510
     },
     {
       "epoch": 4.426877470355731,
-      "grad_norm": 18.13483428955078,
-      "learning_rate": 2.2924901185770755e-06,
-      "loss": 0.7189,
+      "grad_norm": 20.003461837768555,
+      "learning_rate": 3.438735177865613e-06,
+      "loss": 0.7186,
       "step": 23520
     },
     {
       "epoch": 4.428759646150951,
-      "grad_norm": 5.589889049530029,
-      "learning_rate": 2.2849614153961984e-06,
-      "loss": 0.4137,
+      "grad_norm": 5.769153594970703,
+      "learning_rate": 3.4274421230942972e-06,
+      "loss": 0.3721,
       "step": 23530
     },
     {
       "epoch": 4.430641821946169,
-      "grad_norm": 13.841676712036133,
-      "learning_rate": 2.277432712215321e-06,
-      "loss": 0.6273,
+      "grad_norm": 5.754838466644287,
+      "learning_rate": 3.4161490683229816e-06,
+      "loss": 0.5478,
       "step": 23540
     },
     {
       "epoch": 4.432523997741389,
-      "grad_norm": 0.4990882873535156,
-      "learning_rate": 2.269904009034444e-06,
-      "loss": 0.4074,
+      "grad_norm": 0.23657679557800293,
+      "learning_rate": 3.404856013551666e-06,
+      "loss": 0.3118,
       "step": 23550
     },
     {
       "epoch": 4.434406173536608,
-      "grad_norm": 8.081299781799316,
-      "learning_rate": 2.2623753058535668e-06,
-      "loss": 0.5185,
+      "grad_norm": 15.234368324279785,
+      "learning_rate": 3.3935629587803504e-06,
+      "loss": 0.4746,
       "step": 23560
     },
     {
       "epoch": 4.436288349331828,
-      "grad_norm": 3.278000831604004,
-      "learning_rate": 2.2548466026726897e-06,
-      "loss": 0.6386,
+      "grad_norm": 2.7576918601989746,
+      "learning_rate": 3.3822699040090343e-06,
+      "loss": 0.6348,
       "step": 23570
     },
     {
       "epoch": 4.438170525127047,
-      "grad_norm": 21.62383270263672,
-      "learning_rate": 2.2473178994918126e-06,
-      "loss": 0.5482,
+      "grad_norm": 20.79529571533203,
+      "learning_rate": 3.3709768492377187e-06,
+      "loss": 0.4821,
       "step": 23580
     },
     {
       "epoch": 4.440052700922266,
-      "grad_norm": 44.334747314453125,
-      "learning_rate": 2.2397891963109355e-06,
-      "loss": 0.576,
+      "grad_norm": 15.99216365814209,
+      "learning_rate": 3.359683794466403e-06,
+      "loss": 0.4527,
       "step": 23590
     },
     {
       "epoch": 4.441934876717485,
-      "grad_norm": 13.59565258026123,
-      "learning_rate": 2.2322604931300585e-06,
-      "loss": 0.5771,
+      "grad_norm": 11.632766723632812,
+      "learning_rate": 3.3483907396950875e-06,
+      "loss": 0.4984,
       "step": 23600
     },
     {
       "epoch": 4.4438170525127045,
-      "grad_norm": 19.645078659057617,
-      "learning_rate": 2.2247317899491814e-06,
-      "loss": 0.7014,
+      "grad_norm": 13.765467643737793,
+      "learning_rate": 3.337097684923772e-06,
+      "loss": 0.6588,
       "step": 23610
     },
     {
       "epoch": 4.445699228307924,
-      "grad_norm": 7.434754371643066,
-      "learning_rate": 2.2172030867683043e-06,
-      "loss": 0.4733,
+      "grad_norm": 2.9188790321350098,
+      "learning_rate": 3.3258046301524562e-06,
+      "loss": 0.5091,
       "step": 23620
     },
     {
       "epoch": 4.447581404103143,
-      "grad_norm": 49.53765869140625,
-      "learning_rate": 2.2096743835874272e-06,
-      "loss": 0.7781,
+      "grad_norm": 33.282928466796875,
+      "learning_rate": 3.314511575381141e-06,
+      "loss": 0.665,
       "step": 23630
     },
     {
       "epoch": 4.449463579898363,
-      "grad_norm": 2.3648738861083984,
-      "learning_rate": 2.20214568040655e-06,
-      "loss": 0.6185,
+      "grad_norm": 0.6511355042457581,
+      "learning_rate": 3.303218520609825e-06,
+      "loss": 0.5527,
       "step": 23640
     },
     {
       "epoch": 4.451345755693581,
-      "grad_norm": 9.881647109985352,
-      "learning_rate": 2.194616977225673e-06,
-      "loss": 0.7099,
+      "grad_norm": 12.938034057617188,
+      "learning_rate": 3.2919254658385094e-06,
+      "loss": 0.5798,
       "step": 23650
     },
     {
       "epoch": 4.453227931488801,
-      "grad_norm": 15.192410469055176,
-      "learning_rate": 2.187088274044796e-06,
-      "loss": 0.5284,
+      "grad_norm": 19.567174911499023,
+      "learning_rate": 3.2806324110671938e-06,
+      "loss": 0.624,
       "step": 23660
     },
     {
       "epoch": 4.45511010728402,
-      "grad_norm": 0.09341107308864594,
-      "learning_rate": 2.179559570863919e-06,
-      "loss": 0.2563,
+      "grad_norm": 0.04238367825746536,
+      "learning_rate": 3.269339356295878e-06,
+      "loss": 0.2131,
       "step": 23670
     },
     {
       "epoch": 4.45699228307924,
-      "grad_norm": 17.3346004486084,
-      "learning_rate": 2.172030867683042e-06,
-      "loss": 0.5827,
+      "grad_norm": 14.076987266540527,
+      "learning_rate": 3.2580463015245625e-06,
+      "loss": 0.6618,
       "step": 23680
     },
     {
       "epoch": 4.458874458874459,
-      "grad_norm": 12.81985855102539,
-      "learning_rate": 2.1645021645021648e-06,
-      "loss": 0.476,
+      "grad_norm": 7.118105888366699,
+      "learning_rate": 3.246753246753247e-06,
+      "loss": 0.4138,
       "step": 23690
     },
     {
       "epoch": 4.4607566346696785,
-      "grad_norm": 11.144566535949707,
-      "learning_rate": 2.1569734613212877e-06,
-      "loss": 0.4212,
+      "grad_norm": 13.331232070922852,
+      "learning_rate": 3.2354601919819313e-06,
+      "loss": 0.3546,
       "step": 23700
     },
     {
       "epoch": 4.462638810464897,
-      "grad_norm": 30.27758026123047,
-      "learning_rate": 2.1494447581404106e-06,
-      "loss": 0.5779,
+      "grad_norm": 19.263700485229492,
+      "learning_rate": 3.2241671372106153e-06,
+      "loss": 0.4596,
       "step": 23710
     },
     {
       "epoch": 4.4645209862601165,
-      "grad_norm": 29.81888771057129,
-      "learning_rate": 2.1419160549595335e-06,
-      "loss": 0.7226,
+      "grad_norm": 23.13578987121582,
+      "learning_rate": 3.2128740824392996e-06,
+      "loss": 0.6003,
       "step": 23720
     },
     {
       "epoch": 4.466403162055336,
-      "grad_norm": 0.8965347409248352,
-      "learning_rate": 2.134387351778656e-06,
-      "loss": 0.7259,
+      "grad_norm": 2.132723331451416,
+      "learning_rate": 3.201581027667984e-06,
+      "loss": 0.6584,
       "step": 23730
     },
     {
       "epoch": 4.468285337850555,
-      "grad_norm": 12.240394592285156,
-      "learning_rate": 2.126858648597779e-06,
-      "loss": 0.3844,
+      "grad_norm": 10.117582321166992,
+      "learning_rate": 3.1902879728966684e-06,
+      "loss": 0.3707,
       "step": 23740
     },
     {
       "epoch": 4.470167513645775,
-      "grad_norm": 1.0969513654708862,
-      "learning_rate": 2.1193299454169023e-06,
-      "loss": 0.4925,
+      "grad_norm": 1.6187089681625366,
+      "learning_rate": 3.1789949181253532e-06,
+      "loss": 0.3961,
       "step": 23750
     },
     {
       "epoch": 4.472049689440993,
-      "grad_norm": 6.099503040313721,
-      "learning_rate": 2.111801242236025e-06,
-      "loss": 0.4466,
+      "grad_norm": 8.55710506439209,
+      "learning_rate": 3.1677018633540376e-06,
+      "loss": 0.3721,
       "step": 23760
     },
     {
       "epoch": 4.473931865236213,
-      "grad_norm": 15.072311401367188,
-      "learning_rate": 2.104272539055148e-06,
-      "loss": 0.7378,
+      "grad_norm": 11.327536582946777,
+      "learning_rate": 3.156408808582722e-06,
+      "loss": 0.6747,
       "step": 23770
     },
     {
       "epoch": 4.475814041031432,
-      "grad_norm": 38.78789520263672,
-      "learning_rate": 2.096743835874271e-06,
-      "loss": 0.7411,
+      "grad_norm": 7.744390487670898,
+      "learning_rate": 3.1451157538114064e-06,
+      "loss": 0.6423,
       "step": 23780
     },
     {
       "epoch": 4.477696216826652,
-      "grad_norm": 3.0044000148773193,
-      "learning_rate": 2.089215132693394e-06,
-      "loss": 0.3366,
+      "grad_norm": 1.3057587146759033,
+      "learning_rate": 3.1338226990400903e-06,
+      "loss": 0.2579,
       "step": 23790
     },
     {
       "epoch": 4.479578392621871,
-      "grad_norm": 17.955310821533203,
-      "learning_rate": 2.0816864295125165e-06,
-      "loss": 0.5898,
+      "grad_norm": 10.926691055297852,
+      "learning_rate": 3.1225296442687747e-06,
+      "loss": 0.5395,
       "step": 23800
     },
     {
       "epoch": 4.4814605684170905,
-      "grad_norm": 7.283894062042236,
-      "learning_rate": 2.0741577263316394e-06,
-      "loss": 0.6196,
+      "grad_norm": 13.025545120239258,
+      "learning_rate": 3.111236589497459e-06,
+      "loss": 0.5213,
       "step": 23810
     },
     {
       "epoch": 4.483342744212309,
-      "grad_norm": 20.742965698242188,
-      "learning_rate": 2.0666290231507623e-06,
-      "loss": 0.563,
+      "grad_norm": 8.658554077148438,
+      "learning_rate": 3.0999435347261435e-06,
+      "loss": 0.522,
       "step": 23820
     },
     {
       "epoch": 4.4852249200075285,
-      "grad_norm": 0.6865695714950562,
-      "learning_rate": 2.0591003199698852e-06,
-      "loss": 0.3824,
+      "grad_norm": 0.7279936671257019,
+      "learning_rate": 3.088650479954828e-06,
+      "loss": 0.3601,
       "step": 23830
     },
     {
       "epoch": 4.487107095802748,
-      "grad_norm": 18.67192268371582,
-      "learning_rate": 2.051571616789008e-06,
-      "loss": 0.63,
+      "grad_norm": 11.042409896850586,
+      "learning_rate": 3.0773574251835122e-06,
+      "loss": 0.5312,
       "step": 23840
     },
     {
       "epoch": 4.488989271597967,
-      "grad_norm": 5.289563179016113,
-      "learning_rate": 2.044042913608131e-06,
-      "loss": 0.6516,
+      "grad_norm": 2.8644516468048096,
+      "learning_rate": 3.0660643704121966e-06,
+      "loss": 0.5416,
       "step": 23850
     },
     {
       "epoch": 4.490871447393187,
-      "grad_norm": 15.691003799438477,
-      "learning_rate": 2.036514210427254e-06,
-      "loss": 0.4024,
+      "grad_norm": 16.289228439331055,
+      "learning_rate": 3.0547713156408806e-06,
+      "loss": 0.37,
       "step": 23860
     },
     {
       "epoch": 4.492753623188406,
-      "grad_norm": 22.759946823120117,
-      "learning_rate": 2.028985507246377e-06,
-      "loss": 0.9504,
+      "grad_norm": 25.66763687133789,
+      "learning_rate": 3.0434782608695654e-06,
+      "loss": 0.9307,
       "step": 23870
     },
     {
       "epoch": 4.494635798983625,
-      "grad_norm": 1.0771783590316772,
-      "learning_rate": 2.0214568040655e-06,
-      "loss": 0.4739,
+      "grad_norm": 3.9312744140625,
+      "learning_rate": 3.0321852060982498e-06,
+      "loss": 0.3846,
       "step": 23880
     },
     {
       "epoch": 4.496517974778844,
-      "grad_norm": 4.770224094390869,
-      "learning_rate": 2.0139281008846228e-06,
-      "loss": 0.8352,
+      "grad_norm": 2.878774404525757,
+      "learning_rate": 3.020892151326934e-06,
+      "loss": 0.6847,
       "step": 23890
     },
     {
       "epoch": 4.498400150574064,
-      "grad_norm": 13.002606391906738,
-      "learning_rate": 2.0063993977037457e-06,
