diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,9353 +1,13025 @@
 {
-  "best_metric": 0.985280588776449,
-  "best_model_checkpoint": "wav2vec2-present/checkpoint-15390",
-  "epoch": 9.996752192270218,
+  "best_metric": 0.9789265722752717,
+  "best_model_checkpoint": "wav2vec2-present/checkpoint-15057",
+  "epoch": 10.0,
   "eval_steps": 500,
-  "global_step": 15390,
+  "global_step": 21510,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.01,
-      "learning_rate": 1.9493177387914228e-07,
-      "loss": 2.564,
+      "epoch": 0.0,
+      "learning_rate": 1.394700139470014e-07,
+      "loss": 2.8932,
       "step": 10
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.8986354775828457e-07,
-      "loss": 2.558,
+      "learning_rate": 2.789400278940028e-07,
+      "loss": 2.8879,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.847953216374269e-07,
-      "loss": 2.5645,
+      "epoch": 0.01,
+      "learning_rate": 4.184100418410042e-07,
+      "loss": 2.8961,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 7.797270955165691e-07,
-      "loss": 2.568,
+      "epoch": 0.02,
+      "learning_rate": 5.578800557880056e-07,
+      "loss": 2.8933,
       "step": 40
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9.746588693957115e-07,
-      "loss": 2.5642,
+      "epoch": 0.02,
+      "learning_rate": 6.97350069735007e-07,
+      "loss": 2.8885,
       "step": 50
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.1695906432748538e-06,
-      "loss": 2.5622,
+      "epoch": 0.03,
+      "learning_rate": 8.368200836820084e-07,
+      "loss": 2.8916,
       "step": 60
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.364522417153996e-06,
-      "loss": 2.5609,
+      "epoch": 0.03,
+      "learning_rate": 9.762900976290097e-07,
+      "loss": 2.8944,
       "step": 70
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.5594541910331383e-06,
-      "loss": 2.5598,
+      "epoch": 0.04,
+      "learning_rate": 1.1157601115760112e-06,
+      "loss": 2.8856,
       "step": 80
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.7543859649122807e-06,
-      "loss": 2.552,
+      "epoch": 0.04,
+      "learning_rate": 1.2552301255230125e-06,
+      "loss": 2.8918,
       "step": 90
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.949317738791423e-06,
-      "loss": 2.5565,
+      "epoch": 0.05,
+      "learning_rate": 1.394700139470014e-06,
+      "loss": 2.8833,
       "step": 100
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 2.1442495126705653e-06,
-      "loss": 2.5584,
+      "epoch": 0.05,
+      "learning_rate": 1.5341701534170154e-06,
+      "loss": 2.8872,
       "step": 110
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.3391812865497075e-06,
-      "loss": 2.5492,
+      "epoch": 0.06,
+      "learning_rate": 1.6736401673640167e-06,
+      "loss": 2.8886,
       "step": 120
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 2.5341130604288498e-06,
-      "loss": 2.5532,
+      "epoch": 0.06,
+      "learning_rate": 1.8131101813110183e-06,
+      "loss": 2.8821,
       "step": 130
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 2.729044834307992e-06,
-      "loss": 2.5497,
+      "epoch": 0.07,
+      "learning_rate": 1.9525801952580194e-06,
+      "loss": 2.8814,
       "step": 140
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 2.9239766081871343e-06,
-      "loss": 2.549,
+      "epoch": 0.07,
+      "learning_rate": 2.092050209205021e-06,
+      "loss": 2.8848,
       "step": 150
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 3.1189083820662765e-06,
-      "loss": 2.546,
+      "epoch": 0.07,
+      "learning_rate": 2.2315202231520224e-06,
+      "loss": 2.8769,
       "step": 160
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 3.313840155945419e-06,
-      "loss": 2.5487,
+      "epoch": 0.08,
+      "learning_rate": 2.3709902370990235e-06,
+      "loss": 2.8785,
       "step": 170
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3.5087719298245615e-06,
-      "loss": 2.5346,
+      "epoch": 0.08,
+      "learning_rate": 2.510460251046025e-06,
+      "loss": 2.8798,
       "step": 180
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 3.7037037037037037e-06,
-      "loss": 2.5267,
+      "epoch": 0.09,
+      "learning_rate": 2.6499302649930266e-06,
+      "loss": 2.8646,
       "step": 190
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 3.898635477582846e-06,
-      "loss": 2.529,
+      "epoch": 0.09,
+      "learning_rate": 2.789400278940028e-06,
+      "loss": 2.8738,
       "step": 200
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.093567251461988e-06,
-      "loss": 2.5306,
+      "epoch": 0.1,
+      "learning_rate": 2.9288702928870297e-06,
+      "loss": 2.8659,
       "step": 210
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.2884990253411305e-06,
-      "loss": 2.5213,
+      "epoch": 0.1,
+      "learning_rate": 3.068340306834031e-06,
+      "loss": 2.8625,
       "step": 220
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.483430799220272e-06,
-      "loss": 2.512,
+      "epoch": 0.11,
+      "learning_rate": 3.207810320781032e-06,
+      "loss": 2.8687,
       "step": 230
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.678362573099415e-06,
-      "loss": 2.5016,
+      "epoch": 0.11,
+      "learning_rate": 3.3472803347280334e-06,
+      "loss": 2.861,
       "step": 240
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.873294346978558e-06,
-      "loss": 2.5006,
+      "epoch": 0.12,
+      "learning_rate": 3.486750348675035e-06,
+      "loss": 2.8517,
       "step": 250
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 5.0682261208576995e-06,
-      "loss": 2.5025,
+      "epoch": 0.12,
+      "learning_rate": 3.6262203626220365e-06,
+      "loss": 2.8432,
       "step": 260
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 5.263157894736842e-06,
-      "loss": 2.498,
+      "epoch": 0.13,
+      "learning_rate": 3.7656903765690376e-06,
+      "loss": 2.8486,
       "step": 270
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 5.458089668615984e-06,
-      "loss": 2.4611,
+      "epoch": 0.13,
+      "learning_rate": 3.905160390516039e-06,
+      "loss": 2.8497,
       "step": 280
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 5.653021442495127e-06,
-      "loss": 2.4517,
+      "epoch": 0.13,
+      "learning_rate": 4.044630404463041e-06,
+      "loss": 2.8504,
       "step": 290
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 5.8479532163742686e-06,
-      "loss": 2.4287,
+      "epoch": 0.14,
+      "learning_rate": 4.184100418410042e-06,
+      "loss": 2.8517,
       "step": 300
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 6.042884990253411e-06,
-      "loss": 2.4189,
+      "epoch": 0.14,
+      "learning_rate": 4.323570432357044e-06,
+      "loss": 2.8413,
       "step": 310
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 6.237816764132553e-06,
-      "loss": 2.3796,
+      "epoch": 0.15,
+      "learning_rate": 4.463040446304045e-06,
+      "loss": 2.8281,
       "step": 320
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 6.432748538011696e-06,
-      "loss": 2.3646,
+      "epoch": 0.15,
+      "learning_rate": 4.602510460251047e-06,
+      "loss": 2.8376,
       "step": 330
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 6.627680311890838e-06,
-      "loss": 2.3719,
+      "epoch": 0.16,
+      "learning_rate": 4.741980474198047e-06,
+      "loss": 2.8325,
       "step": 340
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 6.82261208576998e-06,
-      "loss": 2.3398,
+      "epoch": 0.16,
+      "learning_rate": 4.881450488145049e-06,
+      "loss": 2.8125,
       "step": 350
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 7.017543859649123e-06,
-      "loss": 2.2933,
+      "epoch": 0.17,
+      "learning_rate": 5.02092050209205e-06,
+      "loss": 2.7992,
       "step": 360
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 7.212475633528265e-06,
-      "loss": 2.3268,
+      "epoch": 0.17,
+      "learning_rate": 5.160390516039051e-06,
+      "loss": 2.7948,
       "step": 370
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 7.4074074074074075e-06,
-      "loss": 2.2196,
+      "epoch": 0.18,
+      "learning_rate": 5.299860529986053e-06,
+      "loss": 2.7747,
       "step": 380
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 7.602339181286549e-06,
-      "loss": 2.1905,
+      "epoch": 0.18,
+      "learning_rate": 5.439330543933054e-06,
+      "loss": 2.7668,
       "step": 390
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 7.797270955165692e-06,
-      "loss": 2.2123,
+      "epoch": 0.19,
+      "learning_rate": 5.578800557880056e-06,
+      "loss": 2.7652,
       "step": 400
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 7.992202729044834e-06,
-      "loss": 2.1655,
+      "epoch": 0.19,
+      "learning_rate": 5.718270571827057e-06,
+      "loss": 2.7773,
       "step": 410
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 8.187134502923976e-06,
-      "loss": 2.1171,
+      "epoch": 0.2,
+      "learning_rate": 5.857740585774059e-06,
+      "loss": 2.7606,
       "step": 420
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 8.38206627680312e-06,
-      "loss": 2.056,
+      "epoch": 0.2,
+      "learning_rate": 5.9972105997210605e-06,
+      "loss": 2.7637,
       "step": 430
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 8.576998050682261e-06,
-      "loss": 2.0953,
+      "epoch": 0.2,
+      "learning_rate": 6.136680613668062e-06,
+      "loss": 2.7201,
       "step": 440
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 8.771929824561403e-06,
-      "loss": 2.0215,
+      "epoch": 0.21,
+      "learning_rate": 6.276150627615063e-06,
+      "loss": 2.7098,
       "step": 450
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 8.966861598440545e-06,
-      "loss": 1.9845,
+      "epoch": 0.21,
+      "learning_rate": 6.415620641562064e-06,
+      "loss": 2.7002,
       "step": 460
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.161793372319688e-06,
-      "loss": 1.9909,
+      "epoch": 0.22,
+      "learning_rate": 6.555090655509066e-06,
+      "loss": 2.6489,
       "step": 470
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.35672514619883e-06,
-      "loss": 1.9478,
+      "epoch": 0.22,
+      "learning_rate": 6.680613668061368e-06,
+      "loss": 2.6457,
       "step": 480
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.551656920077972e-06,
-      "loss": 1.9098,
+      "epoch": 0.23,
+      "learning_rate": 6.820083682008369e-06,
+      "loss": 2.6344,
       "step": 490
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 9.746588693957115e-06,
-      "loss": 1.8296,
+      "epoch": 0.23,
+      "learning_rate": 6.95955369595537e-06,
+      "loss": 2.6388,
       "step": 500
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 9.941520467836257e-06,
-      "loss": 1.7486,
+      "epoch": 0.24,
+      "learning_rate": 7.099023709902371e-06,
+      "loss": 2.614,
       "step": 510
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.0136452241715399e-05,
-      "loss": 1.7578,
+      "epoch": 0.24,
+      "learning_rate": 7.238493723849372e-06,
+      "loss": 2.6101,
       "step": 520
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.0331384015594541e-05,
-      "loss": 1.708,
+      "epoch": 0.25,
+      "learning_rate": 7.377963737796374e-06,
+      "loss": 2.5636,
       "step": 530
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.0526315789473684e-05,
-      "loss": 1.5767,
+      "epoch": 0.25,
+      "learning_rate": 7.517433751743376e-06,
+      "loss": 2.5611,
       "step": 540
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.0721247563352826e-05,
-      "loss": 1.7068,
+      "epoch": 0.26,
+      "learning_rate": 7.656903765690377e-06,
+      "loss": 2.5388,
       "step": 550
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.0916179337231968e-05,
-      "loss": 1.6142,
+      "epoch": 0.26,
+      "learning_rate": 7.796373779637378e-06,
+      "loss": 2.5398,
       "step": 560
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.111111111111111e-05,
-      "loss": 1.584,
+      "epoch": 0.26,
+      "learning_rate": 7.93584379358438e-06,
+      "loss": 2.4898,
       "step": 570
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.1306042884990253e-05,
-      "loss": 1.523,
+      "epoch": 0.27,
+      "learning_rate": 8.07531380753138e-06,
+      "loss": 2.4637,
       "step": 580
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.1500974658869395e-05,
-      "loss": 1.4793,
+      "epoch": 0.27,
+      "learning_rate": 8.214783821478383e-06,
+      "loss": 2.4287,
       "step": 590
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.1695906432748537e-05,
-      "loss": 1.4863,
+      "epoch": 0.28,
+      "learning_rate": 8.354253835425383e-06,
+      "loss": 2.4482,
       "step": 600
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.1871345029239766e-05,
-      "loss": 1.4778,
+      "epoch": 0.28,
+      "learning_rate": 8.493723849372385e-06,
+      "loss": 2.4344,
       "step": 610
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.2066276803118908e-05,
-      "loss": 1.3872,
+      "epoch": 0.29,
+      "learning_rate": 8.633193863319387e-06,
+      "loss": 2.336,
       "step": 620
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.226120857699805e-05,
-      "loss": 1.412,
+      "epoch": 0.29,
+      "learning_rate": 8.77266387726639e-06,
+      "loss": 2.3321,
       "step": 630
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.2456140350877193e-05,
-      "loss": 1.4054,
+      "epoch": 0.3,
+      "learning_rate": 8.912133891213389e-06,
+      "loss": 2.3578,
       "step": 640
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.2651072124756335e-05,
-      "loss": 1.4,
+      "epoch": 0.3,
+      "learning_rate": 9.05160390516039e-06,
+      "loss": 2.3621,
       "step": 650
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.2846003898635477e-05,
-      "loss": 1.2976,
+      "epoch": 0.31,
+      "learning_rate": 9.191073919107393e-06,
+      "loss": 2.3569,
       "step": 660
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.3040935672514619e-05,
-      "loss": 1.3261,
+      "epoch": 0.31,
+      "learning_rate": 9.330543933054392e-06,
+      "loss": 2.2825,
       "step": 670
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.3235867446393762e-05,
-      "loss": 1.2583,
+      "epoch": 0.32,
+      "learning_rate": 9.470013947001395e-06,
+      "loss": 2.3117,
       "step": 680
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.3430799220272904e-05,
-      "loss": 1.3031,
+      "epoch": 0.32,
+      "learning_rate": 9.609483960948396e-06,
+      "loss": 2.2235,
       "step": 690
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.3625730994152046e-05,
-      "loss": 1.1888,
+      "epoch": 0.33,
+      "learning_rate": 9.748953974895399e-06,
+      "loss": 2.1834,
       "step": 700
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.3820662768031188e-05,
-      "loss": 1.2747,
+      "epoch": 0.33,
+      "learning_rate": 9.888423988842398e-06,
+      "loss": 2.1857,
       "step": 710
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.4015594541910331e-05,
-      "loss": 1.1909,
+      "epoch": 0.33,
+      "learning_rate": 1.0027894002789401e-05,
+      "loss": 2.1783,
       "step": 720
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.4210526315789473e-05,
-      "loss": 1.1954,
+      "epoch": 0.34,
+      "learning_rate": 1.0167364016736402e-05,
+      "loss": 2.151,
       "step": 730
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.4405458089668615e-05,
-      "loss": 1.1938,
+      "epoch": 0.34,
+      "learning_rate": 1.0306834030683405e-05,
+      "loss": 2.1209,
       "step": 740
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.4600389863547759e-05,
-      "loss": 1.1446,
+      "epoch": 0.35,
+      "learning_rate": 1.0446304044630404e-05,
+      "loss": 2.1109,
       "step": 750
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.47953216374269e-05,
-      "loss": 1.0875,
+      "epoch": 0.35,
+      "learning_rate": 1.0585774058577405e-05,
+      "loss": 2.056,
       "step": 760
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.4990253411306042e-05,
-      "loss": 1.1029,
+      "epoch": 0.36,
+      "learning_rate": 1.0725244072524408e-05,
+      "loss": 1.9979,
       "step": 770
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.5185185185185186e-05,
-      "loss": 0.9905,
+      "epoch": 0.36,
+      "learning_rate": 1.0864714086471408e-05,
+      "loss": 1.9882,
       "step": 780
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.538011695906433e-05,
-      "loss": 1.175,
+      "epoch": 0.37,
+      "learning_rate": 1.100418410041841e-05,
+      "loss": 1.9189,
       "step": 790
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.5555555555555555e-05,
-      "loss": 1.0142,
+      "epoch": 0.37,
+      "learning_rate": 1.1143654114365412e-05,
+      "loss": 1.9409,
       "step": 800
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.5750487329434697e-05,
-      "loss": 1.054,
+      "epoch": 0.38,
+      "learning_rate": 1.1283124128312414e-05,
+      "loss": 1.9101,
       "step": 810
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.594541910331384e-05,
-      "loss": 1.019,
+      "epoch": 0.38,
+      "learning_rate": 1.1422594142259414e-05,
+      "loss": 2.0193,
       "step": 820
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.614035087719298e-05,
-      "loss": 0.98,
+      "epoch": 0.39,
+      "learning_rate": 1.1562064156206417e-05,
+      "loss": 1.9348,
       "step": 830
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.6335282651072126e-05,
-      "loss": 0.8901,
+      "epoch": 0.39,
+      "learning_rate": 1.1701534170153418e-05,
+      "loss": 1.9025,
       "step": 840
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.6530214424951268e-05,
-      "loss": 0.9091,
+      "epoch": 0.4,
+      "learning_rate": 1.1841004184100417e-05,
+      "loss": 1.8682,
       "step": 850
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.672514619883041e-05,
-      "loss": 0.8969,
+      "epoch": 0.4,
+      "learning_rate": 1.198047419804742e-05,