-      "loss": 0.4923,
+      "grad_norm": 8.510382652282715,
+      "learning_rate": 3.0095990965556185e-06,
+      "loss": 0.4522,
       "step": 23900
     },
     {
       "epoch": 4.500282326369283,
-      "grad_norm": 5.516331672668457,
-      "learning_rate": 1.9988706945228686e-06,
-      "loss": 0.4354,
+      "grad_norm": 9.131433486938477,
+      "learning_rate": 2.998306041784303e-06,
+      "loss": 0.3852,
       "step": 23910
     },
     {
       "epoch": 4.5021645021645025,
-      "grad_norm": 16.686582565307617,
-      "learning_rate": 1.9913419913419915e-06,
-      "loss": 0.638,
+      "grad_norm": 14.799365043640137,
+      "learning_rate": 2.9870129870129873e-06,
+      "loss": 0.3988,
       "step": 23920
     },
     {
       "epoch": 4.504046677959721,
-      "grad_norm": 25.16171646118164,
-      "learning_rate": 1.9838132881611144e-06,
-      "loss": 0.649,
+      "grad_norm": 25.332460403442383,
+      "learning_rate": 2.9757199322416712e-06,
+      "loss": 0.5088,
       "step": 23930
     },
     {
       "epoch": 4.5059288537549405,
-      "grad_norm": 22.84310531616211,
-      "learning_rate": 1.9762845849802374e-06,
-      "loss": 0.7048,
+      "grad_norm": 15.210441589355469,
+      "learning_rate": 2.9644268774703556e-06,
+      "loss": 0.5733,
       "step": 23940
     },
     {
       "epoch": 4.50781102955016,
-      "grad_norm": 11.860065460205078,
-      "learning_rate": 1.9687558817993603e-06,
-      "loss": 0.8613,
+      "grad_norm": 17.5853271484375,
+      "learning_rate": 2.95313382269904e-06,
+      "loss": 0.8435,
       "step": 23950
     },
     {
       "epoch": 4.509693205345379,
-      "grad_norm": 8.635574340820312,
-      "learning_rate": 1.961227178618483e-06,
-      "loss": 0.5576,
+      "grad_norm": 5.9127020835876465,
+      "learning_rate": 2.9418407679277244e-06,
+      "loss": 0.405,
       "step": 23960
     },
     {
       "epoch": 4.511575381140599,
-      "grad_norm": 11.803613662719727,
-      "learning_rate": 1.953698475437606e-06,
-      "loss": 0.3823,
+      "grad_norm": 15.873945236206055,
+      "learning_rate": 2.9305477131564088e-06,
+      "loss": 0.3017,
       "step": 23970
     },
     {
       "epoch": 4.513457556935818,
-      "grad_norm": 4.785434246063232,
-      "learning_rate": 1.9461697722567286e-06,
-      "loss": 0.9674,
+      "grad_norm": 6.168259620666504,
+      "learning_rate": 2.919254658385093e-06,
+      "loss": 0.8213,
       "step": 23980
     },
     {
       "epoch": 4.515339732731037,
-      "grad_norm": 135.4927215576172,
-      "learning_rate": 1.938641069075852e-06,
-      "loss": 0.7317,
+      "grad_norm": 14.95897102355957,
+      "learning_rate": 2.907961603613778e-06,
+      "loss": 0.7194,
       "step": 23990
     },
     {
       "epoch": 4.517221908526256,
-      "grad_norm": 11.872318267822266,
-      "learning_rate": 1.931112365894975e-06,
-      "loss": 0.485,
+      "grad_norm": 9.933831214904785,
+      "learning_rate": 2.896668548842462e-06,
+      "loss": 0.4408,
       "step": 24000
     },
     {
       "epoch": 4.519104084321476,
-      "grad_norm": 0.5789291858673096,
-      "learning_rate": 1.923583662714098e-06,
-      "loss": 0.5606,
+      "grad_norm": 0.1559751182794571,
+      "learning_rate": 2.8853754940711463e-06,
+      "loss": 0.5496,
       "step": 24010
     },
     {
       "epoch": 4.520986260116695,
-      "grad_norm": 7.3305439949035645,
-      "learning_rate": 1.9160549595332207e-06,
-      "loss": 0.6314,
+      "grad_norm": 16.345056533813477,
+      "learning_rate": 2.8740824392998307e-06,
+      "loss": 0.5864,
       "step": 24020
     },
     {
       "epoch": 4.5228684359119145,
-      "grad_norm": 13.919968605041504,
-      "learning_rate": 1.9085262563523437e-06,
-      "loss": 0.6887,
+      "grad_norm": 28.054203033447266,
+      "learning_rate": 2.862789384528515e-06,
+      "loss": 0.7297,
       "step": 24030
     },
     {
       "epoch": 4.524750611707134,
-      "grad_norm": 1.8676209449768066,
-      "learning_rate": 1.9009975531714664e-06,
-      "loss": 0.5487,
+      "grad_norm": 14.966489791870117,
+      "learning_rate": 2.8514963297571994e-06,
+      "loss": 0.3968,
       "step": 24040
     },
     {
       "epoch": 4.526632787502352,
-      "grad_norm": 22.899232864379883,
-      "learning_rate": 1.8934688499905893e-06,
-      "loss": 0.7641,
+      "grad_norm": 20.097759246826172,
+      "learning_rate": 2.840203274985884e-06,
+      "loss": 0.7483,
       "step": 24050
     },
     {
       "epoch": 4.528514963297572,
-      "grad_norm": 5.90631103515625,
-      "learning_rate": 1.8859401468097122e-06,
-      "loss": 0.7595,
+      "grad_norm": 19.95496940612793,
+      "learning_rate": 2.828910220214568e-06,
+      "loss": 0.6654,
       "step": 24060
     },
     {
       "epoch": 4.530397139092791,
-      "grad_norm": 9.364104270935059,
-      "learning_rate": 1.8784114436288351e-06,
-      "loss": 0.2766,
+      "grad_norm": 8.588127136230469,
+      "learning_rate": 2.8176171654432526e-06,
+      "loss": 0.2225,
       "step": 24070
     },
     {
       "epoch": 4.532279314888011,
-      "grad_norm": 57.38972854614258,
-      "learning_rate": 1.8708827404479578e-06,
-      "loss": 0.6298,
+      "grad_norm": 10.798761367797852,
+      "learning_rate": 2.8063241106719366e-06,
+      "loss": 0.5675,
       "step": 24080
     },
     {
       "epoch": 4.53416149068323,
-      "grad_norm": 0.9203323125839233,
-      "learning_rate": 1.8633540372670808e-06,
-      "loss": 0.6828,
+      "grad_norm": 0.21323099732398987,
+      "learning_rate": 2.795031055900621e-06,
+      "loss": 0.4654,
       "step": 24090
     },
     {
       "epoch": 4.536043666478449,
-      "grad_norm": 16.27019500732422,
-      "learning_rate": 1.8558253340862037e-06,
-      "loss": 0.8687,
+      "grad_norm": 12.417448997497559,
+      "learning_rate": 2.7837380011293053e-06,
+      "loss": 0.7423,
       "step": 24100
     },
     {
       "epoch": 4.537925842273668,
-      "grad_norm": 10.179085731506348,
-      "learning_rate": 1.8482966309053266e-06,
-      "loss": 0.5654,
+      "grad_norm": 8.83974838256836,
+      "learning_rate": 2.7724449463579897e-06,
+      "loss": 0.3925,
       "step": 24110
     },
     {
       "epoch": 4.539808018068888,
-      "grad_norm": 31.10147476196289,
-      "learning_rate": 1.8407679277244497e-06,
-      "loss": 0.6024,
+      "grad_norm": 28.08986473083496,
+      "learning_rate": 2.7611518915866745e-06,
+      "loss": 0.5835,
       "step": 24120
     },
     {
       "epoch": 4.541690193864107,
-      "grad_norm": 5.746549606323242,
-      "learning_rate": 1.8332392245435727e-06,
-      "loss": 0.6073,
+      "grad_norm": 2.058382034301758,
+      "learning_rate": 2.749858836815359e-06,
+      "loss": 0.5558,
       "step": 24130
     },
     {
       "epoch": 4.543572369659326,
-      "grad_norm": 18.98149299621582,
-      "learning_rate": 1.8257105213626956e-06,
-      "loss": 0.6642,
+      "grad_norm": 11.856590270996094,
+      "learning_rate": 2.7385657820440433e-06,
+      "loss": 0.6624,
       "step": 24140
     },
     {
       "epoch": 4.545454545454545,
-      "grad_norm": 11.173285484313965,
-      "learning_rate": 1.8181818181818183e-06,
-      "loss": 0.4425,
+      "grad_norm": 28.79555320739746,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.4069,
       "step": 24150
     },
     {
       "epoch": 4.547336721249764,
-      "grad_norm": 68.92263793945312,
-      "learning_rate": 1.8106531150009412e-06,
-      "loss": 0.4738,
+      "grad_norm": 0.156916081905365,
+      "learning_rate": 2.7159796725014116e-06,
+      "loss": 0.3982,
       "step": 24160
     },
     {
       "epoch": 4.549218897044984,
-      "grad_norm": 17.539867401123047,
-      "learning_rate": 1.8031244118200641e-06,
-      "loss": 0.6192,
+      "grad_norm": 22.804506301879883,
+      "learning_rate": 2.704686617730096e-06,
+      "loss": 0.5942,
       "step": 24170
     },
     {
       "epoch": 4.551101072840203,
-      "grad_norm": 22.743282318115234,
-      "learning_rate": 1.795595708639187e-06,
-      "loss": 0.7029,
+      "grad_norm": 12.148880004882812,
+      "learning_rate": 2.6933935629587804e-06,
+      "loss": 0.6395,
       "step": 24180
     },
     {
       "epoch": 4.552983248635423,
-      "grad_norm": 14.960015296936035,
-      "learning_rate": 1.78806700545831e-06,
-      "loss": 0.3663,
+      "grad_norm": 17.041955947875977,
+      "learning_rate": 2.6821005081874648e-06,
+      "loss": 0.349,
       "step": 24190
     },
     {
       "epoch": 4.554865424430642,
-      "grad_norm": 16.499238967895508,
-      "learning_rate": 1.780538302277433e-06,
-      "loss": 0.5681,
+      "grad_norm": 15.154557228088379,
+      "learning_rate": 2.670807453416149e-06,
+      "loss": 0.5323,
       "step": 24200
     },
     {
       "epoch": 4.556747600225862,
-      "grad_norm": 7.2052388191223145,
-      "learning_rate": 1.7730095990965556e-06,
-      "loss": 0.3522,
+      "grad_norm": 5.193390369415283,
+      "learning_rate": 2.6595143986448335e-06,
+      "loss": 0.306,
       "step": 24210
     },
     {
       "epoch": 4.55862977602108,
-      "grad_norm": 10.154321670532227,
-      "learning_rate": 1.7654808959156785e-06,
-      "loss": 0.6202,
+      "grad_norm": 14.101144790649414,
+      "learning_rate": 2.648221343873518e-06,
+      "loss": 0.5806,
       "step": 24220
     },
     {
       "epoch": 4.5605119518162995,
-      "grad_norm": 3.8168325424194336,
-      "learning_rate": 1.7579521927348015e-06,
-      "loss": 0.5148,
+      "grad_norm": 4.696810722351074,
+      "learning_rate": 2.636928289102202e-06,
+      "loss": 0.4525,
       "step": 24230
     },
     {
       "epoch": 4.562394127611519,
-      "grad_norm": 5.976160526275635,
-      "learning_rate": 1.7504234895539246e-06,
-      "loss": 0.6931,
+      "grad_norm": 2.615445613861084,
+      "learning_rate": 2.6256352343308867e-06,
+      "loss": 0.4967,
       "step": 24240
     },
     {
       "epoch": 4.564276303406738,
-      "grad_norm": 10.809183120727539,
-      "learning_rate": 1.7428947863730475e-06,
-      "loss": 0.6397,
+      "grad_norm": 35.27506637573242,
+      "learning_rate": 2.614342179559571e-06,
+      "loss": 0.4971,
       "step": 24250
     },
     {
       "epoch": 4.566158479201958,
-      "grad_norm": 6.282565116882324,
-      "learning_rate": 1.7353660831921704e-06,
-      "loss": 0.4463,
+      "grad_norm": 8.616056442260742,
+      "learning_rate": 2.6030491247882554e-06,
+      "loss": 0.3323,
       "step": 24260
     },
     {
       "epoch": 4.568040654997176,
-      "grad_norm": 1.7869082689285278,
-      "learning_rate": 1.7278373800112933e-06,
-      "loss": 0.3967,
+      "grad_norm": 1.696595311164856,
+      "learning_rate": 2.59175607001694e-06,
+      "loss": 0.343,
       "step": 24270
     },
     {
       "epoch": 4.569922830792396,
-      "grad_norm": 7.395685195922852,
-      "learning_rate": 1.720308676830416e-06,
-      "loss": 0.5906,
+      "grad_norm": 8.711554527282715,
+      "learning_rate": 2.580463015245624e-06,
+      "loss": 0.5896,