+      "loss": 1.8374,
       "step": 860
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.692007797270955e-05,
-      "loss": 0.8246,
+      "epoch": 0.4,
+      "learning_rate": 1.2119944211994421e-05,
+      "loss": 1.7407,
       "step": 870
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.7115009746588693e-05,
-      "loss": 1.0441,
+      "epoch": 0.41,
+      "learning_rate": 1.2259414225941424e-05,
+      "loss": 1.7579,
       "step": 880
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.7309941520467835e-05,
-      "loss": 0.8832,
+      "epoch": 0.41,
+      "learning_rate": 1.2398884239888423e-05,
+      "loss": 1.7693,
       "step": 890
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.7504873294346977e-05,
-      "loss": 0.8898,
+      "epoch": 0.42,
+      "learning_rate": 1.2538354253835426e-05,
+      "loss": 1.725,
       "step": 900
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.7699805068226122e-05,
-      "loss": 0.8001,
+      "epoch": 0.42,
+      "learning_rate": 1.2677824267782427e-05,
+      "loss": 1.7028,
       "step": 910
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.7894736842105264e-05,
-      "loss": 0.7232,
+      "epoch": 0.43,
+      "learning_rate": 1.281729428172943e-05,
+      "loss": 1.728,
       "step": 920
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.8089668615984406e-05,
-      "loss": 0.8311,
+      "epoch": 0.43,
+      "learning_rate": 1.295676429567643e-05,
+      "loss": 1.6861,
       "step": 930
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.8284600389863547e-05,
-      "loss": 0.7449,
+      "epoch": 0.44,
+      "learning_rate": 1.309623430962343e-05,
+      "loss": 1.6026,
       "step": 940
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.847953216374269e-05,
-      "loss": 0.8162,
+      "epoch": 0.44,
+      "learning_rate": 1.3235704323570433e-05,
+      "loss": 1.6721,
       "step": 950
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.867446393762183e-05,
-      "loss": 0.7072,
+      "epoch": 0.45,
+      "learning_rate": 1.3375174337517433e-05,
+      "loss": 1.6533,
       "step": 960
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.8869395711500973e-05,
-      "loss": 0.7233,
+      "epoch": 0.45,
+      "learning_rate": 1.3514644351464436e-05,
+      "loss": 1.5843,
       "step": 970
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.9064327485380115e-05,
-      "loss": 0.8823,
+      "epoch": 0.46,
+      "learning_rate": 1.3654114365411437e-05,
+      "loss": 1.6248,
       "step": 980
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.925925925925926e-05,
-      "loss": 0.71,
+      "epoch": 0.46,
+      "learning_rate": 1.379358437935844e-05,
+      "loss": 1.5543,
       "step": 990
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.9454191033138402e-05,
-      "loss": 0.6826,
+      "epoch": 0.46,
+      "learning_rate": 1.3933054393305439e-05,
+      "loss": 1.6078,
       "step": 1000
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.9649122807017544e-05,
-      "loss": 0.7723,
+      "epoch": 0.47,
+      "learning_rate": 1.4072524407252442e-05,
+      "loss": 1.4884,
       "step": 1010
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.9844054580896685e-05,
-      "loss": 0.7333,
+      "epoch": 0.47,
+      "learning_rate": 1.4211994421199443e-05,
+      "loss": 1.5583,
       "step": 1020
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 2.0038986354775827e-05,
-      "loss": 0.6399,
+      "epoch": 0.48,
+      "learning_rate": 1.4337517433751745e-05,
+      "loss": 1.4856,
       "step": 1030
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 2.023391812865497e-05,
-      "loss": 0.5773,
+      "epoch": 0.48,
+      "learning_rate": 1.4476987447698744e-05,
+      "loss": 1.4153,
       "step": 1040
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 2.042884990253411e-05,
-      "loss": 0.5524,
+      "epoch": 0.49,
+      "learning_rate": 1.4616457461645747e-05,
+      "loss": 1.5187,
       "step": 1050
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 2.0623781676413256e-05,
-      "loss": 0.6107,
+      "epoch": 0.49,
+      "learning_rate": 1.4755927475592748e-05,
+      "loss": 1.4212,
       "step": 1060
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 2.0818713450292398e-05,
-      "loss": 0.515,
+      "epoch": 0.5,
+      "learning_rate": 1.4895397489539749e-05,
+      "loss": 1.3842,
       "step": 1070
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 2.101364522417154e-05,
-      "loss": 0.5981,
+      "epoch": 0.5,
+      "learning_rate": 1.5034867503486752e-05,
+      "loss": 1.3595,
       "step": 1080
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 2.120857699805068e-05,
-      "loss": 0.6844,
+      "epoch": 0.51,
+      "learning_rate": 1.5174337517433753e-05,
+      "loss": 1.3853,
       "step": 1090
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 2.1403508771929824e-05,
-      "loss": 0.4908,
+      "epoch": 0.51,
+      "learning_rate": 1.5313807531380754e-05,
+      "loss": 1.2961,
       "step": 1100
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 2.1598440545808965e-05,
-      "loss": 0.5733,
+      "epoch": 0.52,
+      "learning_rate": 1.5453277545327754e-05,
+      "loss": 1.2973,
       "step": 1110
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 2.1793372319688107e-05,
-      "loss": 0.4578,
+      "epoch": 0.52,
+      "learning_rate": 1.5592747559274756e-05,
+      "loss": 1.2672,
       "step": 1120
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 2.1988304093567252e-05,
-      "loss": 0.4443,
+      "epoch": 0.53,
+      "learning_rate": 1.573221757322176e-05,
+      "loss": 1.3161,
       "step": 1130
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 2.2183235867446394e-05,
-      "loss": 0.4854,
+      "epoch": 0.53,
+      "learning_rate": 1.587168758716876e-05,
+      "loss": 1.2184,
       "step": 1140
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 2.2378167641325536e-05,
-      "loss": 0.4891,
+      "epoch": 0.53,
+      "learning_rate": 1.601115760111576e-05,
+      "loss": 1.2968,
       "step": 1150
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 2.2573099415204678e-05,
-      "loss": 0.4287,
+      "epoch": 0.54,
+      "learning_rate": 1.615062761506276e-05,
+      "loss": 1.3205,
       "step": 1160
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 2.2748538011695907e-05,
-      "loss": 0.7449,
+      "epoch": 0.54,
+      "learning_rate": 1.6290097629009764e-05,
+      "loss": 1.1363,
       "step": 1170
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.294346978557505e-05,
-      "loss": 0.4233,
+      "epoch": 0.55,
+      "learning_rate": 1.6429567642956766e-05,
+      "loss": 1.1579,
       "step": 1180
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 2.313840155945419e-05,
-      "loss": 0.749,
+      "epoch": 0.55,
+      "learning_rate": 1.6569037656903766e-05,
+      "loss": 1.1291,
       "step": 1190
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 2.3333333333333336e-05,
-      "loss": 0.4584,
+      "epoch": 0.56,
+      "learning_rate": 1.6708507670850765e-05,
+      "loss": 1.1111,
       "step": 1200
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.3528265107212478e-05,
-      "loss": 0.379,
+      "epoch": 0.56,
+      "learning_rate": 1.6847977684797768e-05,
+      "loss": 1.195,
       "step": 1210
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 2.372319688109162e-05,
-      "loss": 0.5038,
+      "epoch": 0.57,
+      "learning_rate": 1.698744769874477e-05,
+      "loss": 1.1692,
       "step": 1220
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 2.391812865497076e-05,
-      "loss": 0.4659,
+      "epoch": 0.57,
+      "learning_rate": 1.7126917712691774e-05,
+      "loss": 1.091,
       "step": 1230
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.4113060428849903e-05,
-      "loss": 0.5737,
+      "epoch": 0.58,
+      "learning_rate": 1.7266387726638773e-05,
+      "loss": 1.0087,
       "step": 1240
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 2.4307992202729045e-05,
-      "loss": 0.4579,
+      "epoch": 0.58,
+      "learning_rate": 1.7405857740585773e-05,
+      "loss": 1.0821,
       "step": 1250
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.4502923976608187e-05,
-      "loss": 0.4404,
+      "epoch": 0.59,
+      "learning_rate": 1.754532775453278e-05,
+      "loss": 1.0786,
       "step": 1260
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 2.4697855750487332e-05,
-      "loss": 0.3332,
+      "epoch": 0.59,
+      "learning_rate": 1.7684797768479778e-05,
+      "loss": 1.0286,
       "step": 1270
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 2.4892787524366474e-05,
-      "loss": 0.4733,
+      "epoch": 0.6,
+      "learning_rate": 1.7824267782426778e-05,
+      "loss": 1.0807,
       "step": 1280
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.5087719298245616e-05,
-      "loss": 0.4872,
+      "epoch": 0.6,
+      "learning_rate": 1.796373779637378e-05,
+      "loss": 1.0336,
       "step": 1290
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 2.5282651072124758e-05,
-      "loss": 0.473,
+      "epoch": 0.6,
+      "learning_rate": 1.810320781032078e-05,
+      "loss": 1.0355,
       "step": 1300
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 2.54775828460039e-05,
-      "loss": 0.4366,
+      "epoch": 0.61,
+      "learning_rate": 1.8242677824267783e-05,
+      "loss": 0.9392,
       "step": 1310
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.567251461988304e-05,
-      "loss": 0.3755,
+      "epoch": 0.61,
+      "learning_rate": 1.8382147838214785e-05,
+      "loss": 0.9515,
       "step": 1320
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 2.5867446393762183e-05,
-      "loss": 0.6943,
+      "epoch": 0.62,
+      "learning_rate": 1.8521617852161785e-05,
+      "loss": 0.9423,
       "step": 1330
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 2.6062378167641325e-05,
-      "loss": 0.3036,
+      "epoch": 0.62,
+      "learning_rate": 1.8661087866108784e-05,
+      "loss": 0.8123,
       "step": 1340
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.625730994152047e-05,
-      "loss": 0.4545,
+      "epoch": 0.63,
+      "learning_rate": 1.880055788005579e-05,
+      "loss": 1.0167,
       "step": 1350
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 2.6452241715399612e-05,
-      "loss": 0.3032,
+      "epoch": 0.63,
+      "learning_rate": 1.894002789400279e-05,
+      "loss": 0.9288,
       "step": 1360
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 2.6647173489278754e-05,
-      "loss": 0.3492,
+      "epoch": 0.64,
+      "learning_rate": 1.9079497907949793e-05,
+      "loss": 0.9445,
       "step": 1370
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.6842105263157896e-05,
-      "loss": 0.2846,
+      "epoch": 0.64,
+      "learning_rate": 1.9218967921896792e-05,
+      "loss": 0.8053,
       "step": 1380
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 2.7037037037037037e-05,
-      "loss": 0.5078,
+      "epoch": 0.65,
+      "learning_rate": 1.9358437935843795e-05,
+      "loss": 0.9432,
       "step": 1390
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 2.7212475633528263e-05,
-      "loss": 0.3074,
+      "epoch": 0.65,
+      "learning_rate": 1.9497907949790798e-05,
+      "loss": 0.8761,
       "step": 1400
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.7407407407407408e-05,
-      "loss": 0.2837,
+      "epoch": 0.66,
+      "learning_rate": 1.9637377963737797e-05,
+      "loss": 0.8646,
       "step": 1410
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 2.760233918128655e-05,
-      "loss": 0.4813,
+      "epoch": 0.66,
+      "learning_rate": 1.9776847977684796e-05,
+      "loss": 0.8397,
       "step": 1420
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 2.7797270955165692e-05,
-      "loss": 0.3597,
+      "epoch": 0.66,
+      "learning_rate": 1.99163179916318e-05,
+      "loss": 0.8022,
       "step": 1430
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.7992202729044834e-05,
-      "loss": 0.1955,
+      "epoch": 0.67,
+      "learning_rate": 2.0055788005578802e-05,
+      "loss": 0.6888,
       "step": 1440
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.8187134502923976e-05,
-      "loss": 0.5268,
+      "epoch": 0.67,
+      "learning_rate": 2.01952580195258e-05,
+      "loss": 0.8969,
       "step": 1450
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.8382066276803117e-05,
-      "loss": 0.5699,
+      "epoch": 0.68,
+      "learning_rate": 2.0334728033472804e-05,
+      "loss": 0.7044,
       "step": 1460
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.857699805068226e-05,
-      "loss": 0.3092,
+      "epoch": 0.68,
+      "learning_rate": 2.0474198047419804e-05,
+      "loss": 0.6346,
       "step": 1470
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 2.87719298245614e-05,
-      "loss": 0.3738,
+      "epoch": 0.69,
+      "learning_rate": 2.061366806136681e-05,
+      "loss": 0.7734,
       "step": 1480
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 2.8966861598440546e-05,
-      "loss": 0.5783,
+      "epoch": 0.69,
+      "learning_rate": 2.075313807531381e-05,
+      "loss": 0.7262,
       "step": 1490
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 2.9161793372319688e-05,
-      "loss": 0.3053,
+      "epoch": 0.7,
+      "learning_rate": 2.089260808926081e-05,
+      "loss": 0.6095,
       "step": 1500
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 2.935672514619883e-05,
-      "loss": 0.3565,
+      "epoch": 0.7,
+      "learning_rate": 2.103207810320781e-05,
+      "loss": 0.7233,
       "step": 1510
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 2.9551656920077972e-05,
-      "loss": 0.3092,
+      "epoch": 0.71,
+      "learning_rate": 2.1157601115760115e-05,
+      "loss": 0.7737,
       "step": 1520
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 2.9746588693957114e-05,
-      "loss": 0.5208,
+      "epoch": 0.71,
+      "learning_rate": 2.1297071129707115e-05,
+      "loss": 0.8418,
       "step": 1530
     },
     {
-      "epoch": 1.0,
-      "eval_accuracy": 0.9425022999080037,
-      "eval_loss": 0.22455397248268127,
-      "eval_runtime": 13.2379,
-      "eval_samples_per_second": 164.226,
-      "eval_steps_per_second": 41.094,
-      "step": 1539
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 2.9941520467836255e-05,
-      "loss": 0.4893,
+      "epoch": 0.72,
+      "learning_rate": 2.1436541143654114e-05,
+      "loss": 0.5966,
       "step": 1540
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 2.99848386398094e-05,
-      "loss": 0.2747,
+      "epoch": 0.72,
+      "learning_rate": 2.1576011157601117e-05,
+      "loss": 0.7214,
       "step": 1550
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 2.9963179553822827e-05,
-      "loss": 0.6032,
+      "epoch": 0.73,
+      "learning_rate": 2.1715481171548116e-05,
+      "loss": 0.8499,
       "step": 1560
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 2.9941520467836255e-05,
-      "loss": 0.265,
+      "epoch": 0.73,
+      "learning_rate": 2.185495118549512e-05,
+      "loss": 0.6569,
       "step": 1570
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 2.9919861381849688e-05,
-      "loss": 0.4532,
+      "epoch": 0.73,
+      "learning_rate": 2.1994421199442122e-05,
+      "loss": 0.6469,
       "step": 1580
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 2.9898202295863116e-05,
-      "loss": 0.3197,
+      "epoch": 0.74,
+      "learning_rate": 2.2119944211994422e-05,
+      "loss": 0.6156,
       "step": 1590
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 2.9876543209876545e-05,
-      "loss": 0.3362,
+      "epoch": 0.74,
+      "learning_rate": 2.2259414225941425e-05,
+      "loss": 0.6293,
       "step": 1600
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 2.9854884123889974e-05,
-      "loss": 0.3188,
+      "epoch": 0.75,
+      "learning_rate": 2.2398884239888424e-05,
+      "loss": 0.6863,
       "step": 1610
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 2.9833225037903402e-05,
-      "loss": 0.3177,
+      "epoch": 0.75,
+      "learning_rate": 2.2538354253835427e-05,
+      "loss": 0.6216,
       "step": 1620
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 2.981156595191683e-05,
-      "loss": 0.4515,
+      "epoch": 0.76,
+      "learning_rate": 2.2677824267782426e-05,
+      "loss": 0.6734,
       "step": 1630
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 2.9789906865930256e-05,
-      "loss": 0.2545,
+      "epoch": 0.76,
+      "learning_rate": 2.2817294281729426e-05,
+      "loss": 0.5728,
       "step": 1640
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 2.9768247779943685e-05,
-      "loss": 0.2263,
+      "epoch": 0.77,
+      "learning_rate": 2.2956764295676432e-05,
+      "loss": 0.5693,
       "step": 1650
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 2.9746588693957114e-05,
-      "loss": 0.4115,
+      "epoch": 0.77,
+      "learning_rate": 2.309623430962343e-05,
+      "loss": 0.617,
       "step": 1660
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 2.9724929607970546e-05,
-      "loss": 0.2845,
+      "epoch": 0.78,
+      "learning_rate": 2.3235704323570434e-05,
+      "loss": 0.5586,
       "step": 1670
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 2.9703270521983974e-05,
-      "loss": 0.3104,
+      "epoch": 0.78,
+      "learning_rate": 2.3375174337517434e-05,
+      "loss": 0.7137,
       "step": 1680
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 2.9681611435997403e-05,
-      "loss": 0.3711,
+      "epoch": 0.79,
+      "learning_rate": 2.3514644351464436e-05,
+      "loss": 0.6104,
       "step": 1690
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 2.9659952350010832e-05,
-      "loss": 0.3236,
+      "epoch": 0.79,
+      "learning_rate": 2.365411436541144e-05,
+      "loss": 0.4187,
       "step": 1700
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 2.9638293264024257e-05,
-      "loss": 0.216,
+      "epoch": 0.79,
+      "learning_rate": 2.379358437935844e-05,
+      "loss": 0.6468,
       "step": 1710
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 2.9616634178037686e-05,
-      "loss": 0.3934,
+      "epoch": 0.8,
+      "learning_rate": 2.3933054393305438e-05,
+      "loss": 0.5058,
       "step": 1720
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 2.9594975092051114e-05,
-      "loss": 0.6872,
+      "epoch": 0.8,
+      "learning_rate": 2.407252440725244e-05,
+      "loss": 0.5845,
       "step": 1730
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 2.9573316006064543e-05,
-      "loss": 0.5262,
+      "epoch": 0.81,
+      "learning_rate": 2.4211994421199444e-05,
+      "loss": 0.488,
       "step": 1740
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 2.9551656920077972e-05,
-      "loss": 0.3703,
+      "epoch": 0.81,
+      "learning_rate": 2.4351464435146443e-05,
+      "loss": 0.5375,
       "step": 1750
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 2.9529997834091404e-05,
-      "loss": 0.3206,
+      "epoch": 0.82,
+      "learning_rate": 2.4490934449093446e-05,
+      "loss": 0.6187,
       "step": 1760
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 2.9508338748104833e-05,
-      "loss": 0.3157,
+      "epoch": 0.82,
+      "learning_rate": 2.4630404463040445e-05,
+      "loss": 0.4766,
       "step": 1770
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 2.948667966211826e-05,
-      "loss": 0.3127,
+      "epoch": 0.83,
+      "learning_rate": 2.4769874476987448e-05,
+      "loss": 0.5328,
       "step": 1780
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 2.9465020576131687e-05,
-      "loss": 0.2422,
+      "epoch": 0.83,
+      "learning_rate": 2.490934449093445e-05,
+      "loss": 0.3891,
       "step": 1790
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 2.9443361490145115e-05,
-      "loss": 0.6096,
+      "epoch": 0.84,
+      "learning_rate": 2.504881450488145e-05,
+      "loss": 0.4496,
       "step": 1800
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 2.9421702404158544e-05,
-      "loss": 0.4274,
+      "epoch": 0.84,
+      "learning_rate": 2.5188284518828453e-05,
+      "loss": 0.5077,
       "step": 1810
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 2.9400043318171973e-05,
-      "loss": 0.4009,
+      "epoch": 0.85,
+      "learning_rate": 2.5327754532775453e-05,
+      "loss": 0.6491,
       "step": 1820
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 2.93783842321854e-05,
-      "loss": 0.4787,
+      "epoch": 0.85,
+      "learning_rate": 2.5467224546722455e-05,
+      "loss": 0.5923,
       "step": 1830
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 2.935672514619883e-05,
-      "loss": 0.4098,
+      "epoch": 0.86,
+      "learning_rate": 2.5606694560669458e-05,
+      "loss": 0.5348,
       "step": 1840
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 2.9335066060212262e-05,
-      "loss": 0.4029,
+      "epoch": 0.86,
+      "learning_rate": 2.5746164574616458e-05,
+      "loss": 0.468,
       "step": 1850
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 2.931340697422569e-05,
-      "loss": 0.2639,
+      "epoch": 0.86,
+      "learning_rate": 2.5885634588563457e-05,
+      "loss": 0.6697,
       "step": 1860
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 2.9291747888239116e-05,
-      "loss": 0.4326,
+      "epoch": 0.87,
+      "learning_rate": 2.6025104602510463e-05,
+      "loss": 0.5131,
       "step": 1870
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 2.9270088802252545e-05,
-      "loss": 0.6265,
+      "epoch": 0.87,
+      "learning_rate": 2.6164574616457463e-05,
+      "loss": 0.3658,
       "step": 1880
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 2.9248429716265973e-05,
-      "loss": 0.453,
+      "epoch": 0.88,
+      "learning_rate": 2.6304044630404462e-05,
+      "loss": 0.4172,
       "step": 1890
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 2.9226770630279402e-05,
-      "loss": 0.3825,
+      "epoch": 0.88,
+      "learning_rate": 2.6443514644351465e-05,
+      "loss": 0.5263,
       "step": 1900
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 2.920511154429283e-05,
-      "loss": 0.567,
+      "epoch": 0.89,
+      "learning_rate": 2.6582984658298468e-05,
+      "loss": 0.6777,
       "step": 1910
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 2.918345245830626e-05,
-      "loss": 0.3227,
+      "epoch": 0.89,
+      "learning_rate": 2.672245467224547e-05,
+      "loss": 0.6603,
       "step": 1920
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 2.9161793372319688e-05,
-      "loss": 0.4686,
+      "epoch": 0.9,
+      "learning_rate": 2.686192468619247e-05,
+      "loss": 0.5585,
       "step": 1930
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 2.914013428633312e-05,
-      "loss": 0.4481,
+      "epoch": 0.9,
+      "learning_rate": 2.700139470013947e-05,
+      "loss": 0.3998,
       "step": 1940
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 2.9118475200346546e-05,
-      "loss": 0.3457,
+      "epoch": 0.91,
+      "learning_rate": 2.7140864714086472e-05,
+      "loss": 0.4931,
       "step": 1950
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 2.9096816114359974e-05,
-      "loss": 0.3616,
+      "epoch": 0.91,
+      "learning_rate": 2.7280334728033475e-05,
+      "loss": 0.4954,
       "step": 1960
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 2.9075157028373403e-05,
-      "loss": 0.3722,
+      "epoch": 0.92,
+      "learning_rate": 2.7419804741980474e-05,
+      "loss": 0.3263,
       "step": 1970
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 2.905349794238683e-05,
-      "loss": 0.4628,
+      "epoch": 0.92,
+      "learning_rate": 2.7559274755927477e-05,
+      "loss": 0.3234,
       "step": 1980
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 2.903183885640026e-05,
-      "loss": 0.1889,
+      "epoch": 0.93,
+      "learning_rate": 2.7698744769874477e-05,
+      "loss": 0.4315,
       "step": 1990
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 2.901017977041369e-05,
-      "loss": 0.4129,
+      "epoch": 0.93,
+      "learning_rate": 2.783821478382148e-05,
+      "loss": 0.6674,
       "step": 2000
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 2.8988520684427118e-05,
-      "loss": 0.2512,
+      "epoch": 0.93,
+      "learning_rate": 2.7977684797768482e-05,
+      "loss": 0.4425,
       "step": 2010
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 2.8966861598440546e-05,
-      "loss": 0.3462,
+      "epoch": 0.94,
+      "learning_rate": 2.811715481171548e-05,
+      "loss": 0.5932,
       "step": 2020
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 2.8945202512453975e-05,
-      "loss": 0.2496,
+      "epoch": 0.94,
+      "learning_rate": 2.825662482566248e-05,
+      "loss": 0.4652,
       "step": 2030
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 2.8923543426467404e-05,
-      "loss": 0.2584,
+      "epoch": 0.95,
+      "learning_rate": 2.8396094839609484e-05,
+      "loss": 0.2503,
       "step": 2040
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 2.8901884340480832e-05,
-      "loss": 0.4408,
+      "epoch": 0.95,
+      "learning_rate": 2.8535564853556487e-05,
+      "loss": 0.8047,
       "step": 2050
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 2.888022525449426e-05,
-      "loss": 0.1791,
+      "epoch": 0.96,
+      "learning_rate": 2.867503486750349e-05,
+      "loss": 0.4353,
       "step": 2060
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 2.885856616850769e-05,
-      "loss": 0.2065,
+      "epoch": 0.96,
+      "learning_rate": 2.881450488145049e-05,
+      "loss": 0.4179,
       "step": 2070
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 2.883690708252112e-05,
-      "loss": 0.4957,
+      "epoch": 0.97,
+      "learning_rate": 2.8953974895397488e-05,
+      "loss": 0.6257,
       "step": 2080
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 2.8815247996534547e-05,
-      "loss": 0.1517,
+      "epoch": 0.97,
+      "learning_rate": 2.9093444909344494e-05,
+      "loss": 0.4073,
       "step": 2090
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 2.8793588910547976e-05,
-      "loss": 0.2041,
+      "epoch": 0.98,
+      "learning_rate": 2.9232914923291494e-05,
+      "loss": 0.5766,
       "step": 2100
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 2.87719298245614e-05,
-      "loss": 0.2251,
+      "epoch": 0.98,
+      "learning_rate": 2.9372384937238493e-05,
+      "loss": 0.3263,
       "step": 2110
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 2.8750270738574833e-05,
-      "loss": 0.1827,
+      "epoch": 0.99,
+      "learning_rate": 2.9511854951185496e-05,
+      "loss": 0.4164,
       "step": 2120
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 2.8730777561186917e-05,
-      "loss": 0.1692,
+      "epoch": 0.99,
+      "learning_rate": 2.9651324965132495e-05,
+      "loss": 0.4837,
       "step": 2130
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 2.8709118475200346e-05,
-      "loss": 0.2938,
+      "epoch": 0.99,
+      "learning_rate": 2.9790794979079498e-05,
+      "loss": 0.5648,
       "step": 2140
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 2.8687459389213774e-05,
-      "loss": 0.2557,
+      "epoch": 1.0,
+      "learning_rate": 2.99302649930265e-05,
+      "loss": 0.5859,
       "step": 2150
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 2.8665800303227203e-05,
-      "loss": 0.2886,
+      "epoch": 1.0,
+      "eval_accuracy": 0.9351333552848206,
+      "eval_loss": 0.29516899585723877,
+      "eval_runtime": 17.4865,
+      "eval_samples_per_second": 173.677,
+      "eval_steps_per_second": 43.462,
+      "step": 2151
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9992251665891835e-05,
+      "loss": 0.5367,
       "step": 2160
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 2.8644141217240635e-05,
-      "loss": 0.3589,
+      "epoch": 1.01,
+      "learning_rate": 2.9976754997675502e-05,
+      "loss": 0.2168,
       "step": 2170
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 2.8622482131254064e-05,
-      "loss": 0.3187,
+      "epoch": 1.01,
+      "learning_rate": 2.9961258329459166e-05,
+      "loss": 0.3865,
       "step": 2180
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 2.8600823045267492e-05,
-      "loss": 0.2109,
+      "epoch": 1.02,
+      "learning_rate": 2.9945761661242834e-05,
+      "loss": 0.4129,
       "step": 2190
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 2.857916395928092e-05,
-      "loss": 0.3777,
+      "epoch": 1.02,
+      "learning_rate": 2.99302649930265e-05,
+      "loss": 0.4894,
       "step": 2200
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 2.8557504873294346e-05,
-      "loss": 0.3847,
+      "epoch": 1.03,
+      "learning_rate": 2.991476832481017e-05,
+      "loss": 0.3326,
       "step": 2210
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 2.8535845787307775e-05,
-      "loss": 0.5642,
+      "epoch": 1.03,
+      "learning_rate": 2.9899271656593833e-05,
+      "loss": 0.2733,
       "step": 2220
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 2.8514186701321204e-05,
-      "loss": 0.3966,
+      "epoch": 1.04,
+      "learning_rate": 2.98837749883775e-05,
+      "loss": 0.3845,
       "step": 2230
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 2.8492527615334633e-05,
-      "loss": 0.1387,
+      "epoch": 1.04,
+      "learning_rate": 2.9868278320161168e-05,
+      "loss": 0.3955,
       "step": 2240
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 2.847086852934806e-05,
-      "loss": 0.3419,
+      "epoch": 1.05,
+      "learning_rate": 2.9852781651944835e-05,
+      "loss": 0.7167,
       "step": 2250
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 2.8449209443361493e-05,
-      "loss": 0.3061,
+      "epoch": 1.05,
+      "learning_rate": 2.98372849837285e-05,
+      "loss": 0.3185,
       "step": 2260
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 2.8427550357374922e-05,
-      "loss": 0.6589,
+      "epoch": 1.06,
+      "learning_rate": 2.9821788315512167e-05,
+      "loss": 0.4002,
       "step": 2270
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 2.840589127138835e-05,
-      "loss": 0.2016,
+      "epoch": 1.06,
+      "learning_rate": 2.980629164729583e-05,
+      "loss": 0.3908,
       "step": 2280
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 2.8384232185401776e-05,
-      "loss": 0.2931,
+      "epoch": 1.06,
+      "learning_rate": 2.9790794979079498e-05,
+      "loss": 0.3851,
       "step": 2290
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 2.8362573099415205e-05,
-      "loss": 0.2773,
+      "epoch": 1.07,
+      "learning_rate": 2.9775298310863162e-05,
+      "loss": 0.6374,
       "step": 2300
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 2.8340914013428633e-05,
-      "loss": 0.2194,
+      "epoch": 1.07,
+      "learning_rate": 2.975980164264683e-05,
+      "loss": 0.3323,
       "step": 2310
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.8319254927442062e-05,
-      "loss": 0.2638,
+      "epoch": 1.08,
+      "learning_rate": 2.9744304974430497e-05,
+      "loss": 0.3535,
       "step": 2320
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.829759584145549e-05,
-      "loss": 0.3591,
+      "epoch": 1.08,
+      "learning_rate": 2.9728808306214165e-05,
+      "loss": 0.2928,
       "step": 2330
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.827593675546892e-05,
-      "loss": 0.2546,
+      "epoch": 1.09,
+      "learning_rate": 2.971331163799783e-05,
+      "loss": 0.3505,
       "step": 2340
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.825427766948235e-05,
-      "loss": 0.3646,
+      "epoch": 1.09,
+      "learning_rate": 2.9697814969781496e-05,
+      "loss": 0.4709,
       "step": 2350
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.8232618583495777e-05,
-      "loss": 0.458,
+      "epoch": 1.1,
+      "learning_rate": 2.9682318301565164e-05,
+      "loss": 0.4436,
       "step": 2360
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.8210959497509205e-05,
-      "loss": 0.313,
+      "epoch": 1.1,
+      "learning_rate": 2.966682163334883e-05,
+      "loss": 0.437,
       "step": 2370
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.8189300411522634e-05,
-      "loss": 0.3892,
+      "epoch": 1.11,
+      "learning_rate": 2.9651324965132495e-05,
+      "loss": 0.4205,
       "step": 2380
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.8167641325536063e-05,
-      "loss": 0.4741,
+      "epoch": 1.11,
+      "learning_rate": 2.9635828296916163e-05,
+      "loss": 0.5139,
       "step": 2390
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.814598223954949e-05,
-      "loss": 0.2889,
+      "epoch": 1.12,
+      "learning_rate": 2.962033162869983e-05,
+      "loss": 0.3445,
       "step": 2400
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.812432315356292e-05,
-      "loss": 0.5176,
+      "epoch": 1.12,
+      "learning_rate": 2.9604834960483498e-05,
+      "loss": 0.2945,
       "step": 2410
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.810266406757635e-05,
-      "loss": 0.2147,
+      "epoch": 1.13,
+      "learning_rate": 2.9589338292267162e-05,
+      "loss": 0.7276,
       "step": 2420
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.8081004981589778e-05,
-      "loss": 0.1541,
+      "epoch": 1.13,
+      "learning_rate": 2.957384162405083e-05,
+      "loss": 0.6739,
       "step": 2430
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.8059345895603206e-05,
-      "loss": 0.3722,
+      "epoch": 1.13,
+      "learning_rate": 2.9558344955834497e-05,
+      "loss": 0.4468,
       "step": 2440
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.8037686809616635e-05,
-      "loss": 0.1849,
+      "epoch": 1.14,
+      "learning_rate": 2.9542848287618164e-05,
+      "loss": 0.4889,
       "step": 2450
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.8016027723630064e-05,
-      "loss": 0.3595,
+      "epoch": 1.14,
+      "learning_rate": 2.952735161940183e-05,
+      "loss": 0.4727,
       "step": 2460
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.7994368637643492e-05,
-      "loss": 0.2309,
+      "epoch": 1.15,
+      "learning_rate": 2.9511854951185496e-05,
+      "loss": 0.5521,
       "step": 2470
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 2.797270955165692e-05,
-      "loss": 0.2493,
+      "epoch": 1.15,
+      "learning_rate": 2.9496358282969164e-05,
+      "loss": 0.2647,
       "step": 2480
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 2.795105046567035e-05,
-      "loss": 0.5766,
+      "epoch": 1.16,
+      "learning_rate": 2.948086161475283e-05,
+      "loss": 0.4498,
       "step": 2490
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 2.792939137968378e-05,
-      "loss": 0.4037,
+      "epoch": 1.16,
+      "learning_rate": 2.9465364946536495e-05,
+      "loss": 0.2865,
       "step": 2500
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 2.7907732293697207e-05,
-      "loss": 0.4681,
+      "epoch": 1.17,
+      "learning_rate": 2.9449868278320163e-05,
+      "loss": 0.281,
       "step": 2510
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 2.7886073207710632e-05,
-      "loss": 0.1419,
+      "epoch": 1.17,
+      "learning_rate": 2.943437161010383e-05,
+      "loss": 0.4023,
       "step": 2520
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 2.7864414121724064e-05,
-      "loss": 0.477,
+      "epoch": 1.18,
+      "learning_rate": 2.9418874941887494e-05,
+      "loss": 0.2375,
       "step": 2530
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 2.7842755035737493e-05,
-      "loss": 0.4341,
+      "epoch": 1.18,
+      "learning_rate": 2.9403378273671158e-05,
+      "loss": 0.4883,
       "step": 2540
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 2.7821095949750922e-05,
-      "loss": 0.2381,
+      "epoch": 1.19,
+      "learning_rate": 2.9387881605454826e-05,
+      "loss": 0.4597,
       "step": 2550
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 2.779943686376435e-05,
-      "loss": 0.302,
+      "epoch": 1.19,
+      "learning_rate": 2.9372384937238493e-05,
+      "loss": 0.4231,
       "step": 2560
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 2.777777777777778e-05,