       "step": 24280
     },
     {
       "epoch": 4.571805006587615,
-      "grad_norm": 14.580101013183594,
-      "learning_rate": 1.712779973649539e-06,
-      "loss": 0.5194,
+      "grad_norm": 15.512261390686035,
+      "learning_rate": 2.5691699604743086e-06,
+      "loss": 0.4962,
       "step": 24290
     },
     {
       "epoch": 4.573687182382835,
-      "grad_norm": 29.10459327697754,
-      "learning_rate": 1.705251270468662e-06,
-      "loss": 0.5974,
+      "grad_norm": 10.31464672088623,
+      "learning_rate": 2.5578769057029925e-06,
+      "loss": 0.5446,
       "step": 24300
     },
     {
       "epoch": 4.575569358178054,
-      "grad_norm": 15.645312309265137,
-      "learning_rate": 1.6977225672877848e-06,
-      "loss": 0.5314,
+      "grad_norm": 68.28598022460938,
+      "learning_rate": 2.546583850931677e-06,
+      "loss": 0.5234,
       "step": 24310
     },
     {
       "epoch": 4.577451533973273,
-      "grad_norm": 20.30552101135254,
-      "learning_rate": 1.6901938641069077e-06,
-      "loss": 0.557,
+      "grad_norm": 23.362018585205078,
+      "learning_rate": 2.5352907961603613e-06,
+      "loss": 0.4662,
       "step": 24320
     },
     {
       "epoch": 4.579333709768492,
-      "grad_norm": 0.3066312074661255,
-      "learning_rate": 1.6826651609260305e-06,
-      "loss": 0.4308,
+      "grad_norm": 0.38361719250679016,
+      "learning_rate": 2.5239977413890457e-06,
+      "loss": 0.3885,
       "step": 24330
     },
     {
       "epoch": 4.5812158855637115,
-      "grad_norm": 17.451622009277344,
-      "learning_rate": 1.6751364577451534e-06,
-      "loss": 0.2231,
+      "grad_norm": 13.636258125305176,
+      "learning_rate": 2.51270468661773e-06,
+      "loss": 0.1951,
       "step": 24340
     },
     {
       "epoch": 4.583098061358931,
-      "grad_norm": 20.441179275512695,
-      "learning_rate": 1.6676077545642763e-06,
-      "loss": 0.4294,
+      "grad_norm": 13.915003776550293,
+      "learning_rate": 2.5014116318464144e-06,
+      "loss": 0.2864,
       "step": 24350
     },
     {
       "epoch": 4.58498023715415,
-      "grad_norm": 25.281518936157227,
-      "learning_rate": 1.6600790513833994e-06,
-      "loss": 0.4772,
+      "grad_norm": 28.40807342529297,
+      "learning_rate": 2.4901185770750993e-06,
+      "loss": 0.4974,
       "step": 24360
     },
     {
       "epoch": 4.58686241294937,
-      "grad_norm": 21.41700553894043,
-      "learning_rate": 1.6525503482025224e-06,
-      "loss": 0.7507,
+      "grad_norm": 21.473955154418945,
+      "learning_rate": 2.478825522303783e-06,
+      "loss": 0.6872,
       "step": 24370
     },
     {
       "epoch": 4.588744588744589,
-      "grad_norm": 10.977895736694336,
-      "learning_rate": 1.6450216450216453e-06,
-      "loss": 0.4713,
+      "grad_norm": 39.9091911315918,
+      "learning_rate": 2.4675324675324676e-06,
+      "loss": 0.4134,
       "step": 24380
     },
     {
       "epoch": 4.590626764539808,
-      "grad_norm": 2.4768362045288086,
-      "learning_rate": 1.6374929418407682e-06,
-      "loss": 0.5988,
+      "grad_norm": 18.155563354492188,
+      "learning_rate": 2.456239412761152e-06,
+      "loss": 0.6036,
       "step": 24390
     },
     {
       "epoch": 4.592508940335027,
-      "grad_norm": 11.520479202270508,
-      "learning_rate": 1.629964238659891e-06,
-      "loss": 0.6857,
+      "grad_norm": 9.114380836486816,
+      "learning_rate": 2.4449463579898364e-06,
+      "loss": 0.6799,
       "step": 24400
     },
     {
       "epoch": 4.594391116130247,
-      "grad_norm": 14.980639457702637,
-      "learning_rate": 1.6224355354790138e-06,
-      "loss": 0.7079,
+      "grad_norm": 6.3043904304504395,
+      "learning_rate": 2.4336533032185207e-06,
+      "loss": 0.626,
       "step": 24410
     },
     {
       "epoch": 4.596273291925466,
-      "grad_norm": 15.762008666992188,
-      "learning_rate": 1.6149068322981367e-06,
-      "loss": 0.4411,
+      "grad_norm": 16.44568634033203,
+      "learning_rate": 2.422360248447205e-06,
+      "loss": 0.3922,
       "step": 24420
     },
     {
       "epoch": 4.5981554677206855,
-      "grad_norm": 19.70981216430664,
-      "learning_rate": 1.6073781291172597e-06,
-      "loss": 0.8163,
+      "grad_norm": 9.025197982788086,
+      "learning_rate": 2.4110671936758895e-06,
+      "loss": 0.7261,
       "step": 24430
     },
     {
       "epoch": 4.600037643515904,
-      "grad_norm": 21.06952667236328,
-      "learning_rate": 1.5998494259363826e-06,
-      "loss": 0.4246,
+      "grad_norm": 19.083044052124023,
+      "learning_rate": 2.399774138904574e-06,
+      "loss": 0.4588,
       "step": 24440
     },
     {
       "epoch": 4.6019198193111235,
-      "grad_norm": 12.949865341186523,
-      "learning_rate": 1.5923207227555055e-06,
-      "loss": 0.4607,
+      "grad_norm": 16.03182029724121,
+      "learning_rate": 2.388481084133258e-06,
+      "loss": 0.4318,
       "step": 24450
     },
     {
       "epoch": 4.603801995106343,
-      "grad_norm": 5.208014965057373,
-      "learning_rate": 1.5847920195746282e-06,
-      "loss": 0.6787,
+      "grad_norm": 1.4416394233703613,
+      "learning_rate": 2.3771880293619422e-06,
+      "loss": 0.658,
       "step": 24460
     },
     {
       "epoch": 4.605684170901562,
-      "grad_norm": 15.924193382263184,
-      "learning_rate": 1.5772633163937511e-06,
-      "loss": 0.5571,
+      "grad_norm": 7.5059075355529785,
+      "learning_rate": 2.3658949745906266e-06,
+      "loss": 0.4453,
       "step": 24470
     },
     {
       "epoch": 4.607566346696782,
-      "grad_norm": 5.009922981262207,
-      "learning_rate": 1.5697346132128743e-06,
-      "loss": 0.7348,
+      "grad_norm": 4.021427631378174,
+      "learning_rate": 2.3546019198193114e-06,
+      "loss": 0.5636,
       "step": 24480
     },
     {
       "epoch": 4.609448522492,
-      "grad_norm": 6.933772563934326,
-      "learning_rate": 1.5622059100319972e-06,
-      "loss": 0.4897,
+      "grad_norm": 5.8382744789123535,
+      "learning_rate": 2.343308865047996e-06,
+      "loss": 0.2773,
       "step": 24490
     },
     {
       "epoch": 4.61133069828722,
-      "grad_norm": 12.554123878479004,
-      "learning_rate": 1.5546772068511201e-06,
-      "loss": 0.5164,
+      "grad_norm": 10.830999374389648,
+      "learning_rate": 2.33201581027668e-06,
+      "loss": 0.5919,
       "step": 24500
     },
     {
       "epoch": 4.613212874082439,
-      "grad_norm": 30.179964065551758,
-      "learning_rate": 1.547148503670243e-06,
-      "loss": 0.4959,
+      "grad_norm": 30.35820198059082,
+      "learning_rate": 2.3207227555053646e-06,
+      "loss": 0.3376,
       "step": 24510
     },
     {
       "epoch": 4.615095049877659,
-      "grad_norm": 16.29331398010254,
-      "learning_rate": 1.539619800489366e-06,
-      "loss": 0.9734,
+      "grad_norm": 17.739852905273438,
+      "learning_rate": 2.3094297007340485e-06,
+      "loss": 0.9373,
       "step": 24520
     },
     {
       "epoch": 4.616977225672878,
-      "grad_norm": 11.80064868927002,
-      "learning_rate": 1.5320910973084887e-06,
-      "loss": 0.4734,
+      "grad_norm": 9.197490692138672,
+      "learning_rate": 2.298136645962733e-06,
+      "loss": 0.3972,
       "step": 24530
     },
     {
       "epoch": 4.6188594014680975,
-      "grad_norm": 22.42782974243164,
-      "learning_rate": 1.5245623941276116e-06,
-      "loss": 0.4877,
+      "grad_norm": 16.6794490814209,
+      "learning_rate": 2.2868435911914173e-06,
+      "loss": 0.3707,
       "step": 24540
     },
     {
       "epoch": 4.620741577263316,
-      "grad_norm": 0.34393635392189026,
-      "learning_rate": 1.5170336909467345e-06,
-      "loss": 0.4987,
+      "grad_norm": 0.1481352001428604,
+      "learning_rate": 2.2755505364201017e-06,
+      "loss": 0.4161,
       "step": 24550
     },
     {
       "epoch": 4.6226237530585355,
-      "grad_norm": 0.2390964925289154,
-      "learning_rate": 1.5095049877658574e-06,
-      "loss": 0.4389,
+      "grad_norm": 0.12366284430027008,
+      "learning_rate": 2.264257481648786e-06,
+      "loss": 0.4102,
       "step": 24560
     },
     {
       "epoch": 4.624505928853755,
-      "grad_norm": 9.884252548217773,
-      "learning_rate": 1.5019762845849804e-06,
-      "loss": 0.698,
+      "grad_norm": 3.985370397567749,
+      "learning_rate": 2.2529644268774704e-06,
+      "loss": 0.6824,
       "step": 24570
     },
     {
       "epoch": 4.626388104648974,
-      "grad_norm": 9.831954002380371,
-      "learning_rate": 1.4944475814041033e-06,
-      "loss": 0.5781,
+      "grad_norm": 4.121525764465332,
+      "learning_rate": 2.241671372106155e-06,
+      "loss": 0.4749,
       "step": 24580
     },
     {
       "epoch": 4.628270280444194,
-      "grad_norm": 30.68934440612793,
-      "learning_rate": 1.486918878223226e-06,
-      "loss": 0.5255,
+      "grad_norm": 3.955786943435669,
+      "learning_rate": 2.2303783173348388e-06,
+      "loss": 0.4139,
       "step": 24590
     },
     {
       "epoch": 4.630152456239413,
-      "grad_norm": 10.3611478805542,
-      "learning_rate": 1.479390175042349e-06,
-      "loss": 0.807,
+      "grad_norm": 17.2208251953125,
+      "learning_rate": 2.219085262563523e-06,
+      "loss": 0.7047,
       "step": 24600
     },
     {
       "epoch": 4.632034632034632,
-      "grad_norm": 9.771163940429688,
-      "learning_rate": 1.471861471861472e-06,
-      "loss": 0.4991,
+      "grad_norm": 7.985387325286865,
+      "learning_rate": 2.207792207792208e-06,
+      "loss": 0.4245,
       "step": 24610
     },
     {
       "epoch": 4.633916807829851,
-      "grad_norm": 11.571956634521484,
-      "learning_rate": 1.464332768680595e-06,
-      "loss": 0.4391,
+      "grad_norm": 10.840561866760254,
+      "learning_rate": 2.1964991530208923e-06,
+      "loss": 0.451,
       "step": 24620
     },
     {
       "epoch": 4.635798983625071,
-      "grad_norm": 1.4063133001327515,
-      "learning_rate": 1.4568040654997179e-06,
-      "loss": 0.6051,
+      "grad_norm": 1.2460243701934814,
+      "learning_rate": 2.1852060982495767e-06,
+      "loss": 0.5374,
       "step": 24630
     },
     {
       "epoch": 4.63768115942029,
-      "grad_norm": 4.350733757019043,
-      "learning_rate": 1.4492753623188408e-06,
-      "loss": 0.5334,
+      "grad_norm": 7.391758918762207,
+      "learning_rate": 2.173913043478261e-06,
+      "loss": 0.5306,
       "step": 24640
     },
     {
       "epoch": 4.6395633352155095,
-      "grad_norm": 21.370582580566406,
-      "learning_rate": 1.4417466591379637e-06,
-      "loss": 0.4807,
+      "grad_norm": 17.164552688598633,
+      "learning_rate": 2.1626199887069455e-06,
+      "loss": 0.4736,
       "step": 24650
     },
     {
       "epoch": 4.641445511010728,
-      "grad_norm": 57.6276969909668,
-      "learning_rate": 1.4342179559570864e-06,
-      "loss": 0.752,
+      "grad_norm": 15.955860137939453,
+      "learning_rate": 2.1513269339356294e-06,
+      "loss": 0.7017,
       "step": 24660
     },
     {
       "epoch": 4.643327686805947,
-      "grad_norm": 5.320174217224121,
-      "learning_rate": 1.4266892527762094e-06,
-      "loss": 0.6826,
+      "grad_norm": 6.390231609344482,
+      "learning_rate": 2.140033879164314e-06,
+      "loss": 0.6922,
       "step": 24670
     },
     {
       "epoch": 4.645209862601167,
-      "grad_norm": 37.81419372558594,
-      "learning_rate": 1.4191605495953323e-06,
-      "loss": 0.8429,
+      "grad_norm": 35.61813735961914,
+      "learning_rate": 2.128740824392998e-06,
+      "loss": 0.7417,
       "step": 24680
     },
     {
       "epoch": 4.647092038396386,
-      "grad_norm": 28.828960418701172,
-      "learning_rate": 1.4116318464144552e-06,
-      "loss": 0.7019,
+      "grad_norm": 23.050209045410156,
+      "learning_rate": 2.1174477696216826e-06,
+      "loss": 0.6244,
       "step": 24690
     },
     {
       "epoch": 4.648974214191606,
-      "grad_norm": 8.05349349975586,
-      "learning_rate": 1.4041031432335781e-06,
-      "loss": 0.6875,
+      "grad_norm": 8.97560977935791,
+      "learning_rate": 2.106154714850367e-06,
+      "loss": 0.6843,
       "step": 24700
     },
     {
       "epoch": 4.650856389986825,
-      "grad_norm": 4.589570999145508,
-      "learning_rate": 1.3965744400527008e-06,
-      "loss": 0.5141,
+      "grad_norm": 5.241304874420166,
+      "learning_rate": 2.0948616600790514e-06,
+      "loss": 0.4725,
       "step": 24710
     },
     {
       "epoch": 4.652738565782044,
-      "grad_norm": 21.772384643554688,
-      "learning_rate": 1.3890457368718238e-06,
-      "loss": 0.6424,
+      "grad_norm": 15.338078498840332,
+      "learning_rate": 2.0835686053077357e-06,
+      "loss": 0.6333,
       "step": 24720
     },
     {
       "epoch": 4.654620741577263,
-      "grad_norm": 27.69976806640625,
-      "learning_rate": 1.3815170336909469e-06,
-      "loss": 0.4256,
+      "grad_norm": 31.047271728515625,
+      "learning_rate": 2.0722755505364205e-06,
+      "loss": 0.2574,
       "step": 24730
     },
     {
       "epoch": 4.656502917372483,
-      "grad_norm": 32.07310104370117,
-      "learning_rate": 1.3739883305100698e-06,
-      "loss": 0.4189,
+      "grad_norm": 39.24232864379883,
+      "learning_rate": 2.0609824957651045e-06,
+      "loss": 0.3953,
       "step": 24740
     },
     {
       "epoch": 4.658385093167702,
-      "grad_norm": 16.623987197875977,
-      "learning_rate": 1.3664596273291927e-06,
-      "loss": 0.5246,
+      "grad_norm": 6.512187957763672,
+      "learning_rate": 2.049689440993789e-06,
+      "loss": 0.398,
       "step": 24750
     },
     {
       "epoch": 4.660267268962921,
-      "grad_norm": 15.414529800415039,
-      "learning_rate": 1.3589309241483157e-06,
-      "loss": 0.7166,
+      "grad_norm": 10.04133415222168,
+      "learning_rate": 2.0383963862224733e-06,
+      "loss": 0.6215,
       "step": 24760
     },
     {
       "epoch": 4.662149444758141,
-      "grad_norm": 14.814608573913574,
-      "learning_rate": 1.3514022209674386e-06,
-      "loss": 0.8627,
+      "grad_norm": 20.212570190429688,
+      "learning_rate": 2.0271033314511576e-06,
+      "loss": 0.8138,
       "step": 24770
     },
     {
       "epoch": 4.664031620553359,
-      "grad_norm": 18.566692352294922,
-      "learning_rate": 1.3438735177865615e-06,
-      "loss": 0.7965,
+      "grad_norm": 8.476882934570312,
+      "learning_rate": 2.015810276679842e-06,
+      "loss": 0.7214,
       "step": 24780
     },
     {
       "epoch": 4.665913796348579,
-      "grad_norm": 7.707961559295654,
-      "learning_rate": 1.3363448146056842e-06,
-      "loss": 0.5256,
+      "grad_norm": 11.082260131835938,
+      "learning_rate": 2.0045172219085264e-06,
+      "loss": 0.4978,
       "step": 24790
     },
     {
       "epoch": 4.667795972143798,
-      "grad_norm": 0.7956398725509644,
-      "learning_rate": 1.3288161114248071e-06,
-      "loss": 0.9007,
+      "grad_norm": 0.5434314012527466,
+      "learning_rate": 1.993224167137211e-06,
+      "loss": 0.72,
       "step": 24800
     },
     {
       "epoch": 4.669678147939018,
-      "grad_norm": 25.115041732788086,
-      "learning_rate": 1.32128740824393e-06,
-      "loss": 0.6778,
+      "grad_norm": 28.41150665283203,
+      "learning_rate": 1.9819311123658948e-06,
+      "loss": 0.6336,
       "step": 24810
     },
     {
       "epoch": 4.671560323734237,
-      "grad_norm": 1.7890321016311646,
-      "learning_rate": 1.313758705063053e-06,
-      "loss": 0.9155,
+      "grad_norm": 1.0246213674545288,
+      "learning_rate": 1.970638057594579e-06,
+      "loss": 0.888,
       "step": 24820
     },
     {
       "epoch": 4.673442499529456,
-      "grad_norm": 16.954694747924805,
-      "learning_rate": 1.3062300018821759e-06,
-      "loss": 0.6522,
+      "grad_norm": 15.647318840026855,
+      "learning_rate": 1.9593450028232635e-06,
+      "loss": 0.6784,
       "step": 24830
     },
     {
       "epoch": 4.675324675324675,
-      "grad_norm": 18.548112869262695,
-      "learning_rate": 1.2987012987012986e-06,
-      "loss": 0.3274,
+      "grad_norm": 16.091299057006836,
+      "learning_rate": 1.948051948051948e-06,
+      "loss": 0.2994,
       "step": 24840
     },
     {
       "epoch": 4.6772068511198945,
-      "grad_norm": 19.337177276611328,
-      "learning_rate": 1.291172595520422e-06,
-      "loss": 0.5718,
+      "grad_norm": 20.402158737182617,
+      "learning_rate": 1.9367588932806327e-06,
+      "loss": 0.4222,
       "step": 24850
     },
     {
       "epoch": 4.679089026915114,
-      "grad_norm": 29.29849624633789,
-      "learning_rate": 1.2836438923395447e-06,
-      "loss": 0.6632,
+      "grad_norm": 29.887250900268555,
+      "learning_rate": 1.925465838509317e-06,
+      "loss": 0.5632,
       "step": 24860
     },
     {
       "epoch": 4.680971202710333,
-      "grad_norm": 0.6499333381652832,
-      "learning_rate": 1.2761151891586676e-06,
-      "loss": 0.3925,
+      "grad_norm": 0.39280015230178833,
+      "learning_rate": 1.9141727837380015e-06,
+      "loss": 0.3391,
       "step": 24870
     },
     {
       "epoch": 4.682853378505552,
-      "grad_norm": 10.113441467285156,
-      "learning_rate": 1.2685864859777905e-06,
-      "loss": 0.4654,
+      "grad_norm": 1.1423434019088745,
+      "learning_rate": 1.9028797289666856e-06,
+      "loss": 0.4052,
       "step": 24880
     },
     {
       "epoch": 4.684735554300771,
-      "grad_norm": 0.6062495708465576,
-      "learning_rate": 1.2610577827969134e-06,
-      "loss": 0.6459,
+      "grad_norm": 0.18918712437152863,
+      "learning_rate": 1.89158667419537e-06,
+      "loss": 0.6111,
       "step": 24890
     },
     {
       "epoch": 4.686617730095991,
-      "grad_norm": 6.477812767028809,
-      "learning_rate": 1.2535290796160363e-06,
-      "loss": 0.4256,
+      "grad_norm": 6.821197986602783,
+      "learning_rate": 1.8802936194240542e-06,
+      "loss": 0.3858,
       "step": 24900
     },
     {
       "epoch": 4.68849990589121,
-      "grad_norm": 11.277091979980469,
-      "learning_rate": 1.246000376435159e-06,
-      "loss": 0.5037,
+      "grad_norm": 18.688823699951172,
+      "learning_rate": 1.8690005646527386e-06,
+      "loss": 0.5372,
       "step": 24910
     },
     {
       "epoch": 4.69038208168643,
-      "grad_norm": 24.859067916870117,
-      "learning_rate": 1.238471673254282e-06,
-      "loss": 0.6871,
+      "grad_norm": 22.61110496520996,
+      "learning_rate": 1.857707509881423e-06,
+      "loss": 0.5846,
       "step": 24920
     },
     {
       "epoch": 4.692264257481649,
-      "grad_norm": 16.682817459106445,
-      "learning_rate": 1.2309429700734049e-06,
-      "loss": 0.4382,
+      "grad_norm": 14.270989418029785,
+      "learning_rate": 1.8464144551101073e-06,
+      "loss": 0.3677,
       "step": 24930
     },
     {
       "epoch": 4.6941464332768685,
-      "grad_norm": 15.542104721069336,
-      "learning_rate": 1.223414266892528e-06,
-      "loss": 0.4241,
+      "grad_norm": 20.81678581237793,
+      "learning_rate": 1.8351214003387917e-06,
+      "loss": 0.3929,
       "step": 24940
     },
     {
       "epoch": 4.696028609072087,
-      "grad_norm": 0.6870865225791931,
-      "learning_rate": 1.2158855637116507e-06,
-      "loss": 0.5352,
+      "grad_norm": 0.7753129601478577,
+      "learning_rate": 1.823828345567476e-06,
+      "loss": 0.512,
       "step": 24950
     },
     {
       "epoch": 4.6979107848673065,
-      "grad_norm": 7.823055744171143,
-      "learning_rate": 1.2083568605307737e-06,
-      "loss": 0.9063,
+      "grad_norm": 7.8035664558410645,
+      "learning_rate": 1.8125352907961605e-06,
+      "loss": 0.8553,
       "step": 24960
     },
     {
       "epoch": 4.699792960662526,
-      "grad_norm": 4.5665483474731445,
-      "learning_rate": 1.2008281573498966e-06,
-      "loss": 0.7457,
+      "grad_norm": 9.686589241027832,
+      "learning_rate": 1.8012422360248449e-06,
+      "loss": 0.7138,
       "step": 24970
     },
     {
       "epoch": 4.701675136457745,
-      "grad_norm": 1.2452117204666138,
-      "learning_rate": 1.1932994541690195e-06,
-      "loss": 0.6592,
+      "grad_norm": 0.7475610375404358,
+      "learning_rate": 1.789949181253529e-06,
+      "loss": 0.6173,
       "step": 24980
     },
     {
       "epoch": 4.703557312252965,
-      "grad_norm": 18.012990951538086,
-      "learning_rate": 1.1857707509881424e-06,
-      "loss": 0.401,
+      "grad_norm": 11.084076881408691,
+      "learning_rate": 1.7786561264822134e-06,
+      "loss": 0.353,
       "step": 24990
     },
     {
       "epoch": 4.705439488048183,
-      "grad_norm": 7.86597204208374,
-      "learning_rate": 1.1782420478072651e-06,
-      "loss": 0.7094,
+      "grad_norm": 33.30714416503906,
+      "learning_rate": 1.7673630717108978e-06,
+      "loss": 0.578,
       "step": 25000
     },
     {
       "epoch": 4.707321663843403,
-      "grad_norm": 32.99113845825195,
-      "learning_rate": 1.1707133446263883e-06,
-      "loss": 0.4918,
+      "grad_norm": 35.408416748046875,
+      "learning_rate": 1.7560700169395822e-06,
+      "loss": 0.4082,
       "step": 25010
     },
     {
       "epoch": 4.709203839638622,
-      "grad_norm": 33.427459716796875,
-      "learning_rate": 1.1631846414455112e-06,
-      "loss": 0.5377,
+      "grad_norm": 37.653106689453125,
+      "learning_rate": 1.7447769621682666e-06,
+      "loss": 0.5041,
       "step": 25020
     },
     {
       "epoch": 4.711086015433842,
-      "grad_norm": 0.6558060050010681,
-      "learning_rate": 1.155655938264634e-06,
-      "loss": 0.6283,
+      "grad_norm": 0.38172101974487305,
+      "learning_rate": 1.733483907396951e-06,
+      "loss": 0.6022,
       "step": 25030
     },
     {
       "epoch": 4.712968191229061,
-      "grad_norm": 0.2545570433139801,
-      "learning_rate": 1.1481272350837568e-06,
-      "loss": 0.4473,
+      "grad_norm": 0.1132197305560112,
+      "learning_rate": 1.7221908526256353e-06,
+      "loss": 0.4891,
       "step": 25040
     },
     {
       "epoch": 4.71485036702428,
-      "grad_norm": 26.013124465942383,
-      "learning_rate": 1.1405985319028797e-06,
-      "loss": 0.6082,
+      "grad_norm": 24.087400436401367,
+      "learning_rate": 1.7108977978543195e-06,