-      "loss": 0.2982,
+      "epoch": 1.19,
+      "learning_rate": 2.935688826902216e-05,
+      "loss": 0.1726,
       "step": 2570
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 2.7756118691791208e-05,
-      "loss": 0.4892,
+      "epoch": 1.2,
+      "learning_rate": 2.9341391600805825e-05,
+      "loss": 0.3009,
       "step": 2580
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 2.7734459605804636e-05,
-      "loss": 0.3519,
+      "epoch": 1.2,
+      "learning_rate": 2.9325894932589492e-05,
+      "loss": 0.6538,
       "step": 2590
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 2.7712800519818062e-05,
-      "loss": 0.207,
+      "epoch": 1.21,
+      "learning_rate": 2.931039826437316e-05,
+      "loss": 0.4962,
       "step": 2600
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 2.769114143383149e-05,
-      "loss": 0.1381,
+      "epoch": 1.21,
+      "learning_rate": 2.9294901596156827e-05,
+      "loss": 0.2876,
       "step": 2610
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 2.7669482347844923e-05,
-      "loss": 0.3716,
+      "epoch": 1.22,
+      "learning_rate": 2.9279404927940495e-05,
+      "loss": 0.2462,
       "step": 2620
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 2.764782326185835e-05,
-      "loss": 0.4123,
+      "epoch": 1.22,
+      "learning_rate": 2.926390825972416e-05,
+      "loss": 0.4799,
       "step": 2630
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 2.762616417587178e-05,
-      "loss": 0.2551,
+      "epoch": 1.23,
+      "learning_rate": 2.9248411591507826e-05,
+      "loss": 0.4754,
       "step": 2640
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 2.760450508988521e-05,
-      "loss": 0.2666,
+      "epoch": 1.23,
+      "learning_rate": 2.9232914923291494e-05,
+      "loss": 0.2026,
       "step": 2650
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 2.7582846003898637e-05,
-      "loss": 0.1336,
+      "epoch": 1.24,
+      "learning_rate": 2.921741825507516e-05,
+      "loss": 0.1991,
       "step": 2660
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 2.7561186917912066e-05,
-      "loss": 0.5134,
+      "epoch": 1.24,
+      "learning_rate": 2.9201921586858825e-05,
+      "loss": 0.2708,
       "step": 2670
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 2.753952783192549e-05,
-      "loss": 0.2047,
+      "epoch": 1.25,
+      "learning_rate": 2.9186424918642493e-05,
+      "loss": 0.1851,
       "step": 2680
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 2.751786874593892e-05,
-      "loss": 0.2533,
+      "epoch": 1.25,
+      "learning_rate": 2.917092825042616e-05,
+      "loss": 0.4528,
       "step": 2690
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 2.749620965995235e-05,
-      "loss": 0.0704,
+      "epoch": 1.26,
+      "learning_rate": 2.9155431582209828e-05,
+      "loss": 0.4303,
       "step": 2700
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 2.747455057396578e-05,
-      "loss": 0.269,
+      "epoch": 1.26,
+      "learning_rate": 2.9139934913993492e-05,
+      "loss": 0.3352,
       "step": 2710
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 2.745289148797921e-05,
-      "loss": 0.1492,
+      "epoch": 1.26,
+      "learning_rate": 2.912443824577716e-05,
+      "loss": 0.3075,
       "step": 2720
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 2.7431232401992638e-05,
-      "loss": 0.1962,
+      "epoch": 1.27,
+      "learning_rate": 2.9108941577560827e-05,
+      "loss": 0.1506,
       "step": 2730
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 2.7409573316006067e-05,
-      "loss": 0.251,
+      "epoch": 1.27,
+      "learning_rate": 2.9093444909344494e-05,
+      "loss": 0.4589,
       "step": 2740
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 2.7387914230019495e-05,
-      "loss": 0.5038,
+      "epoch": 1.28,
+      "learning_rate": 2.907794824112816e-05,
+      "loss": 0.3709,
       "step": 2750
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 2.736625514403292e-05,
-      "loss": 0.0198,
+      "epoch": 1.28,
+      "learning_rate": 2.9062451572911826e-05,
+      "loss": 0.4069,
       "step": 2760
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 2.734459605804635e-05,
-      "loss": 0.3041,
+      "epoch": 1.29,
+      "learning_rate": 2.9046954904695493e-05,
+      "loss": 0.4776,
       "step": 2770
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 2.7322936972059778e-05,
-      "loss": 0.3146,
+      "epoch": 1.29,
+      "learning_rate": 2.9031458236479158e-05,
+      "loss": 0.5512,
       "step": 2780
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 2.7301277886073207e-05,
-      "loss": 0.2433,
+      "epoch": 1.3,
+      "learning_rate": 2.901596156826282e-05,
+      "loss": 0.307,
       "step": 2790
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 2.727961880008664e-05,
-      "loss": 0.5248,
+      "epoch": 1.3,
+      "learning_rate": 2.9002014566868122e-05,
+      "loss": 0.7173,
       "step": 2800
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 2.7257959714100068e-05,
-      "loss": 0.2415,
+      "epoch": 1.31,
+      "learning_rate": 2.898651789865179e-05,
+      "loss": 0.533,
       "step": 2810
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 2.7236300628113496e-05,
-      "loss": 0.3897,
+      "epoch": 1.31,
+      "learning_rate": 2.8971021230435457e-05,
+      "loss": 0.382,
       "step": 2820
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 2.7214641542126925e-05,
-      "loss": 0.4774,
+      "epoch": 1.32,
+      "learning_rate": 2.8955524562219125e-05,
+      "loss": 0.2306,
       "step": 2830
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 2.719298245614035e-05,
-      "loss": 0.1535,
+      "epoch": 1.32,
+      "learning_rate": 2.894002789400279e-05,
+      "loss": 0.5861,
       "step": 2840
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.717132337015378e-05,
-      "loss": 0.7375,
+      "epoch": 1.32,
+      "learning_rate": 2.8924531225786456e-05,
+      "loss": 0.4831,
       "step": 2850
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.7149664284167208e-05,
-      "loss": 0.3043,
+      "epoch": 1.33,
+      "learning_rate": 2.8909034557570124e-05,
+      "loss": 0.1982,
       "step": 2860
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.7128005198180636e-05,
-      "loss": 0.1723,
+      "epoch": 1.33,
+      "learning_rate": 2.889353788935379e-05,
+      "loss": 0.4137,
       "step": 2870
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.7106346112194065e-05,
-      "loss": 0.2247,
+      "epoch": 1.34,
+      "learning_rate": 2.887804122113746e-05,
+      "loss": 0.2809,
       "step": 2880
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 2.7084687026207497e-05,
-      "loss": 0.4143,
+      "epoch": 1.34,
+      "learning_rate": 2.8862544552921123e-05,
+      "loss": 0.4817,
       "step": 2890
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 2.7063027940220926e-05,
-      "loss": 0.3145,
+      "epoch": 1.35,
+      "learning_rate": 2.884704788470479e-05,
+      "loss": 0.3055,
       "step": 2900
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 2.7041368854234354e-05,
-      "loss": 0.2796,
+      "epoch": 1.35,
+      "learning_rate": 2.8831551216488458e-05,
+      "loss": 0.6572,
       "step": 2910
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 2.701970976824778e-05,
-      "loss": 0.1485,
+      "epoch": 1.36,
+      "learning_rate": 2.8816054548272125e-05,
+      "loss": 0.3695,
       "step": 2920
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 2.699805068226121e-05,
-      "loss": 0.3664,
+      "epoch": 1.36,
+      "learning_rate": 2.880055788005579e-05,
+      "loss": 0.1977,
       "step": 2930
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 2.6976391596274637e-05,
-      "loss": 0.298,
+      "epoch": 1.37,
+      "learning_rate": 2.8785061211839454e-05,
+      "loss": 0.2833,
       "step": 2940
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 2.6954732510288066e-05,
-      "loss": 0.5404,
+      "epoch": 1.37,
+      "learning_rate": 2.876956454362312e-05,
+      "loss": 0.2818,
       "step": 2950
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 2.6933073424301494e-05,
-      "loss": 0.0554,
+      "epoch": 1.38,
+      "learning_rate": 2.875406787540679e-05,
+      "loss": 0.4276,
       "step": 2960
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 2.6911414338314923e-05,
-      "loss": 0.3279,
+      "epoch": 1.38,
+      "learning_rate": 2.8738571207190453e-05,
+      "loss": 0.7242,
       "step": 2970
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 2.6889755252328355e-05,
-      "loss": 0.2289,
+      "epoch": 1.39,
+      "learning_rate": 2.872307453897412e-05,
+      "loss": 0.3649,
       "step": 2980
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 2.6868096166341784e-05,
-      "loss": 0.257,
+      "epoch": 1.39,
+      "learning_rate": 2.8707577870757788e-05,
+      "loss": 0.4253,
       "step": 2990
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 2.684643708035521e-05,
-      "loss": 0.5012,
+      "epoch": 1.39,
+      "learning_rate": 2.8692081202541455e-05,
+      "loss": 0.5493,
       "step": 3000
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.6824777994368638e-05,
-      "loss": 0.2418,
+      "epoch": 1.4,
+      "learning_rate": 2.867658453432512e-05,
+      "loss": 0.1422,
       "step": 3010
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.6803118908382067e-05,
-      "loss": 0.0826,
+      "epoch": 1.4,
+      "learning_rate": 2.8661087866108787e-05,
+      "loss": 0.2176,
       "step": 3020
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 2.6781459822395495e-05,
-      "loss": 0.204,
+      "epoch": 1.41,
+      "learning_rate": 2.8645591197892454e-05,
+      "loss": 0.2307,
       "step": 3030
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 2.6759800736408924e-05,
-      "loss": 0.2902,
+      "epoch": 1.41,
+      "learning_rate": 2.863009452967612e-05,
+      "loss": 0.3336,
       "step": 3040
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 2.6738141650422353e-05,
-      "loss": 0.3245,
+      "epoch": 1.42,
+      "learning_rate": 2.8614597861459786e-05,
+      "loss": 0.4291,
       "step": 3050
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 2.671648256443578e-05,
-      "loss": 0.3886,
+      "epoch": 1.42,
+      "learning_rate": 2.8599101193243453e-05,
+      "loss": 0.0654,
       "step": 3060
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 2.6694823478449213e-05,
-      "loss": 0.1232,
+      "epoch": 1.43,
+      "learning_rate": 2.858360452502712e-05,
+      "loss": 0.3041,
       "step": 3070
     },
     {
-      "epoch": 2.0,
-      "eval_accuracy": 0.9655013799448022,
-      "eval_loss": 0.1605590134859085,
-      "eval_runtime": 12.979,
-      "eval_samples_per_second": 167.501,
-      "eval_steps_per_second": 41.914,
-      "step": 3079
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 2.667316439246264e-05,
-      "loss": 0.1282,
+      "epoch": 1.43,
+      "learning_rate": 2.8568107856810788e-05,
+      "loss": 0.4646,
       "step": 3080
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 2.6651505306476067e-05,
-      "loss": 0.4082,
+      "epoch": 1.44,
+      "learning_rate": 2.8552611188594452e-05,
+      "loss": 0.399,
       "step": 3090
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 2.6629846220489496e-05,
-      "loss": 0.5166,
+      "epoch": 1.44,
+      "learning_rate": 2.853711452037812e-05,
+      "loss": 0.2304,
       "step": 3100
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 2.6608187134502925e-05,
-      "loss": 0.3364,
+      "epoch": 1.45,
+      "learning_rate": 2.8521617852161787e-05,
+      "loss": 0.3793,
       "step": 3110
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 2.6586528048516353e-05,
-      "loss": 0.2218,
+      "epoch": 1.45,
+      "learning_rate": 2.8506121183945455e-05,
+      "loss": 0.125,
       "step": 3120
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 2.6564868962529782e-05,
-      "loss": 0.285,
+      "epoch": 1.46,
+      "learning_rate": 2.849062451572912e-05,
+      "loss": 0.4181,
       "step": 3130
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 2.654320987654321e-05,
-      "loss": 0.214,
+      "epoch": 1.46,
+      "learning_rate": 2.8475127847512786e-05,
+      "loss": 0.37,
       "step": 3140
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 2.652155079055664e-05,
-      "loss": 0.547,
+      "epoch": 1.46,
+      "learning_rate": 2.8459631179296454e-05,
+      "loss": 0.501,
       "step": 3150
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 2.6499891704570065e-05,
-      "loss": 0.2743,
+      "epoch": 1.47,
+      "learning_rate": 2.844413451108012e-05,
+      "loss": 0.5434,
       "step": 3160
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 2.6478232618583497e-05,
-      "loss": 0.4205,
+      "epoch": 1.47,
+      "learning_rate": 2.8428637842863785e-05,
+      "loss": 0.4125,
       "step": 3170
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 2.6456573532596926e-05,
-      "loss": 0.1901,
+      "epoch": 1.48,
+      "learning_rate": 2.841314117464745e-05,
+      "loss": 0.4479,
       "step": 3180
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 2.6434914446610354e-05,
-      "loss": 0.1482,
+      "epoch": 1.48,
+      "learning_rate": 2.8397644506431117e-05,
+      "loss": 0.2616,
       "step": 3190
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 2.6413255360623783e-05,
-      "loss": 0.1439,
+      "epoch": 1.49,
+      "learning_rate": 2.8382147838214784e-05,
+      "loss": 0.3377,
       "step": 3200
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 2.639159627463721e-05,
-      "loss": 0.3271,
+      "epoch": 1.49,
+      "learning_rate": 2.836665116999845e-05,
+      "loss": 0.375,
       "step": 3210
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 2.636993718865064e-05,
-      "loss": 0.237,
+      "epoch": 1.5,
+      "learning_rate": 2.8351154501782116e-05,
+      "loss": 0.585,
       "step": 3220
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 2.634827810266407e-05,
-      "loss": 0.3626,
+      "epoch": 1.5,
+      "learning_rate": 2.8335657833565784e-05,
+      "loss": 0.5077,
       "step": 3230
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 2.6326619016677494e-05,
-      "loss": 0.2673,
+      "epoch": 1.51,
+      "learning_rate": 2.832016116534945e-05,
+      "loss": 0.2332,
       "step": 3240
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 2.6304959930690923e-05,
-      "loss": 0.0236,
+      "epoch": 1.51,
+      "learning_rate": 2.8304664497133115e-05,
+      "loss": 0.4414,
       "step": 3250
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 2.6283300844704355e-05,
-      "loss": 0.3381,
+      "epoch": 1.52,
+      "learning_rate": 2.8289167828916783e-05,
+      "loss": 0.2384,
       "step": 3260
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 2.6261641758717784e-05,
-      "loss": 0.0661,
+      "epoch": 1.52,
+      "learning_rate": 2.827367116070045e-05,
+      "loss": 0.3747,
       "step": 3270
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 2.6239982672731212e-05,
-      "loss": 0.4211,
+      "epoch": 1.52,
+      "learning_rate": 2.8258174492484118e-05,
+      "loss": 0.2938,
       "step": 3280
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 2.621832358674464e-05,
-      "loss": 0.3107,
+      "epoch": 1.53,
+      "learning_rate": 2.824267782426778e-05,
+      "loss": 0.3634,
       "step": 3290
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 2.619666450075807e-05,
-      "loss": 0.3576,
+      "epoch": 1.53,
+      "learning_rate": 2.822718115605145e-05,
+      "loss": 0.2502,
       "step": 3300
     },
     {
-      "epoch": 2.15,
-      "learning_rate": 2.61750054147715e-05,
-      "loss": 0.4687,
+      "epoch": 1.54,
+      "learning_rate": 2.8211684487835117e-05,
+      "loss": 0.3573,
       "step": 3310
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 2.6153346328784924e-05,
-      "loss": 0.1991,
+      "epoch": 1.54,
+      "learning_rate": 2.8196187819618784e-05,
+      "loss": 0.3062,
       "step": 3320
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 2.6131687242798352e-05,
-      "loss": 0.5119,
+      "epoch": 1.55,
+      "learning_rate": 2.8180691151402448e-05,
+      "loss": 0.3654,
       "step": 3330
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 2.611002815681178e-05,
-      "loss": 0.1439,
+      "epoch": 1.55,
+      "learning_rate": 2.8165194483186116e-05,
+      "loss": 0.2654,
       "step": 3340
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 2.6088369070825213e-05,
-      "loss": 0.29,
+      "epoch": 1.56,
+      "learning_rate": 2.8149697814969783e-05,
+      "loss": 0.1328,
       "step": 3350
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 2.6066709984838642e-05,
-      "loss": 0.1705,
+      "epoch": 1.56,
+      "learning_rate": 2.813420114675345e-05,
+      "loss": 0.4447,
       "step": 3360
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 2.604505089885207e-05,
-      "loss": 0.1561,
+      "epoch": 1.57,
+      "learning_rate": 2.8118704478537115e-05,
+      "loss": 0.3217,
       "step": 3370
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 2.60233918128655e-05,
-      "loss": 0.5298,
+      "epoch": 1.57,
+      "learning_rate": 2.8103207810320782e-05,
+      "loss": 0.1941,
       "step": 3380
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 2.6001732726878928e-05,
-      "loss": 0.2279,
+      "epoch": 1.58,
+      "learning_rate": 2.808771114210445e-05,
+      "loss": 0.5132,
       "step": 3390
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 2.5980073640892353e-05,
-      "loss": 0.1045,
+      "epoch": 1.58,
+      "learning_rate": 2.8072214473888117e-05,
+      "loss": 0.4109,
       "step": 3400
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 2.5958414554905782e-05,
-      "loss": 0.2926,
+      "epoch": 1.59,
+      "learning_rate": 2.805671780567178e-05,
+      "loss": 0.503,
       "step": 3410
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 2.593675546891921e-05,
-      "loss": 0.1234,