+      "loss": 0.4906,
       "step": 25050
     },
     {
       "epoch": 4.716732542819499,
-      "grad_norm": 6.071534156799316,
-      "learning_rate": 1.1330698287220027e-06,
-      "loss": 0.5555,
+      "grad_norm": 9.231995582580566,
+      "learning_rate": 1.6996047430830039e-06,
+      "loss": 0.6238,
       "step": 25060
     },
     {
       "epoch": 4.7186147186147185,
-      "grad_norm": 24.877159118652344,
-      "learning_rate": 1.1255411255411256e-06,
-      "loss": 1.0349,
+      "grad_norm": 22.86023712158203,
+      "learning_rate": 1.6883116883116885e-06,
+      "loss": 0.9478,
       "step": 25070
     },
     {
       "epoch": 4.720496894409938,
-      "grad_norm": 7.250149250030518,
-      "learning_rate": 1.1180124223602485e-06,
-      "loss": 0.5758,
+      "grad_norm": 7.785012245178223,
+      "learning_rate": 1.6770186335403729e-06,
+      "loss": 0.5008,
       "step": 25080
     },
     {
       "epoch": 4.722379070205157,
-      "grad_norm": 2.08498215675354,
-      "learning_rate": 1.1104837191793714e-06,
-      "loss": 0.8396,
+      "grad_norm": 2.3250386714935303,
+      "learning_rate": 1.665725578769057e-06,
+      "loss": 0.7925,
       "step": 25090
     },
     {
       "epoch": 4.724261246000377,
-      "grad_norm": 12.615907669067383,
-      "learning_rate": 1.1029550159984943e-06,
-      "loss": 0.4626,
+      "grad_norm": 12.125810623168945,
+      "learning_rate": 1.6544325239977414e-06,
+      "loss": 0.3744,
       "step": 25100
     },
     {
       "epoch": 4.726143421795595,
-      "grad_norm": 24.194643020629883,
-      "learning_rate": 1.0954263128176173e-06,
-      "loss": 0.4744,
+      "grad_norm": 14.144658088684082,
+      "learning_rate": 1.6431394692264258e-06,
+      "loss": 0.4073,
       "step": 25110
     },
     {
       "epoch": 4.728025597590815,
-      "grad_norm": 5.291766166687012,
-      "learning_rate": 1.0878976096367402e-06,
-      "loss": 0.3882,
+      "grad_norm": 4.601728916168213,
+      "learning_rate": 1.63184641445511e-06,
+      "loss": 0.3743,
       "step": 25120
     },
     {
       "epoch": 4.729907773386034,
-      "grad_norm": 20.6169376373291,
-      "learning_rate": 1.0803689064558631e-06,
-      "loss": 0.6649,
+      "grad_norm": 9.717355728149414,
+      "learning_rate": 1.6205533596837946e-06,
+      "loss": 0.5983,
       "step": 25130
     },
     {
       "epoch": 4.731789949181254,
-      "grad_norm": 18.83254623413086,
-      "learning_rate": 1.072840203274986e-06,
-      "loss": 0.4599,
+      "grad_norm": 13.010869026184082,
+      "learning_rate": 1.609260304912479e-06,
+      "loss": 0.4777,
       "step": 25140
     },
     {
       "epoch": 4.733672124976473,
-      "grad_norm": 12.41308307647705,
-      "learning_rate": 1.065311500094109e-06,
-      "loss": 0.8391,
+      "grad_norm": 10.841914176940918,
+      "learning_rate": 1.5979672501411633e-06,
+      "loss": 0.7914,
       "step": 25150
     },
     {
       "epoch": 4.7355543007716925,
-      "grad_norm": 13.976269721984863,
-      "learning_rate": 1.0577827969132319e-06,
-      "loss": 0.5846,
+      "grad_norm": 11.253789901733398,
+      "learning_rate": 1.5866741953698475e-06,
+      "loss": 0.5007,
       "step": 25160
     },
     {
       "epoch": 4.737436476566911,
-      "grad_norm": 25.47661590576172,
-      "learning_rate": 1.0502540937323546e-06,
-      "loss": 0.604,
+      "grad_norm": 20.718843460083008,
+      "learning_rate": 1.5753811405985319e-06,
+      "loss": 0.4821,
       "step": 25170
     },
     {
       "epoch": 4.7393186523621305,
-      "grad_norm": 21.08745765686035,
-      "learning_rate": 1.0427253905514775e-06,
-      "loss": 0.4192,
+      "grad_norm": 15.470338821411133,
+      "learning_rate": 1.5640880858272163e-06,
+      "loss": 0.3737,
       "step": 25180
     },
     {
       "epoch": 4.74120082815735,
-      "grad_norm": 0.9242314696311951,
-      "learning_rate": 1.0351966873706006e-06,
-      "loss": 0.3425,
+      "grad_norm": 0.15901845693588257,
+      "learning_rate": 1.5527950310559008e-06,
+      "loss": 0.1535,
       "step": 25190
     },
     {
       "epoch": 4.743083003952569,
-      "grad_norm": 11.735674858093262,
-      "learning_rate": 1.0276679841897233e-06,
-      "loss": 0.5547,
+      "grad_norm": 17.537002563476562,
+      "learning_rate": 1.541501976284585e-06,
+      "loss": 0.3773,
       "step": 25200
     },
     {
       "epoch": 4.744965179747789,
-      "grad_norm": 7.690491199493408,
-      "learning_rate": 1.0201392810088463e-06,
-      "loss": 0.3842,
+      "grad_norm": 6.956587314605713,
+      "learning_rate": 1.5302089215132694e-06,
+      "loss": 0.3261,
       "step": 25210
     },
     {
       "epoch": 4.746847355543007,
-      "grad_norm": 12.007920265197754,
-      "learning_rate": 1.0126105778279692e-06,
-      "loss": 0.679,
+      "grad_norm": 5.067175388336182,
+      "learning_rate": 1.5189158667419538e-06,
+      "loss": 0.6587,
       "step": 25220
     },
     {
       "epoch": 4.748729531338227,
-      "grad_norm": 20.26568603515625,
-      "learning_rate": 1.0050818746470921e-06,
-      "loss": 0.751,
+      "grad_norm": 25.866880416870117,
+      "learning_rate": 1.507622811970638e-06,
+      "loss": 0.6093,
       "step": 25230
     },
     {
       "epoch": 4.750611707133446,
-      "grad_norm": 0.7630726099014282,
-      "learning_rate": 9.97553171466215e-07,
-      "loss": 0.3219,
+      "grad_norm": 2.593472957611084,
+      "learning_rate": 1.4963297571993223e-06,
+      "loss": 0.2919,
       "step": 25240
     },
     {
       "epoch": 4.752493882928666,
-      "grad_norm": 5.351938247680664,
-      "learning_rate": 9.90024468285338e-07,
-      "loss": 0.6458,
+      "grad_norm": 5.722588062286377,
+      "learning_rate": 1.485036702428007e-06,
+      "loss": 0.464,
       "step": 25250
     },
     {
       "epoch": 4.754376058723885,
-      "grad_norm": 11.440779685974121,
-      "learning_rate": 9.824957651044609e-07,
-      "loss": 0.3074,
+      "grad_norm": 12.117194175720215,
+      "learning_rate": 1.4737436476566913e-06,
+      "loss": 0.2929,
       "step": 25260
     },
     {
       "epoch": 4.7562582345191045,
-      "grad_norm": 13.200923919677734,
-      "learning_rate": 9.749670619235838e-07,
-      "loss": 0.7268,
+      "grad_norm": 18.254680633544922,
+      "learning_rate": 1.4624505928853755e-06,
+      "loss": 0.757,
       "step": 25270
     },
     {
       "epoch": 4.758140410314323,
-      "grad_norm": 0.8628940582275391,
-      "learning_rate": 9.674383587427067e-07,
-      "loss": 0.5937,
+      "grad_norm": 0.17077714204788208,
+      "learning_rate": 1.4511575381140599e-06,
+      "loss": 0.4409,
       "step": 25280
     },
     {
       "epoch": 4.760022586109542,
-      "grad_norm": 12.860737800598145,
-      "learning_rate": 9.599096555618294e-07,
-      "loss": 0.6364,
+      "grad_norm": 11.023554801940918,
+      "learning_rate": 1.4398644833427442e-06,
+      "loss": 0.6109,
       "step": 25290
     },
     {
       "epoch": 4.761904761904762,
-      "grad_norm": 18.585010528564453,
-      "learning_rate": 9.523809523809525e-07,
-      "loss": 0.604,
+      "grad_norm": 11.318467140197754,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.41,
       "step": 25300
     },
     {
       "epoch": 4.763786937699981,
-      "grad_norm": 10.778792381286621,
-      "learning_rate": 9.448522492000754e-07,
-      "loss": 0.408,
+      "grad_norm": 0.2806694507598877,
+      "learning_rate": 1.417278373800113e-06,
+      "loss": 0.3642,
       "step": 25310
     },
     {
       "epoch": 4.765669113495201,
-      "grad_norm": 4.942008018493652,
-      "learning_rate": 9.373235460191983e-07,
-      "loss": 0.6934,
+      "grad_norm": 3.844435930252075,
+      "learning_rate": 1.4059853190287974e-06,
+      "loss": 0.5849,
       "step": 25320
     },
     {
       "epoch": 4.76755128929042,
-      "grad_norm": 9.399242401123047,
-      "learning_rate": 9.297948428383212e-07,
-      "loss": 0.4827,
+      "grad_norm": 6.41024923324585,
+      "learning_rate": 1.3946922642574818e-06,
+      "loss": 0.4466,
       "step": 25330
     },
     {
       "epoch": 4.769433465085639,
-      "grad_norm": 7.679773330688477,
-      "learning_rate": 9.22266139657444e-07,
-      "loss": 0.448,
+      "grad_norm": 12.630584716796875,
+      "learning_rate": 1.383399209486166e-06,
+      "loss": 0.4274,
       "step": 25340
     },
     {
       "epoch": 4.771315640880858,
-      "grad_norm": 11.352853775024414,
-      "learning_rate": 9.14737436476567e-07,
-      "loss": 0.5005,
+      "grad_norm": 14.851526260375977,
+      "learning_rate": 1.3721061547148503e-06,
+      "loss": 0.5182,
       "step": 25350
     },
     {
       "epoch": 4.773197816676078,
-      "grad_norm": 15.323731422424316,
-      "learning_rate": 9.072087332956899e-07,
-      "loss": 0.8852,
+      "grad_norm": 10.023591995239258,
+      "learning_rate": 1.3608130999435347e-06,
+      "loss": 0.7274,
       "step": 25360
     },
     {
       "epoch": 4.775079992471297,
-      "grad_norm": 34.29592514038086,
-      "learning_rate": 8.996800301148129e-07,
-      "loss": 0.4909,
+      "grad_norm": 32.864810943603516,
+      "learning_rate": 1.3495200451722193e-06,
+      "loss": 0.464,
       "step": 25370
     },
     {
       "epoch": 4.776962168266516,
-      "grad_norm": 15.671870231628418,
-      "learning_rate": 8.921513269339357e-07,
-      "loss": 0.5851,
+      "grad_norm": 17.206785202026367,
+      "learning_rate": 1.3382269904009035e-06,
+      "loss": 0.4297,
       "step": 25380
     },
     {
       "epoch": 4.778844344061735,
-      "grad_norm": 15.997258186340332,
-      "learning_rate": 8.846226237530586e-07,
-      "loss": 0.5856,
+      "grad_norm": 11.58726978302002,
+      "learning_rate": 1.3269339356295879e-06,
+      "loss": 0.4302,
       "step": 25390
     },
     {
       "epoch": 4.780726519856954,
-      "grad_norm": 7.108194828033447,
-      "learning_rate": 8.770939205721816e-07,
-      "loss": 0.6275,
+      "grad_norm": 5.545811176300049,
+      "learning_rate": 1.3156408808582722e-06,
+      "loss": 0.4928,
       "step": 25400
     },
     {
       "epoch": 4.782608695652174,
-      "grad_norm": 17.355302810668945,
-      "learning_rate": 8.695652173913044e-07,
-      "loss": 0.6376,
+      "grad_norm": 14.361126899719238,
+      "learning_rate": 1.3043478260869566e-06,
+      "loss": 0.6301,
       "step": 25410
     },
     {
       "epoch": 4.784490871447393,
-      "grad_norm": 2.995657205581665,
-      "learning_rate": 8.620365142104273e-07,
-      "loss": 0.8359,
+      "grad_norm": 16.078994750976562,
+      "learning_rate": 1.2930547713156408e-06,
+      "loss": 0.8214,
       "step": 25420
     },
     {
       "epoch": 4.786373047242613,
-      "grad_norm": 11.23421859741211,
-      "learning_rate": 8.545078110295501e-07,
-      "loss": 0.8888,
+      "grad_norm": 18.931100845336914,
+      "learning_rate": 1.2817617165443252e-06,
+      "loss": 0.8551,
       "step": 25430
     },
     {
       "epoch": 4.788255223037831,
-      "grad_norm": 11.181546211242676,