-      "step": 3420
+      "epoch": 1.59,
+      "learning_rate": 2.804122113745545e-05,
+      "loss": 0.259,
+      "step": 3420
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 2.591509638293264e-05,
-      "loss": 0.4732,
+      "epoch": 1.59,
+      "learning_rate": 2.8025724469239113e-05,
+      "loss": 0.1918,
       "step": 3430
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 2.589343729694607e-05,
-      "loss": 0.2754,
+      "epoch": 1.6,
+      "learning_rate": 2.801022780102278e-05,
+      "loss": 0.3181,
       "step": 3440
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 2.58717782109595e-05,
-      "loss": 0.3542,
+      "epoch": 1.6,
+      "learning_rate": 2.7994731132806444e-05,
+      "loss": 0.1778,
       "step": 3450
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 2.585011912497293e-05,
-      "loss": 0.5329,
+      "epoch": 1.61,
+      "learning_rate": 2.7979234464590112e-05,
+      "loss": 0.354,
       "step": 3460
     },
     {
-      "epoch": 2.25,
-      "learning_rate": 2.5828460038986354e-05,
-      "loss": 0.2176,
+      "epoch": 1.61,
+      "learning_rate": 2.796373779637378e-05,
+      "loss": 0.718,
       "step": 3470
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 2.5806800952999783e-05,
-      "loss": 0.1245,
+      "epoch": 1.62,
+      "learning_rate": 2.7948241128157447e-05,
+      "loss": 0.2096,
       "step": 3480
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 2.578514186701321e-05,
-      "loss": 0.1505,
+      "epoch": 1.62,
+      "learning_rate": 2.793274445994111e-05,
+      "loss": 0.5214,
       "step": 3490
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 2.576348278102664e-05,
-      "loss": 0.5114,
+      "epoch": 1.63,
+      "learning_rate": 2.791724779172478e-05,
+      "loss": 0.2787,
       "step": 3500
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 2.574182369504007e-05,
-      "loss": 0.2739,
+      "epoch": 1.63,
+      "learning_rate": 2.7901751123508446e-05,
+      "loss": 0.2631,
       "step": 3510
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 2.5720164609053497e-05,
-      "loss": 0.4071,
+      "epoch": 1.64,
+      "learning_rate": 2.7886254455292113e-05,
+      "loss": 0.3894,
       "step": 3520
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 2.569850552306693e-05,
-      "loss": 0.1159,
+      "epoch": 1.64,
+      "learning_rate": 2.7870757787075778e-05,
+      "loss": 0.4931,
       "step": 3530
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 2.5676846437080358e-05,
-      "loss": 0.3694,
+      "epoch": 1.65,
+      "learning_rate": 2.7855261118859445e-05,
+      "loss": 0.2916,
       "step": 3540
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 2.5655187351093783e-05,
-      "loss": 0.4572,
+      "epoch": 1.65,
+      "learning_rate": 2.7839764450643113e-05,
+      "loss": 0.3037,
       "step": 3550
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 2.5633528265107212e-05,
-      "loss": 0.2517,
+      "epoch": 1.66,
+      "learning_rate": 2.782426778242678e-05,
+      "loss": 0.3069,
       "step": 3560
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 2.561186917912064e-05,
-      "loss": 0.1947,
+      "epoch": 1.66,
+      "learning_rate": 2.7808771114210444e-05,
+      "loss": 0.2568,
       "step": 3570
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 2.559021009313407e-05,
-      "loss": 0.2947,
+      "epoch": 1.66,
+      "learning_rate": 2.779327444599411e-05,
+      "loss": 0.4131,
       "step": 3580
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 2.5568551007147498e-05,
-      "loss": 0.1922,
+      "epoch": 1.67,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.2753,
       "step": 3590
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 2.5546891921160927e-05,
-      "loss": 0.3169,
+      "epoch": 1.67,
+      "learning_rate": 2.7762281109561447e-05,
+      "loss": 0.5791,
       "step": 3600
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 2.5525232835174356e-05,
-      "loss": 0.065,
+      "epoch": 1.68,
+      "learning_rate": 2.774678444134511e-05,
+      "loss": 0.2685,
       "step": 3610
     },
     {
-      "epoch": 2.35,
-      "learning_rate": 2.5503573749187788e-05,
-      "loss": 0.4879,
+      "epoch": 1.68,
+      "learning_rate": 2.7731287773128778e-05,
+      "loss": 0.2439,
       "step": 3620
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 2.5481914663201213e-05,
-      "loss": 0.1695,
+      "epoch": 1.69,
+      "learning_rate": 2.7715791104912446e-05,
+      "loss": 0.6156,
       "step": 3630
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 2.546025557721464e-05,
-      "loss": 0.381,
+      "epoch": 1.69,
+      "learning_rate": 2.7700294436696113e-05,
+      "loss": 0.4206,
       "step": 3640
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 2.543859649122807e-05,
-      "loss": 0.2645,
+      "epoch": 1.7,
+      "learning_rate": 2.7684797768479777e-05,
+      "loss": 0.2462,
       "step": 3650
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 2.54169374052415e-05,
-      "loss": 0.2559,
+      "epoch": 1.7,
+      "learning_rate": 2.7669301100263445e-05,
+      "loss": 0.3283,
       "step": 3660
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 2.5395278319254928e-05,
-      "loss": 0.0956,
+      "epoch": 1.71,
+      "learning_rate": 2.7653804432047112e-05,
+      "loss": 0.4734,
       "step": 3670
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 2.5373619233268356e-05,
-      "loss": 0.3473,
+      "epoch": 1.71,
+      "learning_rate": 2.7638307763830776e-05,
+      "loss": 0.3934,
       "step": 3680
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 2.5351960147281785e-05,
-      "loss": 0.25,
+      "epoch": 1.72,
+      "learning_rate": 2.762281109561444e-05,
+      "loss": 0.2597,
       "step": 3690
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 2.5330301061295214e-05,
-      "loss": 0.3441,
+      "epoch": 1.72,
+      "learning_rate": 2.7607314427398108e-05,
+      "loss": 0.482,
       "step": 3700
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 2.5308641975308642e-05,
-      "loss": 0.2912,
+      "epoch": 1.72,
+      "learning_rate": 2.7591817759181775e-05,
+      "loss": 0.1531,
       "step": 3710
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 2.528698288932207e-05,
-      "loss": 0.2895,
+      "epoch": 1.73,
+      "learning_rate": 2.7576321090965443e-05,
+      "loss": 0.3001,
       "step": 3720
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 2.52653238033355e-05,
-      "loss": 0.1336,
+      "epoch": 1.73,
+      "learning_rate": 2.756082442274911e-05,
+      "loss": 0.3956,
       "step": 3730
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 2.524366471734893e-05,
-      "loss": 0.1,
+      "epoch": 1.74,
+      "learning_rate": 2.7545327754532774e-05,
+      "loss": 0.1822,
       "step": 3740
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 2.5222005631362357e-05,
-      "loss": 0.0469,
+      "epoch": 1.74,
+      "learning_rate": 2.7529831086316442e-05,
+      "loss": 0.5104,
       "step": 3750
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 2.5200346545375786e-05,
-      "loss": 0.1433,
+      "epoch": 1.75,
+      "learning_rate": 2.751433441810011e-05,
+      "loss": 0.4782,
       "step": 3760
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 2.5178687459389215e-05,
-      "loss": 0.1602,
+      "epoch": 1.75,
+      "learning_rate": 2.7498837749883777e-05,
+      "loss": 0.3576,
       "step": 3770
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 2.5157028373402643e-05,
-      "loss": 0.3086,
+      "epoch": 1.76,
+      "learning_rate": 2.748334108166744e-05,
+      "loss": 0.3221,
       "step": 3780
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 2.513536928741607e-05,
-      "loss": 0.112,
+      "epoch": 1.76,
+      "learning_rate": 2.746784441345111e-05,
+      "loss": 0.3456,
       "step": 3790
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 2.51137102014295e-05,
-      "loss": 0.3675,
+      "epoch": 1.77,
+      "learning_rate": 2.7452347745234776e-05,
+      "loss": 0.501,
       "step": 3800
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 2.509205111544293e-05,
-      "loss": 0.1579,
+      "epoch": 1.77,
+      "learning_rate": 2.7436851077018443e-05,
+      "loss": 0.2229,
       "step": 3810
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 2.5070392029456358e-05,
-      "loss": 0.3057,
+      "epoch": 1.78,
+      "learning_rate": 2.7421354408802108e-05,
+      "loss": 0.2876,
       "step": 3820
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 2.5048732943469787e-05,
-      "loss": 0.3646,
+      "epoch": 1.78,
+      "learning_rate": 2.7405857740585775e-05,
+      "loss": 0.3996,
       "step": 3830
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 2.5027073857483215e-05,
-      "loss": 0.2384,
+      "epoch": 1.79,
+      "learning_rate": 2.7390361072369443e-05,
+      "loss": 0.3382,
       "step": 3840
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 2.5005414771496644e-05,
-      "loss": 0.3239,
+      "epoch": 1.79,
+      "learning_rate": 2.737486440415311e-05,
+      "loss": 0.2938,
       "step": 3850
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 2.4983755685510073e-05,
-      "loss": 0.2883,
+      "epoch": 1.79,
+      "learning_rate": 2.7359367735936774e-05,
+      "loss": 0.2648,
       "step": 3860
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 2.4962096599523498e-05,
-      "loss": 0.2288,
+      "epoch": 1.8,
+      "learning_rate": 2.734387106772044e-05,
+      "loss": 0.2285,
       "step": 3870
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 2.4940437513536927e-05,
-      "loss": 0.3717,
+      "epoch": 1.8,
+      "learning_rate": 2.732837439950411e-05,
+      "loss": 0.3115,
       "step": 3880
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 2.491877842755036e-05,
-      "loss": 0.2852,
+      "epoch": 1.81,
+      "learning_rate": 2.7312877731287777e-05,
+      "loss": 0.5288,
       "step": 3890
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 2.4897119341563787e-05,
-      "loss": 0.6269,
+      "epoch": 1.81,
+      "learning_rate": 2.729738106307144e-05,
+      "loss": 0.1673,
       "step": 3900
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 2.4875460255577216e-05,
-      "loss": 0.2522,
+      "epoch": 1.82,
+      "learning_rate": 2.7281884394855108e-05,
+      "loss": 0.3358,
       "step": 3910
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 2.4853801169590645e-05,
-      "loss": 0.3426,
+      "epoch": 1.82,
+      "learning_rate": 2.7266387726638776e-05,
+      "loss": 0.5848,
       "step": 3920
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 2.4832142083604074e-05,
-      "loss": 0.2613,
+      "epoch": 1.83,
+      "learning_rate": 2.725089105842244e-05,
+      "loss": 0.3806,
       "step": 3930
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 2.4810482997617502e-05,
-      "loss": 0.3396,
+      "epoch": 1.83,
+      "learning_rate": 2.7235394390206104e-05,
+      "loss": 0.2968,
       "step": 3940
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 2.4788823911630927e-05,
-      "loss": 0.225,
+      "epoch": 1.84,
+      "learning_rate": 2.721989772198977e-05,
+      "loss": 0.2995,
       "step": 3950
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 2.4767164825644356e-05,
-      "loss": 0.204,
+      "epoch": 1.84,
+      "learning_rate": 2.720440105377344e-05,
+      "loss": 0.3624,
       "step": 3960
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 2.4745505739657785e-05,
-      "loss": 0.6812,
+      "epoch": 1.85,
+      "learning_rate": 2.7188904385557106e-05,
+      "loss": 0.2398,
       "step": 3970
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 2.4723846653671217e-05,
-      "loss": 0.5684,
+      "epoch": 1.85,
+      "learning_rate": 2.717340771734077e-05,
+      "loss": 0.3253,
       "step": 3980
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 2.4702187567684646e-05,
-      "loss": 0.1882,
+      "epoch": 1.85,
+      "learning_rate": 2.7157911049124438e-05,
+      "loss": 0.2992,
       "step": 3990
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 2.4680528481698074e-05,
-      "loss": 0.2217,
+      "epoch": 1.86,
+      "learning_rate": 2.7142414380908105e-05,
+      "loss": 0.5173,
       "step": 4000
     },
     {
-      "epoch": 2.6,
-      "learning_rate": 2.4658869395711503e-05,
-      "loss": 0.0704,
+      "epoch": 1.86,
+      "learning_rate": 2.7126917712691773e-05,
+      "loss": 0.2939,
       "step": 4010
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 2.463721030972493e-05,
-      "loss": 0.3599,
+      "epoch": 1.87,
+      "learning_rate": 2.7111421044475437e-05,
+      "loss": 0.519,
       "step": 4020
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 2.4615551223738357e-05,
-      "loss": 0.2779,
+      "epoch": 1.87,
+      "learning_rate": 2.7095924376259104e-05,
+      "loss": 0.4549,
       "step": 4030
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 2.4593892137751786e-05,
-      "loss": 0.4508,
+      "epoch": 1.88,
+      "learning_rate": 2.7080427708042772e-05,
+      "loss": 0.2915,
       "step": 4040
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 2.4572233051765214e-05,
-      "loss": 0.3388,
+      "epoch": 1.88,
+      "learning_rate": 2.706493103982644e-05,
+      "loss": 0.4378,
       "step": 4050
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 2.4550573965778643e-05,
-      "loss": 0.3605,
+      "epoch": 1.89,
+      "learning_rate": 2.7049434371610103e-05,
+      "loss": 0.3109,
       "step": 4060
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 2.4528914879792075e-05,
-      "loss": 0.3589,
+      "epoch": 1.89,
+      "learning_rate": 2.703393770339377e-05,
+      "loss": 0.5096,
       "step": 4070
     },
     {
-      "epoch": 2.65,
-      "learning_rate": 2.4507255793805504e-05,
-      "loss": 0.2711,
+      "epoch": 1.9,
+      "learning_rate": 2.701844103517744e-05,
+      "loss": 0.3356,
       "step": 4080
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 2.4485596707818932e-05,
-      "loss": 0.1015,
+      "epoch": 1.9,
+      "learning_rate": 2.7002944366961106e-05,
+      "loss": 0.3959,
       "step": 4090
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 2.446393762183236e-05,
-      "loss": 0.4934,
+      "epoch": 1.91,
+      "learning_rate": 2.698744769874477e-05,
+      "loss": 0.499,
       "step": 4100
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 2.4442278535845786e-05,
-      "loss": 0.5104,
+      "epoch": 1.91,
+      "learning_rate": 2.6971951030528437e-05,
+      "loss": 0.3222,
       "step": 4110
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 2.4420619449859215e-05,
-      "loss": 0.2386,
+      "epoch": 1.92,
+      "learning_rate": 2.6956454362312105e-05,
+      "loss": 0.3265,
       "step": 4120
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 2.4398960363872644e-05,
-      "loss": 0.3044,
+      "epoch": 1.92,
+      "learning_rate": 2.6940957694095772e-05,
+      "loss": 0.2492,
       "step": 4130
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 2.4377301277886073e-05,
-      "loss": 0.4077,
+      "epoch": 1.92,
+      "learning_rate": 2.6925461025879437e-05,
+      "loss": 0.6622,
       "step": 4140
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 2.43556421918995e-05,
-      "loss": 0.2545,
+      "epoch": 1.93,
+      "learning_rate": 2.6909964357663104e-05,
+      "loss": 0.2932,
       "step": 4150
     },
     {
-      "epoch": 2.7,
-      "learning_rate": 2.4333983105912933e-05,
-      "loss": 0.2894,
+      "epoch": 1.93,
+      "learning_rate": 2.689446768944677e-05,
+      "loss": 0.8129,
       "step": 4160
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 2.4312324019926362e-05,
-      "loss": 0.3408,
+      "epoch": 1.94,
+      "learning_rate": 2.687897102123044e-05,
+      "loss": 0.1074,
       "step": 4170
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 2.429066493393979e-05,
-      "loss": 0.1438,
+      "epoch": 1.94,
+      "learning_rate": 2.68634743530141e-05,
+      "loss": 0.1133,
       "step": 4180
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 2.4269005847953216e-05,
-      "loss": 0.2897,
+      "epoch": 1.95,
+      "learning_rate": 2.6847977684797767e-05,
+      "loss": 0.4516,
       "step": 4190
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 2.4247346761966645e-05,
-      "loss": 0.2148,
+      "epoch": 1.95,
+      "learning_rate": 2.6832481016581435e-05,
+      "loss": 0.3812,
       "step": 4200
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 2.4225687675980073e-05,
-      "loss": 0.5298,
+      "epoch": 1.96,
+      "learning_rate": 2.6816984348365102e-05,
+      "loss": 0.2656,
       "step": 4210
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 2.4204028589993502e-05,
-      "loss": 0.1486,
+      "epoch": 1.96,
+      "learning_rate": 2.6801487680148766e-05,
+      "loss": 0.1742,
       "step": 4220
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 2.418236950400693e-05,
-      "loss": 0.4059,
+      "epoch": 1.97,
+      "learning_rate": 2.6785991011932434e-05,
+      "loss": 0.3687,
       "step": 4230
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 2.416071041802036e-05,
-      "loss": 0.1608,
+      "epoch": 1.97,
+      "learning_rate": 2.67704943437161e-05,
+      "loss": 0.3627,
       "step": 4240
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 2.413905133203379e-05,
-      "loss": 0.2226,
+      "epoch": 1.98,
+      "learning_rate": 2.675499767549977e-05,
+      "loss": 0.0979,
       "step": 4250
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 2.411739224604722e-05,
-      "loss": 0.3627,
+      "epoch": 1.98,
+      "learning_rate": 2.6739501007283433e-05,
+      "loss": 0.2193,
       "step": 4260
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 2.4095733160060645e-05,
-      "loss": 0.3048,
+      "epoch": 1.99,
+      "learning_rate": 2.67240043390671e-05,
+      "loss": 0.1529,
       "step": 4270
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 2.4074074074074074e-05,
-      "loss": 0.2084,
+      "epoch": 1.99,
+      "learning_rate": 2.6708507670850768e-05,
+      "loss": 0.4351,
       "step": 4280
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 2.4052414988087503e-05,
-      "loss": 0.128,
+      "epoch": 1.99,
+      "learning_rate": 2.6693011002634435e-05,
+      "loss": 0.3699,
       "step": 4290
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 2.403075590210093e-05,
-      "loss": 0.4311,
+      "epoch": 2.0,
+      "learning_rate": 2.66775143344181e-05,
+      "loss": 0.4323,
       "step": 4300
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 2.400909681611436e-05,
-      "loss": 0.2711,
+      "epoch": 2.0,
+      "eval_accuracy": 0.9562067830095489,
+      "eval_loss": 0.2042698711156845,
+      "eval_runtime": 17.0558,
+      "eval_samples_per_second": 178.063,
+      "eval_steps_per_second": 44.56,
+      "step": 4302
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.6662017666201767e-05,
+      "loss": 0.4572,
       "step": 4310
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 2.398743773012779e-05,
-      "loss": 0.2327,
+      "epoch": 2.01,
+      "learning_rate": 2.6646520997985434e-05,
+      "loss": 0.2285,
       "step": 4320
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 2.3965778644141218e-05,
-      "loss": 0.467,
+      "epoch": 2.01,
+      "learning_rate": 2.6631024329769102e-05,
+      "loss": 0.1455,
       "step": 4330
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 2.394411955815465e-05,
-      "loss": 0.5556,
+      "epoch": 2.02,
+      "learning_rate": 2.6615527661552766e-05,
+      "loss": 0.2543,
       "step": 4340
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 2.3922460472168075e-05,
-      "loss": 0.2591,
+      "epoch": 2.02,
+      "learning_rate": 2.6600030993336433e-05,
+      "loss": 0.2346,
       "step": 4350
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 2.3900801386181504e-05,
-      "loss": 0.2438,
+      "epoch": 2.03,
+      "learning_rate": 2.65845343251201e-05,
+      "loss": 0.1568,
       "step": 4360
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 2.3879142300194932e-05,
-      "loss": 0.2504,
+      "epoch": 2.03,
+      "learning_rate": 2.656903765690377e-05,
+      "loss": 0.3764,
       "step": 4370
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 2.385748321420836e-05,
-      "loss": 0.0638,
+      "epoch": 2.04,
+      "learning_rate": 2.6553540988687432e-05,
+      "loss": 0.1831,
       "step": 4380
     },
     {
-      "epoch": 2.85,
-      "learning_rate": 2.383582412822179e-05,
-      "loss": 0.2644,
+      "epoch": 2.04,
+      "learning_rate": 2.65380443204711e-05,
+      "loss": 0.8255,
       "step": 4390
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 2.381416504223522e-05,
-      "loss": 0.1309,
+      "epoch": 2.05,
+      "learning_rate": 2.6522547652254767e-05,
+      "loss": 0.339,
       "step": 4400
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 2.3792505956248647e-05,
-      "loss": 0.2903,
+      "epoch": 2.05,
+      "learning_rate": 2.6507050984038435e-05,
+      "loss": 0.3125,
       "step": 4410
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 2.3770846870262076e-05,
-      "loss": 0.3967,
+      "epoch": 2.05,
+      "learning_rate": 2.64915543158221e-05,
+      "loss": 0.2821,
       "step": 4420
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 2.3749187784275504e-05,
-      "loss": 0.1492,
+      "epoch": 2.06,
+      "learning_rate": 2.6476057647605763e-05,
+      "loss": 0.5594,
       "step": 4430
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 2.3727528698288933e-05,
-      "loss": 0.2094,
+      "epoch": 2.06,
+      "learning_rate": 2.646056097938943e-05,
+      "loss": 0.2578,
       "step": 4440
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 2.3705869612302362e-05,
-      "loss": 0.1203,
+      "epoch": 2.07,
+      "learning_rate": 2.6445064311173098e-05,
+      "loss": 0.2428,
       "step": 4450
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 2.368421052631579e-05,
-      "loss": 0.2751,
+      "epoch": 2.07,
+      "learning_rate": 2.6429567642956762e-05,
+      "loss": 0.2537,
       "step": 4460
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 2.366255144032922e-05,
-      "loss": 0.3923,
+      "epoch": 2.08,
+      "learning_rate": 2.641407097474043e-05,
+      "loss": 0.2695,
       "step": 4470
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 2.3640892354342648e-05,
-      "loss": 0.3586,
+      "epoch": 2.08,
+      "learning_rate": 2.6398574306524097e-05,
+      "loss": 0.3818,
       "step": 4480
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 2.3619233268356076e-05,
-      "loss": 0.1712,
+      "epoch": 2.09,
+      "learning_rate": 2.6383077638307765e-05,
+      "loss": 0.418,
       "step": 4490
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 2.3597574182369505e-05,
-      "loss": 0.2187,
+      "epoch": 2.09,
+      "learning_rate": 2.636758097009143e-05,
+      "loss": 0.5186,
       "step": 4500
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 2.357591509638293e-05,
-      "loss": 0.2134,
+      "epoch": 2.1,
+      "learning_rate": 2.6352084301875096e-05,
+      "loss": 0.1777,
       "step": 4510
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 2.3554256010396363e-05,
-      "loss": 0.1906,
+      "epoch": 2.1,
+      "learning_rate": 2.6336587633658764e-05,
+      "loss": 0.3769,
       "step": 4520
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 2.353259692440979e-05,
-      "loss": 0.2102,
+      "epoch": 2.11,
+      "learning_rate": 2.632109096544243e-05,
+      "loss": 0.4497,
       "step": 4530
     },
     {
-      "epoch": 2.95,
-      "learning_rate": 2.351093783842322e-05,
-      "loss": 0.1499,
+      "epoch": 2.11,
+      "learning_rate": 2.6305594297226095e-05,
+      "loss": 0.3411,
       "step": 4540
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 2.348927875243665e-05,
-      "loss": 0.1415,
+      "epoch": 2.12,
+      "learning_rate": 2.6290097629009763e-05,
+      "loss": 0.4039,
       "step": 4550
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 2.3467619666450077e-05,
-      "loss": 0.146,
+      "epoch": 2.12,
+      "learning_rate": 2.627460096079343e-05,
+      "loss": 0.424,
       "step": 4560
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 2.3445960580463506e-05,
-      "loss": 0.4851,
+      "epoch": 2.12,
+      "learning_rate": 2.6259104292577098e-05,
+      "loss": 0.1011,
       "step": 4570
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 2.342430149447693e-05,
-      "loss": 0.1137,
+      "epoch": 2.13,
+      "learning_rate": 2.6243607624360762e-05,
+      "loss": 0.1579,
       "step": 4580
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 2.340264240849036e-05,
-      "loss": 0.474,
+      "epoch": 2.13,
+      "learning_rate": 2.622811095614443e-05,
+      "loss": 0.4541,
       "step": 4590
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 2.338098332250379e-05,
-      "loss": 0.4388,
+      "epoch": 2.14,
+      "learning_rate": 2.6212614287928097e-05,
+      "loss": 0.4715,
       "step": 4600
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 2.335932423651722e-05,
-      "loss": 0.2855,
+      "epoch": 2.14,
+      "learning_rate": 2.6197117619711764e-05,
+      "loss": 0.2024,
       "step": 4610
     },
     {
-      "epoch": 3.0,
-      "eval_accuracy": 0.9696412143514259,
-      "eval_loss": 0.11773567646741867,
-      "eval_runtime": 12.9815,
-      "eval_samples_per_second": 167.469,
-      "eval_steps_per_second": 41.906,
-      "step": 4618
-    },
-    {
-      "epoch": 3.0,
-      "learning_rate": 2.333766515053065e-05,
-      "loss": 0.333,
+      "epoch": 2.15,
+      "learning_rate": 2.6181620951495432e-05,
+      "loss": 0.1076,
       "step": 4620
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 2.3316006064544078e-05,
-      "loss": 0.1692,
+      "epoch": 2.15,
+      "learning_rate": 2.6166124283279096e-05,
+      "loss": 0.284,
       "step": 4630
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 2.3294346978557507e-05,
-      "loss": 0.1158,
+      "epoch": 2.16,
+      "learning_rate": 2.6150627615062763e-05,
+      "loss": 0.2174,
       "step": 4640
     },
     {
-      "epoch": 3.02,
-      "learning_rate": 2.3272687892570935e-05,
-      "loss": 0.0129,
+      "epoch": 2.16,
+      "learning_rate": 2.613513094684643e-05,
+      "loss": 0.2124,
       "step": 4650
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 2.325102880658436e-05,
-      "loss": 0.2642,
+      "epoch": 2.17,
+      "learning_rate": 2.6119634278630098e-05,
+      "loss": 0.3909,
       "step": 4660
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 2.322936972059779e-05,
-      "loss": 0.5089,
+      "epoch": 2.17,
+      "learning_rate": 2.6104137610413762e-05,
+      "loss": 0.2164,
       "step": 4670
     },
     {
-      "epoch": 3.04,
-      "learning_rate": 2.3207710634611218e-05,
-      "loss": 0.0762,
+      "epoch": 2.18,
+      "learning_rate": 2.6088640942197426e-05,
+      "loss": 0.1218,
       "step": 4680
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 2.3186051548624647e-05,
-      "loss": 0.236,
+      "epoch": 2.18,
+      "learning_rate": 2.6073144273981094e-05,
+      "loss": 0.1378,
       "step": 4690
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 2.316439246263808e-05,
-      "loss": 0.263,
+      "epoch": 2.19,
+      "learning_rate": 2.605764760576476e-05,
+      "loss": 0.3855,
       "step": 4700
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 2.3142733376651508e-05,
-      "loss": 0.0997,
+      "epoch": 2.19,
+      "learning_rate": 2.6042150937548426e-05,
+      "loss": 0.0628,
       "step": 4710
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 2.3121074290664936e-05,
-      "loss": 0.3237,
+      "epoch": 2.19,
+      "learning_rate": 2.6026654269332093e-05,
+      "loss": 0.2678,
       "step": 4720
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 2.3099415204678365e-05,
-      "loss": 0.0913,
+      "epoch": 2.2,
+      "learning_rate": 2.601115760111576e-05,
+      "loss": 0.1977,
       "step": 4730
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 2.307775611869179e-05,
-      "loss": 0.154,
+      "epoch": 2.2,
+      "learning_rate": 2.5995660932899428e-05,
+      "loss": 0.1521,
       "step": 4740
     },
     {
-      "epoch": 3.09,
-      "learning_rate": 2.305609703270522e-05,
-      "loss": 0.1299,
+      "epoch": 2.21,
+      "learning_rate": 2.5980164264683092e-05,
+      "loss": 0.4095,
       "step": 4750
     },
     {
-      "epoch": 3.09,
-      "learning_rate": 2.3034437946718648e-05,
-      "loss": 0.2991,
+      "epoch": 2.21,
+      "learning_rate": 2.596466759646676e-05,
+      "loss": 0.2332,
       "step": 4760
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 2.3012778860732076e-05,
-      "loss": 0.2063,
+      "epoch": 2.22,
+      "learning_rate": 2.5949170928250427e-05,
+      "loss": 0.3119,
       "step": 4770
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 2.2991119774745505e-05,
-      "loss": 0.4323,
+      "epoch": 2.22,
+      "learning_rate": 2.5933674260034095e-05,
+      "loss": 0.267,
       "step": 4780
     },
     {
-      "epoch": 3.11,
-      "learning_rate": 2.2969460688758937e-05,
-      "loss": 0.3903,
+      "epoch": 2.23,
+      "learning_rate": 2.591817759181776e-05,
+      "loss": 0.1504,
       "step": 4790
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 2.2947801602772366e-05,
-      "loss": 0.2083,
+      "epoch": 2.23,
+      "learning_rate": 2.5902680923601426e-05,
+      "loss": 0.208,
       "step": 4800
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 2.2926142516785794e-05,
-      "loss": 0.2063,
+      "epoch": 2.24,
+      "learning_rate": 2.5887184255385094e-05,
+      "loss": 0.4964,
       "step": 4810
     },
     {
-      "epoch": 3.13,
-      "learning_rate": 2.290448343079922e-05,
-      "loss": 0.4494,
+      "epoch": 2.24,
+      "learning_rate": 2.587168758716876e-05,
+      "loss": 0.369,
       "step": 4820
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 2.288282434481265e-05,
-      "loss": 0.5917,
+      "epoch": 2.25,
+      "learning_rate": 2.5856190918952425e-05,
+      "loss": 0.347,
       "step": 4830
     },
     {
-      "epoch": 3.14,
-      "learning_rate": 2.2861165258826077e-05,
-      "loss": 0.175,
+      "epoch": 2.25,
+      "learning_rate": 2.5840694250736093e-05,
+      "loss": 0.3967,
       "step": 4840
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 2.2839506172839506e-05,
-      "loss": 0.1678,
+      "epoch": 2.25,
+      "learning_rate": 2.582519758251976e-05,
+      "loss": 0.2601,
       "step": 4850
     },
     {
-      "epoch": 3.16,
-      "learning_rate": 2.2817847086852934e-05,
-      "loss": 0.397,
+      "epoch": 2.26,
+      "learning_rate": 2.5809700914303428e-05,
+      "loss": 0.0724,
       "step": 4860
     },
     {
-      "epoch": 3.16,
-      "learning_rate": 2.2796188000866363e-05,
-      "loss": 0.3491,
+      "epoch": 2.26,
+      "learning_rate": 2.5794204246087092e-05,
+      "loss": 0.2819,
       "step": 4870
     },
     {
-      "epoch": 3.17,
-      "learning_rate": 2.2774528914879795e-05,
-      "loss": 0.1067,
+      "epoch": 2.27,
+      "learning_rate": 2.577870757787076e-05,
+      "loss": 0.1272,
       "step": 4880
     },
     {
-      "epoch": 3.18,
-      "learning_rate": 2.2752869828893224e-05,
-      "loss": 0.2016,
+      "epoch": 2.27,
+      "learning_rate": 2.5763210909654427e-05,
+      "loss": 0.327,
       "step": 4890
     },
     {
-      "epoch": 3.18,
-      "learning_rate": 2.273121074290665e-05,
-      "loss": 0.2037,
+      "epoch": 2.28,
+      "learning_rate": 2.5747714241438094e-05,
+      "loss": 0.6902,
       "step": 4900
     },
     {
-      "epoch": 3.19,
-      "learning_rate": 2.2709551656920078e-05,
-      "loss": 0.2404,
+      "epoch": 2.28,
+      "learning_rate": 2.5732217573221758e-05,
+      "loss": 0.3627,
       "step": 4910
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 2.2687892570933507e-05,
-      "loss": 0.1275,
+      "epoch": 2.29,
+      "learning_rate": 2.5716720905005426e-05,
+      "loss": 0.2974,
       "step": 4920
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 2.2666233484946935e-05,
-      "loss": 0.0866,
+      "epoch": 2.29,
+      "learning_rate": 2.570122423678909e-05,
+      "loss": 0.3346,
       "step": 4930
     },
     {
-      "epoch": 3.21,
-      "learning_rate": 2.2644574398960364e-05,
-      "loss": 0.2417,
+      "epoch": 2.3,
+      "learning_rate": 2.5685727568572757e-05,
+      "loss": 0.108,
       "step": 4940
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 2.2622915312973793e-05,
-      "loss": 0.0546,
+      "epoch": 2.3,
+      "learning_rate": 2.567023090035642e-05,
+      "loss": 0.1823,
       "step": 4950
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 2.260125622698722e-05,
-      "loss": 0.2839,
+      "epoch": 2.31,
+      "learning_rate": 2.565473423214009e-05,
+      "loss": 0.301,
       "step": 4960
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 2.2579597141000653e-05,
-      "loss": 0.1398,
+      "epoch": 2.31,
+      "learning_rate": 2.5639237563923756e-05,
+      "loss": 0.3014,
       "step": 4970
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 2.255793805501408e-05,
-      "loss": 0.3108,
+      "epoch": 2.32,
+      "learning_rate": 2.5623740895707424e-05,
+      "loss": 0.2479,
       "step": 4980
     },
     {
-      "epoch": 3.24,
-      "learning_rate": 2.2536278969027507e-05,
-      "loss": 0.4113,
+      "epoch": 2.32,
+      "learning_rate": 2.5608244227491088e-05,
+      "loss": 0.2455,
       "step": 4990
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 2.2514619883040936e-05,
-      "loss": 0.15,
+      "epoch": 2.32,
+      "learning_rate": 2.5592747559274755e-05,
+      "loss": 0.4883,
       "step": 5000
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 2.2492960797054365e-05,
-      "loss": 0.1093,
+      "epoch": 2.33,
+      "learning_rate": 2.5577250891058423e-05,
+      "loss": 0.2843,
       "step": 5010
     },
     {
-      "epoch": 3.26,
-      "learning_rate": 2.2471301711067793e-05,
-      "loss": 0.3478,
+      "epoch": 2.33,
+      "learning_rate": 2.556175422284209e-05,
+      "loss": 0.5232,
       "step": 5020
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 2.2449642625081222e-05,
-      "loss": 0.229,
+      "epoch": 2.34,
+      "learning_rate": 2.5546257554625755e-05,
+      "loss": 0.1704,
       "step": 5030
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 2.242798353909465e-05,
-      "loss": 0.2711,
+      "epoch": 2.34,
+      "learning_rate": 2.5530760886409422e-05,
+      "loss": 0.6146,
       "step": 5040
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 2.240632445310808e-05,
-      "loss": 0.1047,
+      "epoch": 2.35,
+      "learning_rate": 2.551526421819309e-05,
+      "loss": 0.2083,
       "step": 5050
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 2.2384665367121508e-05,
-      "loss": 0.1989,
+      "epoch": 2.35,
+      "learning_rate": 2.5499767549976757e-05,
+      "loss": 0.2732,
       "step": 5060
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 2.2363006281134937e-05,
-      "loss": 0.1658,
+      "epoch": 2.36,
+      "learning_rate": 2.548427088176042e-05,
+      "loss": 0.4491,
       "step": 5070
     },
     {
-      "epoch": 3.3,
-      "learning_rate": 2.2341347195148366e-05,
-      "loss": 0.1486,
+      "epoch": 2.36,
+      "learning_rate": 2.546877421354409e-05,
+      "loss": 0.3135,
       "step": 5080
     },
     {
-      "epoch": 3.31,
-      "learning_rate": 2.2319688109161794e-05,
-      "loss": 0.1316,
+      "epoch": 2.37,
+      "learning_rate": 2.5453277545327756e-05,
+      "loss": 0.2734,
       "step": 5090
     },
     {
-      "epoch": 3.31,
-      "learning_rate": 2.2298029023175223e-05,
-      "loss": 0.087,
+      "epoch": 2.37,
+      "learning_rate": 2.5437780877111424e-05,
+      "loss": 0.1615,
       "step": 5100
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 2.227636993718865e-05,
-      "loss": 0.3231,
+      "epoch": 2.38,
+      "learning_rate": 2.5422284208895088e-05,