-      "learning_rate": 8.469791078486731e-07,
-      "loss": 0.8431,
+      "grad_norm": 12.793909072875977,
+      "learning_rate": 1.2704686617730098e-06,
+      "loss": 0.8418,
       "step": 25440
     },
     {
       "epoch": 4.790137398833051,
-      "grad_norm": 17.23093032836914,
-      "learning_rate": 8.394504046677961e-07,
-      "loss": 0.4511,
+      "grad_norm": 9.514110565185547,
+      "learning_rate": 1.259175607001694e-06,
+      "loss": 0.4481,
       "step": 25450
     },
     {
       "epoch": 4.79201957462827,
-      "grad_norm": 17.615921020507812,
-      "learning_rate": 8.31921701486919e-07,
-      "loss": 0.7012,
+      "grad_norm": 2.824892997741699,
+      "learning_rate": 1.2478825522303783e-06,
+      "loss": 0.7443,
       "step": 25460
     },
     {
       "epoch": 4.7939017504234895,
-      "grad_norm": 3.298682451248169,
-      "learning_rate": 8.243929983060418e-07,
-      "loss": 0.5764,
+      "grad_norm": 0.1960848569869995,
+      "learning_rate": 1.2365894974590627e-06,
+      "loss": 0.5792,
       "step": 25470
     },
     {
       "epoch": 4.795783926218709,
-      "grad_norm": 0.7209872603416443,
-      "learning_rate": 8.168642951251647e-07,
-      "loss": 0.5459,
+      "grad_norm": 0.339618057012558,
+      "learning_rate": 1.225296442687747e-06,
+      "loss": 0.5427,
       "step": 25480
     },
     {
       "epoch": 4.797666102013928,
-      "grad_norm": 10.0849609375,
-      "learning_rate": 8.093355919442876e-07,
-      "loss": 0.4159,
+      "grad_norm": 6.723775863647461,
+      "learning_rate": 1.2140033879164313e-06,
+      "loss": 0.3267,
       "step": 25490
     },
     {
       "epoch": 4.799548277809148,
-      "grad_norm": 2.3432960510253906,
-      "learning_rate": 8.018068887634107e-07,
-      "loss": 0.5565,
+      "grad_norm": 1.1799036264419556,
+      "learning_rate": 1.2027103331451158e-06,
+      "loss": 0.3805,
       "step": 25500
     },
     {
       "epoch": 4.801430453604366,
-      "grad_norm": 1.5209139585494995,
-      "learning_rate": 7.942781855825335e-07,
-      "loss": 0.3799,
+      "grad_norm": 3.111079216003418,
+      "learning_rate": 1.1914172783738002e-06,
+      "loss": 0.3117,
       "step": 25510
     },
     {
       "epoch": 4.803312629399586,
-      "grad_norm": 8.557663917541504,
-      "learning_rate": 7.867494824016564e-07,
-      "loss": 0.6082,
+      "grad_norm": 0.1729825735092163,
+      "learning_rate": 1.1801242236024846e-06,
+      "loss": 0.4701,
       "step": 25520
     },
     {
       "epoch": 4.805194805194805,
-      "grad_norm": 42.79642105102539,
-      "learning_rate": 7.792207792207792e-07,
-      "loss": 0.5106,
+      "grad_norm": 8.52800464630127,
+      "learning_rate": 1.1688311688311688e-06,
+      "loss": 0.4286,
       "step": 25530
     },
     {
       "epoch": 4.807076980990025,
-      "grad_norm": 20.128494262695312,
-      "learning_rate": 7.716920760399021e-07,
-      "loss": 0.3495,
+      "grad_norm": 10.847475051879883,
+      "learning_rate": 1.1575381140598532e-06,
+      "loss": 0.3005,
       "step": 25540
     },
     {
       "epoch": 4.808959156785244,
-      "grad_norm": 8.818672180175781,
-      "learning_rate": 7.641633728590251e-07,
-      "loss": 0.4237,
+      "grad_norm": 11.9356689453125,
+      "learning_rate": 1.1462450592885375e-06,
+      "loss": 0.4045,
       "step": 25550
     },
     {
       "epoch": 4.810841332580463,
-      "grad_norm": 10.33991813659668,
-      "learning_rate": 7.566346696781481e-07,
-      "loss": 0.6665,
+      "grad_norm": 7.127353668212891,
+      "learning_rate": 1.134952004517222e-06,
+      "loss": 0.5741,
       "step": 25560
     },
     {
       "epoch": 4.812723508375682,
-      "grad_norm": 15.527596473693848,
-      "learning_rate": 7.491059664972709e-07,
-      "loss": 0.5248,
+      "grad_norm": 16.197805404663086,
+      "learning_rate": 1.1236589497459063e-06,
+      "loss": 0.5069,
       "step": 25570
     },
     {
       "epoch": 4.8146056841709015,
-      "grad_norm": 1.0982279777526855,
-      "learning_rate": 7.415772633163938e-07,
-      "loss": 0.4951,
+      "grad_norm": 0.5227009057998657,
+      "learning_rate": 1.1123658949745907e-06,
+      "loss": 0.4081,
       "step": 25580
     },
     {
       "epoch": 4.816487859966121,
-      "grad_norm": 10.367071151733398,
-      "learning_rate": 7.340485601355168e-07,
-      "loss": 0.5674,
+      "grad_norm": 16.148591995239258,
+      "learning_rate": 1.101072840203275e-06,
+      "loss": 0.5699,
       "step": 25590
     },
     {
       "epoch": 4.81837003576134,
-      "grad_norm": 5.20026969909668,
-      "learning_rate": 7.265198569546396e-07,
-      "loss": 0.528,
+      "grad_norm": 1.3249871730804443,
+      "learning_rate": 1.0897797854319592e-06,
+      "loss": 0.4742,
       "step": 25600
     },
     {
       "epoch": 4.820252211556559,
-      "grad_norm": 15.308959007263184,
-      "learning_rate": 7.189911537737625e-07,
-      "loss": 0.716,
+      "grad_norm": 11.252976417541504,
+      "learning_rate": 1.0784867306606436e-06,
+      "loss": 0.5348,
       "step": 25610
     },
     {
       "epoch": 4.822134387351778,
-      "grad_norm": 35.8542366027832,
-      "learning_rate": 7.114624505928855e-07,
-      "loss": 0.6604,
+      "grad_norm": 9.052772521972656,
+      "learning_rate": 1.0671936758893282e-06,
+      "loss": 0.5361,
       "step": 25620
     },
     {
       "epoch": 4.824016563146998,
-      "grad_norm": 45.377445220947266,
-      "learning_rate": 7.039337474120083e-07,
-      "loss": 0.5864,
+      "grad_norm": 34.760986328125,
+      "learning_rate": 1.0559006211180126e-06,
+      "loss": 0.4909,
       "step": 25630
     },
     {
       "epoch": 4.825898738942217,
-      "grad_norm": 37.29194259643555,
-      "learning_rate": 6.964050442311313e-07,
-      "loss": 0.4791,
+      "grad_norm": 18.605308532714844,
+      "learning_rate": 1.0446075663466968e-06,
+      "loss": 0.3664,
       "step": 25640
     },
     {
       "epoch": 4.827780914737437,
-      "grad_norm": 0.8685435652732849,
-      "learning_rate": 6.888763410502542e-07,
-      "loss": 0.3548,
+      "grad_norm": 1.8343195915222168,
+      "learning_rate": 1.0333145115753812e-06,
+      "loss": 0.2793,
       "step": 25650
     },
     {
       "epoch": 4.829663090532656,
-      "grad_norm": 10.831360816955566,
-      "learning_rate": 6.81347637869377e-07,
-      "loss": 0.4429,
+      "grad_norm": 6.831780433654785,
+      "learning_rate": 1.0220214568040655e-06,
+      "loss": 0.4601,
       "step": 25660
     },
     {
       "epoch": 4.8315452663278755,
-      "grad_norm": 15.885283470153809,
-      "learning_rate": 6.738189346884999e-07,
-      "loss": 0.4418,
+      "grad_norm": 9.190847396850586,
+      "learning_rate": 1.0107284020327497e-06,
+      "loss": 0.3852,
       "step": 25670
     },
     {
       "epoch": 4.833427442123094,
-      "grad_norm": 12.996732711791992,
-      "learning_rate": 6.662902315076229e-07,
-      "loss": 0.3234,
+      "grad_norm": 10.883723258972168,
+      "learning_rate": 9.994353472614343e-07,
+      "loss": 0.3296,
       "step": 25680
     },
     {
       "epoch": 4.8353096179183135,
-      "grad_norm": 17.03729248046875,
-      "learning_rate": 6.587615283267459e-07,
-      "loss": 0.4515,
+      "grad_norm": 11.370898246765137,
+      "learning_rate": 9.881422924901187e-07,
+      "loss": 0.3896,
       "step": 25690
     },
     {
       "epoch": 4.837191793713533,
-      "grad_norm": 4.803049087524414,
-      "learning_rate": 6.512328251458687e-07,
-      "loss": 0.5221,
+      "grad_norm": 0.8723617792129517,
+      "learning_rate": 9.76849237718803e-07,
+      "loss": 0.4617,
       "step": 25700
     },
     {
       "epoch": 4.839073969508752,
-      "grad_norm": 8.547290802001953,
-      "learning_rate": 6.437041219649916e-07,
-      "loss": 0.3565,
+      "grad_norm": 2.9830756187438965,
+      "learning_rate": 9.655561829474872e-07,
+      "loss": 0.3264,
       "step": 25710
     },
     {
       "epoch": 4.840956145303972,
-      "grad_norm": 7.503337860107422,
-      "learning_rate": 6.361754187841144e-07,
-      "loss": 0.6951,
+      "grad_norm": 9.031828880310059,
+      "learning_rate": 9.542631281761716e-07,
+      "loss": 0.5838,
       "step": 25720
     },
     {
       "epoch": 4.84283832109919,
-      "grad_norm": 541.9999389648438,
-      "learning_rate": 6.286467156032373e-07,
-      "loss": 0.2751,
+      "grad_norm": 37.80723571777344,
+      "learning_rate": 9.42970073404856e-07,
+      "loss": 0.2824,
       "step": 25730
     },
     {
       "epoch": 4.84472049689441,
-      "grad_norm": 12.69980239868164,
-      "learning_rate": 6.211180124223603e-07,
-      "loss": 0.4805,
+      "grad_norm": 9.098519325256348,
+      "learning_rate": 9.316770186335404e-07,
+      "loss": 0.4279,
       "step": 25740
     },
     {
       "epoch": 4.846602672689629,
-      "grad_norm": 8.881111145019531,
-      "learning_rate": 6.135893092414832e-07,
-      "loss": 0.502,
+      "grad_norm": 12.211234092712402,
+      "learning_rate": 9.203839638622247e-07,
+      "loss": 0.5721,
       "step": 25750
     },
     {
       "epoch": 4.848484848484849,
-      "grad_norm": 8.829740524291992,
-      "learning_rate": 6.060606060606061e-07,
-      "loss": 0.5651,
+      "grad_norm": 1.9672722816467285,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.432,
       "step": 25760
     },
     {
       "epoch": 4.850367024280068,
-      "grad_norm": 21.692197799682617,
-      "learning_rate": 5.98531902879729e-07,
-      "loss": 0.6097,
+      "grad_norm": 18.3715763092041,
+      "learning_rate": 8.977978543195934e-07,
+      "loss": 0.5275,
       "step": 25770
     },
     {
       "epoch": 4.852249200075287,
-      "grad_norm": 12.04153060913086,
-      "learning_rate": 5.910031996988519e-07,
-      "loss": 0.4574,
+      "grad_norm": 6.322066307067871,
+      "learning_rate": 8.865047995482778e-07,
+      "loss": 0.4312,
       "step": 25780
     },
     {
       "epoch": 4.854131375870506,
-      "grad_norm": 10.96838092803955,
-      "learning_rate": 5.834744965179749e-07,
-      "loss": 0.5796,
+      "grad_norm": 13.303924560546875,
+      "learning_rate": 8.752117447769622e-07,
+      "loss": 0.5374,
       "step": 25790
     },
     {
       "epoch": 4.8560135516657255,
-      "grad_norm": 17.027944564819336,
-      "learning_rate": 5.759457933370977e-07,
-      "loss": 0.6143,
+      "grad_norm": 18.152463912963867,
+      "learning_rate": 8.639186900056466e-07,
+      "loss": 0.4755,
       "step": 25800
     },
     {
       "epoch": 4.857895727460945,
-      "grad_norm": 2.2098312377929688,
-      "learning_rate": 5.684170901562206e-07,
-      "loss": 0.5982,
+      "grad_norm": 1.1153311729431152,
+      "learning_rate": 8.526256352343308e-07,
+      "loss": 0.5785,
       "step": 25810
     },
     {
       "epoch": 4.859777903256164,
-      "grad_norm": 10.510016441345215,
-      "learning_rate": 5.608883869753435e-07,
-      "loss": 0.5048,
+      "grad_norm": 17.98719024658203,
+      "learning_rate": 8.413325804630153e-07,
+      "loss": 0.4314,
       "step": 25820
     },
     {
       "epoch": 4.861660079051384,
-      "grad_norm": 3.657031536102295,
-      "learning_rate": 5.533596837944664e-07,
-      "loss": 0.5796,
+      "grad_norm": 5.713428020477295,
+      "learning_rate": 8.300395256916996e-07,
+      "loss": 0.5436,
       "step": 25830
     },
     {
       "epoch": 4.863542254846602,
-      "grad_norm": 5.817481517791748,
-      "learning_rate": 5.458309806135894e-07,
-      "loss": 0.3377,
+      "grad_norm": 17.330768585205078,
+      "learning_rate": 8.18746470920384e-07,
+      "loss": 0.3292,
       "step": 25840
     },
     {
       "epoch": 4.865424430641822,
-      "grad_norm": 0.44938135147094727,
-      "learning_rate": 5.383022774327123e-07,
-      "loss": 0.6224,
+      "grad_norm": 0.1503809541463852,
+      "learning_rate": 8.074534161490684e-07,
+      "loss": 0.575,
       "step": 25850
     },
     {
       "epoch": 4.867306606437041,
-      "grad_norm": 67.99935913085938,
-      "learning_rate": 5.307735742518352e-07,
-      "loss": 0.8013,
+      "grad_norm": 29.405603408813477,
+      "learning_rate": 7.961603613777526e-07,
+      "loss": 0.7215,
       "step": 25860
     },
     {
       "epoch": 4.869188782232261,
-      "grad_norm": 34.21944046020508,
-      "learning_rate": 5.23244871070958e-07,
-      "loss": 0.3706,
+      "grad_norm": 5.5133209228515625,
+      "learning_rate": 7.84867306606437e-07,
+      "loss": 0.2398,
       "step": 25870
     },
     {
       "epoch": 4.87107095802748,
-      "grad_norm": 12.056262016296387,
-      "learning_rate": 5.15716167890081e-07,
-      "loss": 0.7562,
+      "grad_norm": 10.060201644897461,
+      "learning_rate": 7.735742518351214e-07,
+      "loss": 0.7002,
       "step": 25880
     },
     {
       "epoch": 4.8729531338226995,
-      "grad_norm": 8.577810287475586,
-      "learning_rate": 5.081874647092039e-07,
-      "loss": 0.5821,
+      "grad_norm": 31.25115966796875,
+      "learning_rate": 7.622811970638058e-07,
+      "loss": 0.5119,
       "step": 25890
     },
     {
       "epoch": 4.874835309617918,
-      "grad_norm": 23.58852767944336,
-      "learning_rate": 5.006587615283268e-07,
-      "loss": 0.7323,
+      "grad_norm": 11.812898635864258,
+      "learning_rate": 7.509881422924901e-07,
+      "loss": 0.622,
       "step": 25900
     },
     {
       "epoch": 4.876717485413137,
-      "grad_norm": 17.26483726501465,
-      "learning_rate": 4.931300583474497e-07,
-      "loss": 0.4348,
+      "grad_norm": 6.675577163696289,
+      "learning_rate": 7.396950875211746e-07,
+      "loss": 0.3038,
       "step": 25910
     },
     {
       "epoch": 4.878599661208357,
-      "grad_norm": 23.311981201171875,
-      "learning_rate": 4.856013551665726e-07,
-      "loss": 0.7779,
+      "grad_norm": 9.983612060546875,
+      "learning_rate": 7.284020327498588e-07,
+      "loss": 0.7158,
       "step": 25920
     },
     {
       "epoch": 4.880481837003576,
-      "grad_norm": 19.61308479309082,
-      "learning_rate": 4.780726519856954e-07,
-      "loss": 0.8464,
+      "grad_norm": 8.77939510345459,
+      "learning_rate": 7.171089779785432e-07,
+      "loss": 0.7233,
       "step": 25930
     },
     {
       "epoch": 4.882364012798796,
-      "grad_norm": 7.677709579467773,
-      "learning_rate": 4.705439488048184e-07,
-      "loss": 0.8081,
+      "grad_norm": 11.423147201538086,
+      "learning_rate": 7.058159232072276e-07,
+      "loss": 0.8275,
       "step": 25940
     },
     {
       "epoch": 4.884246188594014,
-      "grad_norm": 13.238296508789062,
-      "learning_rate": 4.6301524562394134e-07,
-      "loss": 0.6175,
+      "grad_norm": 11.282443046569824,
+      "learning_rate": 6.94522868435912e-07,
+      "loss": 0.6037,
       "step": 25950
     },
     {
       "epoch": 4.886128364389234,
-      "grad_norm": 25.431175231933594,
-      "learning_rate": 4.554865424430642e-07,
-      "loss": 0.6642,
+      "grad_norm": 34.89209747314453,
+      "learning_rate": 6.832298136645963e-07,
+      "loss": 0.6057,
       "step": 25960
     },
     {
       "epoch": 4.888010540184453,
-      "grad_norm": 12.603880882263184,
-      "learning_rate": 4.4795783926218713e-07,
-      "loss": 0.6011,
+      "grad_norm": 13.334453582763672,
+      "learning_rate": 6.719367588932806e-07,
+      "loss": 0.5336,
       "step": 25970
     },
     {
       "epoch": 4.889892715979673,
-      "grad_norm": 43.16578674316406,
-      "learning_rate": 4.4042913608131005e-07,
-      "loss": 0.4965,
+      "grad_norm": 43.71876525878906,
+      "learning_rate": 6.60643704121965e-07,
+      "loss": 0.4498,
       "step": 25980
     },
     {
       "epoch": 4.891774891774892,
-      "grad_norm": 9.851216316223145,
-      "learning_rate": 4.329004329004329e-07,
-      "loss": 0.4479,
+      "grad_norm": 4.845729351043701,
+      "learning_rate": 6.493506493506493e-07,
+      "loss": 0.4448,
       "step": 25990
     },
     {
       "epoch": 4.893657067570111,
-      "grad_norm": 11.268643379211426,
-      "learning_rate": 4.253717297195559e-07,
-      "loss": 0.5438,
+      "grad_norm": 10.815642356872559,
+      "learning_rate": 6.380575945793338e-07,
+      "loss": 0.5838,
       "step": 26000
     },
     {
       "epoch": 4.89553924336533,
-      "grad_norm": 24.760019302368164,
-      "learning_rate": 4.1784302653867876e-07,
-      "loss": 0.5283,
+      "grad_norm": 17.238981246948242,
+      "learning_rate": 6.267645398080181e-07,
+      "loss": 0.444,
       "step": 26010
     },
     {
       "epoch": 4.897421419160549,
-      "grad_norm": 25.229755401611328,
-      "learning_rate": 4.1031432335780163e-07,
-      "loss": 0.784,
+      "grad_norm": 10.10824966430664,
+      "learning_rate": 6.154714850367024e-07,
+      "loss": 0.6663,
       "step": 26020
     },
     {
       "epoch": 4.899303594955769,
-      "grad_norm": 9.104552268981934,
-      "learning_rate": 4.027856201769246e-07,
-      "loss": 0.4686,
+      "grad_norm": 33.35478973388672,
+      "learning_rate": 6.041784302653868e-07,
+      "loss": 0.4297,
       "step": 26030
     },
     {
       "epoch": 4.901185770750988,
-      "grad_norm": 6.206742286682129,
-      "learning_rate": 3.9525691699604747e-07,
-      "loss": 0.7664,
+      "grad_norm": 0.8835696578025818,
+      "learning_rate": 5.928853754940712e-07,
+      "loss": 0.6774,
       "step": 26040
     },
     {
       "epoch": 4.903067946546208,
-      "grad_norm": 15.448500633239746,
-      "learning_rate": 3.8772821381517034e-07,
-      "loss": 0.3457,
+      "grad_norm": 5.606777191162109,
+      "learning_rate": 5.815923207227555e-07,
+      "loss": 0.3105,
       "step": 26050
     },
     {
       "epoch": 4.904950122341427,
-      "grad_norm": 31.12100601196289,
-      "learning_rate": 3.801995106342933e-07,
-      "loss": 0.5846,
+      "grad_norm": 13.368224143981934,
+      "learning_rate": 5.7029926595144e-07,
+      "loss": 0.6274,
       "step": 26060
     },
     {
       "epoch": 4.906832298136646,
-      "grad_norm": 14.74261474609375,
-      "learning_rate": 3.726708074534162e-07,
-      "loss": 0.6615,
+      "grad_norm": 17.68413543701172,
+      "learning_rate": 5.590062111801243e-07,
+      "loss": 0.5218,
       "step": 26070
     },
     {
       "epoch": 4.908714473931865,
-      "grad_norm": 26.152849197387695,
-      "learning_rate": 3.6514210427253905e-07,
-      "loss": 0.6266,
+      "grad_norm": 31.081344604492188,
+      "learning_rate": 5.477131564088085e-07,
+      "loss": 0.5692,
       "step": 26080
     },
     {
       "epoch": 4.9105966497270845,
-      "grad_norm": 27.336023330688477,
-      "learning_rate": 3.5761340109166203e-07,
-      "loss": 0.8022,
+      "grad_norm": 11.856854438781738,
+      "learning_rate": 5.36420101637493e-07,
+      "loss": 0.6386,
       "step": 26090
     },
     {
       "epoch": 4.912478825522304,
-      "grad_norm": 4.440485000610352,
-      "learning_rate": 3.500846979107849e-07,
-      "loss": 0.3889,
+      "grad_norm": 6.689940929412842,
+      "learning_rate": 5.251270468661773e-07,
+      "loss": 0.3447,
       "step": 26100
     },
     {
       "epoch": 4.914361001317523,
-      "grad_norm": 31.62215805053711,
-      "learning_rate": 3.4255599472990776e-07,
-      "loss": 0.2929,
+      "grad_norm": 36.05409622192383,
+      "learning_rate": 5.138339920948617e-07,
+      "loss": 0.2413,
       "step": 26110
     },
     {
       "epoch": 4.916243177112742,
-      "grad_norm": 18.474529266357422,
-      "learning_rate": 3.350272915490307e-07,
-      "loss": 0.7286,
+      "grad_norm": 21.475088119506836,
+      "learning_rate": 5.02540937323546e-07,
+      "loss": 0.6389,
       "step": 26120
     },
     {
       "epoch": 4.918125352907961,
-      "grad_norm": 1.3210591077804565,
-      "learning_rate": 3.274985883681536e-07,
-      "loss": 0.5917,
+      "grad_norm": 1.5936756134033203,
+      "learning_rate": 4.912478825522304e-07,
+      "loss": 0.4807,
       "step": 26130
     },
     {
       "epoch": 4.920007528703181,
-      "grad_norm": 22.190696716308594,
-      "learning_rate": 3.1996988518727653e-07,
-      "loss": 0.4259,
+      "grad_norm": 28.940305709838867,
+      "learning_rate": 4.799548277809147e-07,
+      "loss": 0.4051,
       "step": 26140
     },
     {
       "epoch": 4.9218897044984,
-      "grad_norm": 32.777915954589844,
-      "learning_rate": 3.1244118200639945e-07,
-      "loss": 0.458,
+      "grad_norm": 5.989145755767822,
+      "learning_rate": 4.686617730095991e-07,
+      "loss": 0.442,
       "step": 26150
     },
     {
       "epoch": 4.92377188029362,
-      "grad_norm": 8.841329574584961,
-      "learning_rate": 3.049124788255223e-07,
-      "loss": 0.6918,
+      "grad_norm": 11.877305030822754,
+      "learning_rate": 4.573687182382835e-07,
+      "loss": 0.6841,
       "step": 26160
     },
     {
       "epoch": 4.925654056088838,
-      "grad_norm": 12.418993949890137,
-      "learning_rate": 2.9738377564464524e-07,
-      "loss": 0.581,
+      "grad_norm": 10.223360061645508,
+      "learning_rate": 4.460756634669678e-07,
+      "loss": 0.5501,
       "step": 26170
     },
     {
       "epoch": 4.927536231884058,
-      "grad_norm": 8.253694534301758,
-      "learning_rate": 2.8985507246376816e-07,
-      "loss": 0.7705,
+      "grad_norm": 6.764021396636963,
+      "learning_rate": 4.347826086956522e-07,
+      "loss": 0.711,
       "step": 26180
     },
     {
       "epoch": 4.929418407679277,
-      "grad_norm": 9.630890846252441,
-      "learning_rate": 2.823263692828911e-07,
-      "loss": 0.594,
+      "grad_norm": 9.022865295410156,
+      "learning_rate": 4.2348955392433657e-07,
+      "loss": 0.5528,
       "step": 26190
     },
     {
       "epoch": 4.9313005834744965,
-      "grad_norm": 0.2443116456270218,
-      "learning_rate": 2.7479766610201395e-07,
-      "loss": 0.9983,
+      "grad_norm": 0.19420573115348816,
+      "learning_rate": 4.121964991530209e-07,
+      "loss": 0.8545,
       "step": 26200
     },
     {
       "epoch": 4.933182759269716,
-      "grad_norm": 12.538666725158691,
-      "learning_rate": 2.6726896292113687e-07,
-      "loss": 0.8449,