+      "loss": 0.2685,
       "step": 5110
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 2.225471085120208e-05,
-      "loss": 0.1468,
+      "epoch": 2.38,
+      "learning_rate": 2.5406787540678755e-05,
+      "loss": 0.4313,
       "step": 5120
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 2.223305176521551e-05,
-      "loss": 0.1442,
+      "epoch": 2.38,
+      "learning_rate": 2.5391290872462423e-05,
+      "loss": 0.3828,
       "step": 5130
     },
     {
-      "epoch": 3.34,
-      "learning_rate": 2.2211392679228934e-05,
-      "loss": 0.3304,
+      "epoch": 2.39,
+      "learning_rate": 2.537579420424609e-05,
+      "loss": 0.1468,
       "step": 5140
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 2.2189733593242366e-05,
-      "loss": 0.2766,
+      "epoch": 2.39,
+      "learning_rate": 2.5360297536029754e-05,
+      "loss": 0.1752,
       "step": 5150
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 2.2168074507255795e-05,
-      "loss": 0.1563,
+      "epoch": 2.4,
+      "learning_rate": 2.534480086781342e-05,
+      "loss": 0.3589,
       "step": 5160
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 2.2146415421269224e-05,
-      "loss": 0.1921,
+      "epoch": 2.4,
+      "learning_rate": 2.532930419959709e-05,
+      "loss": 0.2682,
       "step": 5170
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 2.2124756335282652e-05,
-      "loss": 0.1017,
+      "epoch": 2.41,
+      "learning_rate": 2.5313807531380753e-05,
+      "loss": 0.4267,
       "step": 5180
     },
     {
-      "epoch": 3.37,
-      "learning_rate": 2.210309724929608e-05,
-      "loss": 0.2529,
+      "epoch": 2.41,
+      "learning_rate": 2.5298310863164417e-05,
+      "loss": 0.2382,
       "step": 5190
     },
     {
-      "epoch": 3.38,
-      "learning_rate": 2.208143816330951e-05,
-      "loss": 0.0399,
+      "epoch": 2.42,
+      "learning_rate": 2.5282814194948085e-05,
+      "loss": 0.2842,
       "step": 5200
     },
     {
-      "epoch": 3.38,
-      "learning_rate": 2.205977907732294e-05,
-      "loss": 0.417,
+      "epoch": 2.42,
+      "learning_rate": 2.5267317526731752e-05,
+      "loss": 0.2339,
       "step": 5210
     },
     {
-      "epoch": 3.39,
-      "learning_rate": 2.2038119991336364e-05,
-      "loss": 0.6172,
+      "epoch": 2.43,
+      "learning_rate": 2.525182085851542e-05,
+      "loss": 0.2628,
       "step": 5220
     },
     {
-      "epoch": 3.4,
-      "learning_rate": 2.2016460905349792e-05,
-      "loss": 0.1713,
+      "epoch": 2.43,
+      "learning_rate": 2.5236324190299084e-05,
+      "loss": 0.4855,
       "step": 5230
     },
     {
-      "epoch": 3.4,
-      "learning_rate": 2.1994801819363224e-05,
-      "loss": 0.4274,
+      "epoch": 2.44,
+      "learning_rate": 2.522082752208275e-05,
+      "loss": 0.2352,
       "step": 5240
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 2.1973142733376653e-05,
-      "loss": 0.3577,
+      "epoch": 2.44,
+      "learning_rate": 2.520533085386642e-05,
+      "loss": 0.1419,
       "step": 5250
     },
     {
-      "epoch": 3.42,
-      "learning_rate": 2.1951483647390082e-05,
-      "loss": 0.2066,
+      "epoch": 2.45,
+      "learning_rate": 2.5189834185650086e-05,
+      "loss": 1.0735,
       "step": 5260
     },
     {
-      "epoch": 3.42,
-      "learning_rate": 2.192982456140351e-05,
-      "loss": 0.3221,
+      "epoch": 2.45,
+      "learning_rate": 2.517433751743375e-05,
+      "loss": 0.5396,
       "step": 5270
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 2.190816547541694e-05,
-      "loss": 0.1747,
+      "epoch": 2.45,
+      "learning_rate": 2.5158840849217418e-05,
+      "loss": 0.3396,
       "step": 5280
     },
     {
-      "epoch": 3.44,
-      "learning_rate": 2.1886506389430368e-05,
-      "loss": 0.1228,
+      "epoch": 2.46,
+      "learning_rate": 2.5143344181001085e-05,
+      "loss": 0.4928,
       "step": 5290
     },
     {
-      "epoch": 3.44,
-      "learning_rate": 2.1864847303443793e-05,
-      "loss": 0.1929,
+      "epoch": 2.46,
+      "learning_rate": 2.5127847512784753e-05,
+      "loss": 0.1598,
       "step": 5300
     },
     {
-      "epoch": 3.45,
-      "learning_rate": 2.1843188217457222e-05,
-      "loss": 0.2565,
+      "epoch": 2.47,
+      "learning_rate": 2.5112350844568417e-05,
+      "loss": 0.4678,
       "step": 5310
     },
     {
-      "epoch": 3.46,
-      "learning_rate": 2.182152913147065e-05,
-      "loss": 0.159,
+      "epoch": 2.47,
+      "learning_rate": 2.5096854176352085e-05,
+      "loss": 0.359,
       "step": 5320
     },
     {
-      "epoch": 3.46,
-      "learning_rate": 2.1799870045484083e-05,
-      "loss": 0.1915,
+      "epoch": 2.48,
+      "learning_rate": 2.5081357508135752e-05,
+      "loss": 0.2221,
       "step": 5330
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 2.177821095949751e-05,
-      "loss": 0.2584,
+      "epoch": 2.48,
+      "learning_rate": 2.506586083991942e-05,
+      "loss": 0.1983,
       "step": 5340
     },
     {
-      "epoch": 3.48,
-      "learning_rate": 2.175655187351094e-05,
-      "loss": 0.162,
+      "epoch": 2.49,
+      "learning_rate": 2.5050364171703084e-05,
+      "loss": 0.4923,
       "step": 5350
     },
     {
-      "epoch": 3.48,
-      "learning_rate": 2.173489278752437e-05,
-      "loss": 0.1473,
+      "epoch": 2.49,
+      "learning_rate": 2.503486750348675e-05,
+      "loss": 0.103,
       "step": 5360
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 2.1713233701537797e-05,
-      "loss": 0.4753,
+      "epoch": 2.5,
+      "learning_rate": 2.501937083527042e-05,
+      "loss": 0.1654,
       "step": 5370
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 2.1691574615551223e-05,
-      "loss": 0.1635,
+      "epoch": 2.5,
+      "learning_rate": 2.5003874167054086e-05,
+      "loss": 0.5142,
       "step": 5380
     },
     {
-      "epoch": 3.5,
-      "learning_rate": 2.166991552956465e-05,
-      "loss": 0.2713,
+      "epoch": 2.51,
+      "learning_rate": 2.498837749883775e-05,
+      "loss": 0.3632,
       "step": 5390
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 2.164825644357808e-05,
-      "loss": 0.4024,
+      "epoch": 2.51,
+      "learning_rate": 2.4972880830621418e-05,
+      "loss": 0.0935,
       "step": 5400
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 2.162659735759151e-05,
-      "loss": 0.3595,
+      "epoch": 2.52,
+      "learning_rate": 2.4957384162405085e-05,
+      "loss": 0.4564,
       "step": 5410
     },
     {
-      "epoch": 3.52,
-      "learning_rate": 2.160493827160494e-05,
-      "loss": 0.2345,
+      "epoch": 2.52,
+      "learning_rate": 2.4941887494188753e-05,
+      "loss": 0.4198,
       "step": 5420
     },
     {
-      "epoch": 3.53,
-      "learning_rate": 2.158327918561837e-05,
-      "loss": 0.2353,
+      "epoch": 2.52,
+      "learning_rate": 2.4926390825972413e-05,
+      "loss": 0.1952,
       "step": 5430
     },
     {
-      "epoch": 3.53,
-      "learning_rate": 2.1561620099631798e-05,
-      "loss": 0.3144,
+      "epoch": 2.53,
+      "learning_rate": 2.491089415775608e-05,
+      "loss": 0.3286,
       "step": 5440
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 2.1539961013645227e-05,
-      "loss": 0.558,
+      "epoch": 2.53,
+      "learning_rate": 2.4895397489539748e-05,
+      "loss": 0.2146,
       "step": 5450
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 2.1518301927658652e-05,
-      "loss": 0.1033,
+      "epoch": 2.54,
+      "learning_rate": 2.4879900821323416e-05,
+      "loss": 0.3435,
       "step": 5460
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 2.149664284167208e-05,
-      "loss": 0.2233,
+      "epoch": 2.54,
+      "learning_rate": 2.4864404153107083e-05,
+      "loss": 0.3927,
       "step": 5470
     },
     {
-      "epoch": 3.56,
-      "learning_rate": 2.147498375568551e-05,
-      "loss": 0.2178,
+      "epoch": 2.55,
+      "learning_rate": 2.4848907484890747e-05,
+      "loss": 0.3311,
       "step": 5480
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 2.1453324669698938e-05,
-      "loss": 0.0916,
+      "epoch": 2.55,
+      "learning_rate": 2.4833410816674415e-05,
+      "loss": 0.3258,
       "step": 5490
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 2.1431665583712367e-05,
-      "loss": 0.051,
+      "epoch": 2.56,
+      "learning_rate": 2.4817914148458082e-05,
+      "loss": 0.2766,
       "step": 5500
     },
     {
-      "epoch": 3.58,
-      "learning_rate": 2.14100064977258e-05,
-      "loss": 0.3294,
+      "epoch": 2.56,
+      "learning_rate": 2.480241748024175e-05,
+      "loss": 0.311,
       "step": 5510
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 2.1388347411739228e-05,
-      "loss": 0.1862,
+      "epoch": 2.57,
+      "learning_rate": 2.4786920812025414e-05,
+      "loss": 0.5541,
       "step": 5520
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 2.1366688325752653e-05,
-      "loss": 0.2174,
+      "epoch": 2.57,
+      "learning_rate": 2.477142414380908e-05,
+      "loss": 0.2717,
       "step": 5530
     },
     {
-      "epoch": 3.6,
-      "learning_rate": 2.134502923976608e-05,
-      "loss": 0.2679,
+      "epoch": 2.58,
+      "learning_rate": 2.475592747559275e-05,
+      "loss": 0.2383,
       "step": 5540
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 2.132337015377951e-05,
-      "loss": 0.1024,
+      "epoch": 2.58,
+      "learning_rate": 2.4740430807376416e-05,
+      "loss": 0.1974,
       "step": 5550
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 2.130171106779294e-05,
-      "loss": 0.2108,
+      "epoch": 2.58,
+      "learning_rate": 2.472493413916008e-05,
+      "loss": 0.3493,
       "step": 5560
     },
     {
-      "epoch": 3.62,
-      "learning_rate": 2.1280051981806368e-05,
-      "loss": 0.1707,
+      "epoch": 2.59,
+      "learning_rate": 2.4709437470943748e-05,
+      "loss": 0.4594,
       "step": 5570
     },
     {
-      "epoch": 3.62,
-      "learning_rate": 2.1258392895819796e-05,
-      "loss": 0.1842,
+      "epoch": 2.59,
+      "learning_rate": 2.4693940802727415e-05,
+      "loss": 0.3031,
       "step": 5580
     },
     {
-      "epoch": 3.63,
-      "learning_rate": 2.1236733809833225e-05,
-      "loss": 0.0681,
+      "epoch": 2.6,
+      "learning_rate": 2.4678444134511083e-05,
+      "loss": 0.1899,
       "step": 5590
     },
     {
-      "epoch": 3.64,
-      "learning_rate": 2.1215074723846657e-05,
-      "loss": 0.3176,
+      "epoch": 2.6,
+      "learning_rate": 2.4662947466294747e-05,
+      "loss": 0.2106,
       "step": 5600
     },
     {
-      "epoch": 3.64,
-      "learning_rate": 2.1193415637860082e-05,
-      "loss": 0.3924,
+      "epoch": 2.61,
+      "learning_rate": 2.4647450798078414e-05,
+      "loss": 0.366,
       "step": 5610
     },
     {
-      "epoch": 3.65,
-      "learning_rate": 2.117175655187351e-05,
-      "loss": 0.307,
+      "epoch": 2.61,
+      "learning_rate": 2.4631954129862082e-05,
+      "loss": 0.173,
       "step": 5620
     },
     {
-      "epoch": 3.66,
-      "learning_rate": 2.115009746588694e-05,
-      "loss": 0.164,
+      "epoch": 2.62,
+      "learning_rate": 2.461645746164575e-05,
+      "loss": 0.286,
       "step": 5630
     },
     {
-      "epoch": 3.66,
-      "learning_rate": 2.112843837990037e-05,
-      "loss": 0.1099,
+      "epoch": 2.62,
+      "learning_rate": 2.4600960793429414e-05,
+      "loss": 0.3543,
       "step": 5640
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 2.1106779293913797e-05,
-      "loss": 0.3696,
+      "epoch": 2.63,
+      "learning_rate": 2.458546412521308e-05,
+      "loss": 0.3302,
       "step": 5650
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 2.1085120207927226e-05,
-      "loss": 0.2124,
+      "epoch": 2.63,
+      "learning_rate": 2.456996745699675e-05,
+      "loss": 0.3367,
       "step": 5660
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 2.1063461121940655e-05,
-      "loss": 0.212,
+      "epoch": 2.64,
+      "learning_rate": 2.4554470788780416e-05,
+      "loss": 0.3913,
       "step": 5670
     },
     {
-      "epoch": 3.69,
-      "learning_rate": 2.1041802035954083e-05,
-      "loss": 0.2727,
+      "epoch": 2.64,
+      "learning_rate": 2.4538974120564077e-05,
+      "loss": 0.421,
       "step": 5680
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 2.1020142949967512e-05,
-      "loss": 0.2698,
+      "epoch": 2.65,
+      "learning_rate": 2.4523477452347744e-05,
+      "loss": 0.4009,
       "step": 5690
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 2.099848386398094e-05,
-      "loss": 0.4189,
+      "epoch": 2.65,
+      "learning_rate": 2.450798078413141e-05,
+      "loss": 0.1738,
       "step": 5700
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 2.097682477799437e-05,
-      "loss": 0.1711,
+      "epoch": 2.65,
+      "learning_rate": 2.449248411591508e-05,
+      "loss": 0.2611,
       "step": 5710
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 2.0955165692007798e-05,
-      "loss": 0.1526,
+      "epoch": 2.66,
+      "learning_rate": 2.4476987447698743e-05,
+      "loss": 0.4684,
       "step": 5720
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 2.0933506606021227e-05,
-      "loss": 0.1894,
+      "epoch": 2.66,
+      "learning_rate": 2.446149077948241e-05,
+      "loss": 0.2115,
       "step": 5730
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 2.0911847520034655e-05,
-      "loss": 0.2144,
+      "epoch": 2.67,
+      "learning_rate": 2.4445994111266078e-05,
+      "loss": 0.1809,
       "step": 5740
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 2.0890188434048084e-05,
-      "loss": 0.1443,
+      "epoch": 2.67,
+      "learning_rate": 2.4430497443049746e-05,
+      "loss": 0.2061,
       "step": 5750
     },
     {
-      "epoch": 3.74,
-      "learning_rate": 2.0868529348061513e-05,
-      "loss": 0.0814,
+      "epoch": 2.68,
+      "learning_rate": 2.441500077483341e-05,
+      "loss": 0.3943,
       "step": 5760
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 2.0846870262074938e-05,
-      "loss": 0.2765,
+      "epoch": 2.68,
+      "learning_rate": 2.4399504106617077e-05,
+      "loss": 0.3,
       "step": 5770
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 2.082521117608837e-05,
-      "loss": 0.1244,
+      "epoch": 2.69,
+      "learning_rate": 2.4384007438400745e-05,
+      "loss": 0.222,
       "step": 5780
     },
     {
-      "epoch": 3.76,
-      "learning_rate": 2.08035520901018e-05,
-      "loss": 0.2107,
+      "epoch": 2.69,
+      "learning_rate": 2.4368510770184412e-05,
+      "loss": 0.3432,
       "step": 5790
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 2.0781893004115227e-05,
-      "loss": 0.0134,
+      "epoch": 2.7,
+      "learning_rate": 2.4353014101968076e-05,
+      "loss": 0.19,
       "step": 5800
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 2.0760233918128656e-05,
-      "loss": 0.1578,
+      "epoch": 2.7,
+      "learning_rate": 2.4337517433751744e-05,
+      "loss": 0.323,
       "step": 5810
     },
     {
-      "epoch": 3.78,
-      "learning_rate": 2.074074074074074e-05,
-      "loss": 0.1113,
+      "epoch": 2.71,
+      "learning_rate": 2.432202076553541e-05,
+      "loss": 0.3536,
       "step": 5820
     },
     {
-      "epoch": 3.79,
-      "learning_rate": 2.071908165475417e-05,
-      "loss": 0.1734,
+      "epoch": 2.71,
+      "learning_rate": 2.430652409731908e-05,
+      "loss": 0.4613,
       "step": 5830
     },
     {
-      "epoch": 3.79,
-      "learning_rate": 2.06974225687676e-05,
-      "loss": 0.2593,
+      "epoch": 2.72,
+      "learning_rate": 2.4291027429102743e-05,
+      "loss": 0.1771,
       "step": 5840
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 2.067576348278103e-05,
-      "loss": 0.4005,
+      "epoch": 2.72,
+      "learning_rate": 2.427553076088641e-05,
+      "loss": 0.3451,
       "step": 5850
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 2.0654104396794458e-05,
-      "loss": 0.1157,
+      "epoch": 2.72,
+      "learning_rate": 2.4260034092670078e-05,
+      "loss": 0.2012,
       "step": 5860
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 2.0632445310807883e-05,
-      "loss": 0.1248,
+      "epoch": 2.73,
+      "learning_rate": 2.4244537424453745e-05,
+      "loss": 0.3176,
       "step": 5870
     },
     {
-      "epoch": 3.82,
-      "learning_rate": 2.0610786224821312e-05,
-      "loss": 0.1108,
+      "epoch": 2.73,
+      "learning_rate": 2.422904075623741e-05,
+      "loss": 0.2653,
       "step": 5880
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 2.058912713883474e-05,
-      "loss": 0.1242,
+      "epoch": 2.74,
+      "learning_rate": 2.4213544088021077e-05,
+      "loss": 0.3311,
       "step": 5890
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 2.056746805284817e-05,
-      "loss": 0.1443,
+      "epoch": 2.74,
+      "learning_rate": 2.4198047419804744e-05,
+      "loss": 0.1197,
       "step": 5900
     },
     {
-      "epoch": 3.84,
-      "learning_rate": 2.0545808966861598e-05,
-      "loss": 0.1957,
+      "epoch": 2.75,
+      "learning_rate": 2.4182550751588412e-05,
+      "loss": 0.1639,
       "step": 5910
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 2.0524149880875027e-05,
-      "loss": 0.323,
+      "epoch": 2.75,
+      "learning_rate": 2.4167054083372076e-05,
+      "loss": 0.4772,
       "step": 5920
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 2.050249079488846e-05,
-      "loss": 0.3315,
+      "epoch": 2.76,
+      "learning_rate": 2.415155741515574e-05,
+      "loss": 0.2448,
       "step": 5930
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 2.0480831708901888e-05,
-      "loss": 0.311,
+      "epoch": 2.76,
+      "learning_rate": 2.4136060746939408e-05,
+      "loss": 0.1919,
       "step": 5940
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 2.0459172622915313e-05,
-      "loss": 0.1231,
+      "epoch": 2.77,
+      "learning_rate": 2.4120564078723075e-05,
+      "loss": 0.1686,
       "step": 5950
     },
     {
-      "epoch": 3.87,
-      "learning_rate": 2.043751353692874e-05,
-      "loss": 0.1074,
+      "epoch": 2.77,
+      "learning_rate": 2.410506741050674e-05,
+      "loss": 0.1621,