+      "grad_norm": 32.37028884887695,
+      "learning_rate": 4.009034443817053e-07,
+      "loss": 0.6968,
       "step": 26210
     },
     {
       "epoch": 4.935064935064935,
-      "grad_norm": 7.572700023651123,
-      "learning_rate": 2.597402597402598e-07,
-      "loss": 0.4725,
+      "grad_norm": 5.03609561920166,
+      "learning_rate": 3.8961038961038966e-07,
+      "loss": 0.5307,
       "step": 26220
     },
     {
       "epoch": 4.936947110860155,
-      "grad_norm": 7.338517189025879,
-      "learning_rate": 2.5221155655938266e-07,
-      "loss": 0.8398,
+      "grad_norm": 7.052425861358643,
+      "learning_rate": 3.78317334839074e-07,
+      "loss": 0.8121,
       "step": 26230
     },
     {
       "epoch": 4.938829286655373,
-      "grad_norm": 4.189211368560791,
-      "learning_rate": 2.446828533785056e-07,
-      "loss": 0.5246,
+      "grad_norm": 0.8062214255332947,
+      "learning_rate": 3.670242800677583e-07,
+      "loss": 0.4944,
       "step": 26240
     },
     {
       "epoch": 4.940711462450593,
-      "grad_norm": 10.774524688720703,
-      "learning_rate": 2.3715415019762845e-07,
-      "loss": 0.4691,
+      "grad_norm": 10.001143455505371,
+      "learning_rate": 3.5573122529644265e-07,
+      "loss": 0.3683,
       "step": 26250
     },
     {
       "epoch": 4.942593638245812,
-      "grad_norm": 0.5024207234382629,
-      "learning_rate": 2.2962544701675137e-07,
-      "loss": 0.5102,
+      "grad_norm": 1.526486873626709,
+      "learning_rate": 3.4443817052512703e-07,
+      "loss": 0.3605,
       "step": 26260
     },
     {
       "epoch": 4.944475814041032,
-      "grad_norm": 5.542119026184082,
-      "learning_rate": 2.220967438358743e-07,
-      "loss": 0.5087,
+      "grad_norm": 11.505463600158691,
+      "learning_rate": 3.331451157538114e-07,
+      "loss": 0.498,
       "step": 26270
     },
     {
       "epoch": 4.946357989836251,
-      "grad_norm": 17.346418380737305,
-      "learning_rate": 2.145680406549972e-07,
-      "loss": 0.512,
+      "grad_norm": 18.725603103637695,
+      "learning_rate": 3.2185206098249575e-07,
+      "loss": 0.3436,
       "step": 26280
     },
     {
       "epoch": 4.94824016563147,
-      "grad_norm": 7.274904251098633,
-      "learning_rate": 2.0703933747412008e-07,
-      "loss": 0.5541,
+      "grad_norm": 8.331537246704102,
+      "learning_rate": 3.1055900621118013e-07,
+      "loss": 0.5627,
       "step": 26290
     },
     {
       "epoch": 4.950122341426689,
-      "grad_norm": 7.677774429321289,
-      "learning_rate": 1.99510634293243e-07,
-      "loss": 0.5137,
+      "grad_norm": 10.584824562072754,
+      "learning_rate": 2.992659514398645e-07,
+      "loss": 0.4182,
       "step": 26300
     },
     {
       "epoch": 4.9520045172219085,
-      "grad_norm": 0.39065951108932495,
-      "learning_rate": 1.919819311123659e-07,
-      "loss": 0.6084,
+      "grad_norm": 0.15740695595741272,
+      "learning_rate": 2.8797289666854884e-07,
+      "loss": 0.5521,
       "step": 26310
     },
     {
       "epoch": 4.953886693017128,
-      "grad_norm": 28.504114151000977,
-      "learning_rate": 1.8445322793148882e-07,
-      "loss": 0.5155,
+      "grad_norm": 11.263009071350098,
+      "learning_rate": 2.766798418972332e-07,
+      "loss": 0.4944,
       "step": 26320
     },
     {
       "epoch": 4.955768868812347,
-      "grad_norm": 11.31637954711914,
-      "learning_rate": 1.7692452475061174e-07,
-      "loss": 0.7137,
+      "grad_norm": 7.119065284729004,
+      "learning_rate": 2.653867871259176e-07,
+      "loss": 0.7362,
       "step": 26330
     },
     {
       "epoch": 4.957651044607566,
-      "grad_norm": 1.688535213470459,
-      "learning_rate": 1.693958215697346e-07,
-      "loss": 0.2389,
+      "grad_norm": 1.8257107734680176,
+      "learning_rate": 2.5409373235460193e-07,
+      "loss": 0.1585,
       "step": 26340
     },
     {
       "epoch": 4.959533220402785,
-      "grad_norm": 8.698450088500977,
-      "learning_rate": 1.6186711838885753e-07,
-      "loss": 0.4807,
+      "grad_norm": 33.10144805908203,
+      "learning_rate": 2.4280067758328626e-07,
+      "loss": 0.4476,
       "step": 26350
     },
     {
       "epoch": 4.961415396198005,
-      "grad_norm": 30.824499130249023,
-      "learning_rate": 1.5433841520798043e-07,
-      "loss": 0.5214,
+      "grad_norm": 2.7676703929901123,
+      "learning_rate": 2.3150762281197064e-07,
+      "loss": 0.366,
       "step": 26360
     },
     {
       "epoch": 4.963297571993224,
-      "grad_norm": 2.2312653064727783,
-      "learning_rate": 1.4680971202710335e-07,
-      "loss": 0.6085,
+      "grad_norm": 0.12121053785085678,
+      "learning_rate": 2.2021456804065503e-07,
+      "loss": 0.4664,
       "step": 26370
     },
     {
       "epoch": 4.965179747788444,
-      "grad_norm": 2.4065072536468506,
-      "learning_rate": 1.3928100884622625e-07,
-      "loss": 0.4064,
+      "grad_norm": 0.2563054859638214,
+      "learning_rate": 2.0892151326933935e-07,
+      "loss": 0.3163,
       "step": 26380
     },
     {
       "epoch": 4.967061923583663,
-      "grad_norm": 17.495332717895508,
-      "learning_rate": 1.3175230566534914e-07,
-      "loss": 0.5247,
+      "grad_norm": 15.612191200256348,
+      "learning_rate": 1.976284584980237e-07,
+      "loss": 0.4692,
       "step": 26390
     },
     {
       "epoch": 4.9689440993788825,
-      "grad_norm": 14.244489669799805,
-      "learning_rate": 1.2422360248447206e-07,
-      "loss": 0.5497,
+      "grad_norm": 12.518272399902344,
+      "learning_rate": 1.8633540372670807e-07,
+      "loss": 0.5011,
       "step": 26400
     },
     {
       "epoch": 4.970826275174101,
-      "grad_norm": 6.616152286529541,
-      "learning_rate": 1.1669489930359497e-07,
-      "loss": 0.7547,
+      "grad_norm": 8.736549377441406,
+      "learning_rate": 1.7504234895539245e-07,
+      "loss": 0.7325,
       "step": 26410
     },
     {
       "epoch": 4.9727084509693205,
-      "grad_norm": 36.595096588134766,
-      "learning_rate": 1.0916619612271786e-07,
-      "loss": 0.6431,
+      "grad_norm": 45.901878356933594,
+      "learning_rate": 1.637492941840768e-07,
+      "loss": 0.5119,
       "step": 26420
     },
     {
       "epoch": 4.97459062676454,
-      "grad_norm": 17.054536819458008,
-      "learning_rate": 1.0163749294184077e-07,
-      "loss": 0.4411,
+      "grad_norm": 15.459980010986328,
+      "learning_rate": 1.5245623941276113e-07,
+      "loss": 0.4057,
       "step": 26430
     },
     {
       "epoch": 4.976472802559759,
-      "grad_norm": 14.136749267578125,
-      "learning_rate": 9.410878976096368e-08,
-      "loss": 0.8405,
+      "grad_norm": 12.372354507446289,
+      "learning_rate": 1.4116318464144551e-07,
+      "loss": 0.9019,
       "step": 26440
     },
     {
       "epoch": 4.978354978354979,
-      "grad_norm": 9.22201156616211,
-      "learning_rate": 8.658008658008659e-08,
-      "loss": 0.3544,
+      "grad_norm": 8.161259651184082,
+      "learning_rate": 1.2987012987012987e-07,
+      "loss": 0.3162,
       "step": 26450
     },
     {
       "epoch": 4.980237154150197,
-      "grad_norm": 19.09043312072754,
-      "learning_rate": 7.905138339920948e-08,
-      "loss": 0.5576,
+      "grad_norm": 2.586444854736328,
+      "learning_rate": 1.1857707509881423e-07,
+      "loss": 0.6493,
       "step": 26460
     },
     {
       "epoch": 4.982119329945417,
-      "grad_norm": 4.943528652191162,
-      "learning_rate": 7.152268021833239e-08,
-      "loss": 0.3464,
+      "grad_norm": 0.4139041602611542,
+      "learning_rate": 1.0728402032749858e-07,
+      "loss": 0.3598,
       "step": 26470
     },
     {
       "epoch": 4.984001505740636,
-      "grad_norm": 1.3135852813720703,
-      "learning_rate": 6.39939770374553e-08,
-      "loss": 0.5733,
+      "grad_norm": 0.5446306467056274,
+      "learning_rate": 9.599096555618295e-08,
+      "loss": 0.4985,
       "step": 26480
     },
     {
       "epoch": 4.985883681535856,
-      "grad_norm": 19.596176147460938,
-      "learning_rate": 5.64652738565782e-08,
-      "loss": 0.4562,
+      "grad_norm": 14.608466148376465,
+      "learning_rate": 8.46979107848673e-08,
+      "loss": 0.3913,
       "step": 26490
     },
     {
       "epoch": 4.987765857331075,
-      "grad_norm": 31.44451141357422,
-      "learning_rate": 4.893657067570112e-08,
-      "loss": 0.6244,
+      "grad_norm": 27.869840621948242,
+      "learning_rate": 7.340485601355166e-08,
+      "loss": 0.5136,
       "step": 26500
     },
     {
       "epoch": 4.989648033126294,
-      "grad_norm": 10.986721992492676,
-      "learning_rate": 4.1407867494824025e-08,
-      "loss": 0.6193,
+      "grad_norm": 7.235048294067383,
+      "learning_rate": 6.211180124223603e-08,
+      "loss": 0.5271,
       "step": 26510
     },
     {
       "epoch": 4.991530208921513,
-      "grad_norm": 10.592276573181152,
-      "learning_rate": 3.3879164313946926e-08,
-      "loss": 0.649,
+      "grad_norm": 16.698986053466797,
+      "learning_rate": 5.0818746470920386e-08,
+      "loss": 0.6373,
       "step": 26520
     },
     {
       "epoch": 4.993412384716732,
-      "grad_norm": 11.80045223236084,
-      "learning_rate": 2.635046113306983e-08,
-      "loss": 0.5307,
+      "grad_norm": 7.239467620849609,
+      "learning_rate": 3.952569169960475e-08,
+      "loss": 0.4937,
       "step": 26530
     },
     {
       "epoch": 4.995294560511952,
-      "grad_norm": 24.187894821166992,
-      "learning_rate": 1.8821757952192736e-08,
-      "loss": 0.7837,
+      "grad_norm": 33.553871154785156,
+      "learning_rate": 2.8232636928289104e-08,
+      "loss": 0.6561,
       "step": 26540
     },
     {
       "epoch": 4.997176736307171,
-      "grad_norm": 8.47293472290039,
-      "learning_rate": 1.1293054771315641e-08,
-      "loss": 0.6338,
+      "grad_norm": 11.409107208251953,
+      "learning_rate": 1.693958215697346e-08,
+      "loss": 0.5426,
       "step": 26550
     },
     {
       "epoch": 4.999058912102391,
-      "grad_norm": 63.684627532958984,
-      "learning_rate": 3.7643515904385476e-09,
-      "loss": 0.4886,
+      "grad_norm": 26.004404067993164,
+      "learning_rate": 5.6465273856578205e-09,
+      "loss": 0.4763,
       "step": 26560
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.9241333333333334,
-      "eval_loss": 0.2954551577568054,
-      "eval_runtime": 116.8437,
-      "eval_samples_per_second": 64.188,
-      "eval_steps_per_second": 8.028,
+      "eval_accuracy": 0.9252,
+      "eval_loss": 0.2850923240184784,
+      "eval_runtime": 277.3592,
+      "eval_samples_per_second": 27.041,
+      "eval_steps_per_second": 3.382,
       "step": 26565
     },
     {
       "epoch": 5.0,
       "step": 26565,
       "total_flos": 1.64815115092992e+19,
-      "train_loss": 0.0,
-      "train_runtime": 0.0821,
-      "train_samples_per_second": 2587573.233,
-      "train_steps_per_second": 323477.096
+      "train_loss": 1.0248974845253678,
+      "train_runtime": 12727.7584,
+      "train_samples_per_second": 16.696,
+      "train_steps_per_second": 2.087
     }
   ],
   "logging_steps": 10,