       "step": 5960
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 2.041585445094217e-05,
-      "loss": 0.1604,
+      "epoch": 2.78,
+      "learning_rate": 2.4089570742290407e-05,
+      "loss": 0.3617,
       "step": 5970
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 2.03941953649556e-05,
-      "loss": 0.3031,
+      "epoch": 2.78,
+      "learning_rate": 2.4074074074074074e-05,
+      "loss": 0.6375,
       "step": 5980
     },
     {
-      "epoch": 3.89,
-      "learning_rate": 2.0372536278969028e-05,
-      "loss": 0.4937,
+      "epoch": 2.78,
+      "learning_rate": 2.405857740585774e-05,
+      "loss": 0.1808,
       "step": 5990
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 2.0350877192982456e-05,
-      "loss": 0.1419,
+      "epoch": 2.79,
+      "learning_rate": 2.4043080737641406e-05,
+      "loss": 0.2978,
       "step": 6000
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 2.0329218106995885e-05,
-      "loss": 0.2078,
+      "epoch": 2.79,
+      "learning_rate": 2.4027584069425073e-05,
+      "loss": 0.1222,
       "step": 6010
     },
     {
-      "epoch": 3.91,
-      "learning_rate": 2.0307559021009317e-05,
-      "loss": 0.1859,
+      "epoch": 2.8,
+      "learning_rate": 2.401208740120874e-05,
+      "loss": 0.3246,
       "step": 6020
     },
     {
-      "epoch": 3.92,
-      "learning_rate": 2.0285899935022742e-05,
-      "loss": 0.2744,
+      "epoch": 2.8,
+      "learning_rate": 2.3996590732992408e-05,
+      "loss": 0.3239,
       "step": 6030
     },
     {
-      "epoch": 3.92,
-      "learning_rate": 2.026424084903617e-05,
-      "loss": 0.039,
+      "epoch": 2.81,
+      "learning_rate": 2.3981094064776072e-05,
+      "loss": 0.1893,
       "step": 6040
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 2.02425817630496e-05,
-      "loss": 0.2164,
+      "epoch": 2.81,
+      "learning_rate": 2.396559739655974e-05,
+      "loss": 0.2506,
       "step": 6050
     },
     {
-      "epoch": 3.94,
-      "learning_rate": 2.022092267706303e-05,
-      "loss": 0.231,
+      "epoch": 2.82,
+      "learning_rate": 2.3950100728343407e-05,
+      "loss": 0.2423,
       "step": 6060
     },
     {
-      "epoch": 3.94,
-      "learning_rate": 2.0199263591076457e-05,
-      "loss": 0.182,
+      "epoch": 2.82,
+      "learning_rate": 2.3934604060127075e-05,
+      "loss": 0.6559,
       "step": 6070
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 2.0177604505089886e-05,
-      "loss": 0.0891,
+      "epoch": 2.83,
+      "learning_rate": 2.391910739191074e-05,
+      "loss": 0.2882,
       "step": 6080
     },
     {
-      "epoch": 3.96,
-      "learning_rate": 2.0155945419103314e-05,
-      "loss": 0.2888,
+      "epoch": 2.83,
+      "learning_rate": 2.3903610723694406e-05,
+      "loss": 0.4461,
       "step": 6090
     },
     {
-      "epoch": 3.96,
-      "learning_rate": 2.0134286333116743e-05,
-      "loss": 0.2661,
+      "epoch": 2.84,
+      "learning_rate": 2.3888114055478074e-05,
+      "loss": 0.3032,
       "step": 6100
     },
     {
-      "epoch": 3.97,
-      "learning_rate": 2.0112627247130172e-05,
-      "loss": 0.2692,
+      "epoch": 2.84,
+      "learning_rate": 2.387261738726174e-05,
+      "loss": 0.1458,
       "step": 6110
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 2.00909681611436e-05,
-      "loss": 0.1445,
+      "epoch": 2.85,
+      "learning_rate": 2.3857120719045405e-05,
+      "loss": 0.296,
       "step": 6120
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 2.006930907515703e-05,
-      "loss": 0.2447,
+      "epoch": 2.85,
+      "learning_rate": 2.3841624050829073e-05,
+      "loss": 0.186,
       "step": 6130
     },
     {
-      "epoch": 3.99,
-      "learning_rate": 2.0047649989170458e-05,
-      "loss": 0.1887,
+      "epoch": 2.85,
+      "learning_rate": 2.382612738261274e-05,
+      "loss": 0.2157,
       "step": 6140
     },
     {
-      "epoch": 3.99,
-      "learning_rate": 2.0025990903183887e-05,
-      "loss": 0.2405,
+      "epoch": 2.86,
+      "learning_rate": 2.3810630714396408e-05,
+      "loss": 0.218,
       "step": 6150
     },
     {
-      "epoch": 4.0,
-      "eval_accuracy": 0.9793008279668813,
-      "eval_loss": 0.10150198638439178,
-      "eval_runtime": 12.9625,
-      "eval_samples_per_second": 167.715,
-      "eval_steps_per_second": 41.967,
-      "step": 6158
-    },
-    {
-      "epoch": 4.0,
-      "learning_rate": 2.0004331817197315e-05,
-      "loss": 0.172,
+      "epoch": 2.86,
+      "learning_rate": 2.3795134046180072e-05,
+      "loss": 0.3147,
       "step": 6160
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 1.9982672731210744e-05,
-      "loss": 0.2265,
+      "epoch": 2.87,
+      "learning_rate": 2.377963737796374e-05,
+      "loss": 0.4664,
       "step": 6170
     },
     {
-      "epoch": 4.01,
-      "learning_rate": 1.9961013645224173e-05,
-      "loss": 0.1351,
+      "epoch": 2.87,
+      "learning_rate": 2.3764140709747403e-05,
+      "loss": 0.3786,
       "step": 6180
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 1.9939354559237598e-05,
-      "loss": 0.2882,
+      "epoch": 2.88,
+      "learning_rate": 2.374864404153107e-05,
+      "loss": 0.1602,
       "step": 6190
     },
     {
-      "epoch": 4.03,
-      "learning_rate": 1.991769547325103e-05,
-      "loss": 0.2578,
+      "epoch": 2.88,
+      "learning_rate": 2.3733147373314735e-05,
+      "loss": 0.0418,
       "step": 6200
     },
     {
-      "epoch": 4.03,
-      "learning_rate": 1.989603638726446e-05,
-      "loss": 0.2349,
+      "epoch": 2.89,
+      "learning_rate": 2.3717650705098403e-05,
+      "loss": 0.2606,
       "step": 6210
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 1.9874377301277887e-05,
-      "loss": 0.3212,
+      "epoch": 2.89,
+      "learning_rate": 2.370215403688207e-05,
+      "loss": 0.2046,
       "step": 6220
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 1.9852718215291316e-05,
-      "loss": 0.2314,
+      "epoch": 2.9,
+      "learning_rate": 2.3686657368665737e-05,
+      "loss": 0.4652,
       "step": 6230
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 1.9831059129304745e-05,
-      "loss": 0.1418,
+      "epoch": 2.9,
+      "learning_rate": 2.36711607004494e-05,
+      "loss": 0.3277,
       "step": 6240
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 1.9809400043318173e-05,
-      "loss": 0.1055,
+      "epoch": 2.91,
+      "learning_rate": 2.365566403223307e-05,
+      "loss": 0.4152,
       "step": 6250
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 1.9787740957331602e-05,
-      "loss": 0.1424,
+      "epoch": 2.91,
+      "learning_rate": 2.3640167364016737e-05,
+      "loss": 0.2766,
       "step": 6260
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 1.9766081871345027e-05,
-      "loss": 0.1625,
+      "epoch": 2.91,
+      "learning_rate": 2.3624670695800404e-05,
+      "loss": 0.4198,
       "step": 6270
     },
     {
-      "epoch": 4.08,
-      "learning_rate": 1.9744422785358456e-05,
-      "loss": 0.2511,
+      "epoch": 2.92,
+      "learning_rate": 2.3609174027584068e-05,
+      "loss": 0.2715,
       "step": 6280
     },
     {
-      "epoch": 4.09,
-      "learning_rate": 1.9722763699371888e-05,
-      "loss": 0.2538,
+      "epoch": 2.92,
+      "learning_rate": 2.3595227026189372e-05,
+      "loss": 0.247,
       "step": 6290
     },
     {
-      "epoch": 4.09,
-      "learning_rate": 1.9701104613385317e-05,
-      "loss": 0.0055,
+      "epoch": 2.93,
+      "learning_rate": 2.3579730357973036e-05,
+      "loss": 0.1352,
       "step": 6300
     },
     {
-      "epoch": 4.1,
-      "learning_rate": 1.9679445527398745e-05,
-      "loss": 0.2292,
+      "epoch": 2.93,
+      "learning_rate": 2.3564233689756704e-05,
+      "loss": 0.2075,
       "step": 6310
     },
     {
-      "epoch": 4.11,
-      "learning_rate": 1.9657786441412174e-05,
-      "loss": 0.1772,
+      "epoch": 2.94,
+      "learning_rate": 2.354873702154037e-05,
+      "loss": 0.4621,
       "step": 6320
     },
     {
-      "epoch": 4.11,
-      "learning_rate": 1.9636127355425603e-05,
-      "loss": 0.2582,
+      "epoch": 2.94,
+      "learning_rate": 2.3533240353324035e-05,
+      "loss": 0.221,
       "step": 6330
     },
     {
-      "epoch": 4.12,
-      "learning_rate": 1.9614468269439028e-05,
-      "loss": 0.1229,
+      "epoch": 2.95,
+      "learning_rate": 2.35177436851077e-05,
+      "loss": 0.1743,
       "step": 6340
     },
     {
-      "epoch": 4.12,
-      "learning_rate": 1.9592809183452457e-05,
-      "loss": 0.0285,
+      "epoch": 2.95,
+      "learning_rate": 2.3502247016891367e-05,
+      "loss": 0.3211,
       "step": 6350
     },
     {
-      "epoch": 4.13,
-      "learning_rate": 1.9571150097465886e-05,
-      "loss": 0.1899,
+      "epoch": 2.96,
+      "learning_rate": 2.3486750348675034e-05,
+      "loss": 0.458,
       "step": 6360
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 1.9549491011479314e-05,
-      "loss": 0.4994,
+      "epoch": 2.96,
+      "learning_rate": 2.3471253680458702e-05,
+      "loss": 0.1484,
       "step": 6370
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 1.9527831925492746e-05,
-      "loss": 0.2171,
+      "epoch": 2.97,
+      "learning_rate": 2.3455757012242366e-05,
+      "loss": 0.2771,
       "step": 6380
     },
     {
-      "epoch": 4.15,
-      "learning_rate": 1.9506172839506175e-05,
-      "loss": 0.2057,
-      "step": 6390
+      "epoch": 2.97,
+      "learning_rate": 2.3440260344026034e-05,
+      "loss": 0.3501,
+      "step": 6390
     },
     {
-      "epoch": 4.16,
-      "learning_rate": 1.9484513753519604e-05,
-      "loss": 0.2704,
+      "epoch": 2.98,
+      "learning_rate": 2.34247636758097e-05,
+      "loss": 0.3302,
       "step": 6400
     },
     {
-      "epoch": 4.16,
-      "learning_rate": 1.9462854667533032e-05,
-      "loss": 0.1213,
+      "epoch": 2.98,
+      "learning_rate": 2.340926700759337e-05,
+      "loss": 0.34,
       "step": 6410
     },
     {
-      "epoch": 4.17,
-      "learning_rate": 1.9441195581546458e-05,
-      "loss": 0.2396,
+      "epoch": 2.98,
+      "learning_rate": 2.3393770339377033e-05,
+      "loss": 0.3771,
       "step": 6420
     },
     {
-      "epoch": 4.18,
-      "learning_rate": 1.9419536495559886e-05,
-      "loss": 0.0396,
+      "epoch": 2.99,
+      "learning_rate": 2.33782736711607e-05,
+      "loss": 0.1353,
       "step": 6430
     },
     {
-      "epoch": 4.18,
-      "learning_rate": 1.9397877409573315e-05,
-      "loss": 0.2299,
+      "epoch": 2.99,
+      "learning_rate": 2.3362777002944368e-05,
+      "loss": 0.2731,
       "step": 6440
     },
     {
-      "epoch": 4.19,
-      "learning_rate": 1.9376218323586744e-05,
-      "loss": 0.0109,
+      "epoch": 3.0,
+      "learning_rate": 2.3347280334728035e-05,
+      "loss": 0.4494,
       "step": 6450
     },
     {
-      "epoch": 4.2,
-      "learning_rate": 1.9354559237600172e-05,
-      "loss": 0.0679,
+      "epoch": 3.0,
+      "eval_accuracy": 0.9644385907145209,
+      "eval_loss": 0.16907474398612976,
+      "eval_runtime": 17.2887,
+      "eval_samples_per_second": 175.664,
+      "eval_steps_per_second": 43.959,
+      "step": 6453
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.33317836665117e-05,
+      "loss": 0.4246,
       "step": 6460
     },
     {
-      "epoch": 4.2,
-      "learning_rate": 1.9332900151613604e-05,
-      "loss": 0.1354,
+      "epoch": 3.01,
+      "learning_rate": 2.3316286998295367e-05,
+      "loss": 0.2103,
       "step": 6470
     },
     {
-      "epoch": 4.21,
-      "learning_rate": 1.9311241065627033e-05,
-      "loss": 0.1422,
+      "epoch": 3.01,
+      "learning_rate": 2.3300790330079034e-05,
+      "loss": 0.4987,
       "step": 6480
     },
     {
-      "epoch": 4.22,
-      "learning_rate": 1.9289581979640462e-05,
-      "loss": 0.3022,
+      "epoch": 3.02,
+      "learning_rate": 2.32852936618627e-05,
+      "loss": 0.2063,
       "step": 6490
     },
     {
-      "epoch": 4.22,
-      "learning_rate": 1.9267922893653887e-05,
-      "loss": 0.274,
+      "epoch": 3.02,
+      "learning_rate": 2.3269796993646366e-05,
+      "loss": 0.1851,
       "step": 6500
     },
     {
-      "epoch": 4.23,
-      "learning_rate": 1.9246263807667316e-05,
-      "loss": 0.1733,
+      "epoch": 3.03,
+      "learning_rate": 2.3254300325430033e-05,
+      "loss": 0.1201,
       "step": 6510
     },
     {
-      "epoch": 4.24,
-      "learning_rate": 1.9224604721680744e-05,
-      "loss": 0.4914,
+      "epoch": 3.03,
+      "learning_rate": 2.32388036572137e-05,
+      "loss": 0.3403,
       "step": 6520
     },
     {
-      "epoch": 4.24,
-      "learning_rate": 1.9202945635694173e-05,
-      "loss": 0.1881,
+      "epoch": 3.04,
+      "learning_rate": 2.3223306988997368e-05,
+      "loss": 0.0493,
       "step": 6530
     },
     {
-      "epoch": 4.25,
-      "learning_rate": 1.9181286549707602e-05,
-      "loss": 0.1131,
+      "epoch": 3.04,
+      "learning_rate": 2.3207810320781032e-05,
+      "loss": 0.2367,
       "step": 6540
     },
     {
-      "epoch": 4.25,
-      "learning_rate": 1.915962746372103e-05,
-      "loss": 0.3501,
+      "epoch": 3.05,
+      "learning_rate": 2.31923136525647e-05,
+      "loss": 0.2026,
       "step": 6550
     },
     {
-      "epoch": 4.26,
-      "learning_rate": 1.9137968377734463e-05,
-      "loss": 0.1951,
+      "epoch": 3.05,
+      "learning_rate": 2.3176816984348367e-05,
+      "loss": 0.1488,
       "step": 6560
     },
     {
-      "epoch": 4.27,
-      "learning_rate": 1.911630929174789e-05,
-      "loss": 0.1627,
+      "epoch": 3.05,
+      "learning_rate": 2.3161320316132035e-05,
+      "loss": 0.4259,
       "step": 6570
     },
     {
-      "epoch": 4.27,
-      "learning_rate": 1.9094650205761317e-05,
-      "loss": 0.0676,
+      "epoch": 3.06,
+      "learning_rate": 2.3145823647915695e-05,
+      "loss": 0.2624,
       "step": 6580
     },
     {
-      "epoch": 4.28,
-      "learning_rate": 1.9072991119774745e-05,
-      "loss": 0.2998,
+      "epoch": 3.06,
+      "learning_rate": 2.3130326979699363e-05,
+      "loss": 0.3641,
       "step": 6590
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 1.9051332033788174e-05,
-      "loss": 0.1903,
+      "epoch": 3.07,
+      "learning_rate": 2.311483031148303e-05,
+      "loss": 0.2493,
       "step": 6600
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 1.9029672947801603e-05,
-      "loss": 0.1043,
+      "epoch": 3.07,
+      "learning_rate": 2.3099333643266698e-05,
+      "loss": 0.1795,
       "step": 6610
     },
     {
-      "epoch": 4.3,
-      "learning_rate": 1.900801386181503e-05,
-      "loss": 0.2515,
+      "epoch": 3.08,
+      "learning_rate": 2.3083836975050365e-05,
+      "loss": 0.2637,
       "step": 6620
     },
     {
-      "epoch": 4.31,
-      "learning_rate": 1.898635477582846e-05,
-      "loss": 0.3629,
+      "epoch": 3.08,
+      "learning_rate": 2.306834030683403e-05,
+      "loss": 0.2974,
       "step": 6630
     },
     {
-      "epoch": 4.31,
-      "learning_rate": 1.896469568984189e-05,
-      "loss": 0.1478,
+      "epoch": 3.09,
+      "learning_rate": 2.3052843638617697e-05,
+      "loss": 0.2703,
       "step": 6640
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 1.894303660385532e-05,
-      "loss": 0.1651,
+      "epoch": 3.09,
+      "learning_rate": 2.3037346970401364e-05,
+      "loss": 0.3835,
       "step": 6650
     },
     {
-      "epoch": 4.33,
-      "learning_rate": 1.8921377517868746e-05,
-      "loss": 0.1962,
+      "epoch": 3.1,
+      "learning_rate": 2.3021850302185032e-05,
+      "loss": 0.183,
       "step": 6660
     },
     {
-      "epoch": 4.33,
-      "learning_rate": 1.8899718431882175e-05,
-      "loss": 0.0931,
+      "epoch": 3.1,
+      "learning_rate": 2.3006353633968696e-05,
+      "loss": 0.1776,
       "step": 6670
     },
     {
-      "epoch": 4.34,
-      "learning_rate": 1.8878059345895603e-05,
-      "loss": 0.1451,
+      "epoch": 3.11,
+      "learning_rate": 2.2990856965752363e-05,
+      "loss": 0.3091,
       "step": 6680
     },
     {
-      "epoch": 4.35,
-      "learning_rate": 1.8856400259909032e-05,
-      "loss": 0.2826,
+      "epoch": 3.11,
+      "learning_rate": 2.297536029753603e-05,
+      "loss": 0.2412,
       "step": 6690
     },
     {
-      "epoch": 4.35,
-      "learning_rate": 1.883474117392246e-05,
-      "loss": 0.0117,
+      "epoch": 3.11,
+      "learning_rate": 2.29598636293197e-05,
+      "loss": 0.1943,
       "step": 6700
     },
     {
-      "epoch": 4.36,
-      "learning_rate": 1.881308208793589e-05,
-      "loss": 0.2076,
+      "epoch": 3.12,
+      "learning_rate": 2.2944366961103363e-05,
+      "loss": 0.2755,
       "step": 6710
     },
     {
-      "epoch": 4.37,
-      "learning_rate": 1.8791423001949318e-05,
-      "loss": 0.2932,
+      "epoch": 3.12,
+      "learning_rate": 2.292887029288703e-05,
+      "loss": 0.3217,
       "step": 6720
     },
     {
-      "epoch": 4.37,
-      "learning_rate": 1.8769763915962747e-05,
-      "loss": 0.0393,
+      "epoch": 3.13,
+      "learning_rate": 2.2913373624670698e-05,
+      "loss": 0.365,
       "step": 6730
     },
     {
-      "epoch": 4.38,
-      "learning_rate": 1.8748104829976176e-05,
-      "loss": 0.222,
+      "epoch": 3.13,
+      "learning_rate": 2.2897876956454365e-05,
+      "loss": 0.2362,
       "step": 6740
     },
     {
-      "epoch": 4.38,
-      "learning_rate": 1.8726445743989604e-05,
-      "loss": 0.2544,
+      "epoch": 3.14,
+      "learning_rate": 2.288238028823803e-05,
+      "loss": 0.3217,
       "step": 6750
     },
     {
-      "epoch": 4.39,
-      "learning_rate": 1.8704786658003033e-05,
-      "loss": 0.0992,
+      "epoch": 3.14,
+      "learning_rate": 2.2866883620021697e-05,
+      "loss": 0.1733,
       "step": 6760
     },
     {
-      "epoch": 4.4,
-      "learning_rate": 1.868312757201646e-05,
-      "loss": 0.1865,
+      "epoch": 3.15,
+      "learning_rate": 2.2851386951805364e-05,
+      "loss": 0.2835,
       "step": 6770
     },
     {
-      "epoch": 4.4,
-      "learning_rate": 1.866146848602989e-05,
-      "loss": 0.0594,
+      "epoch": 3.15,
+      "learning_rate": 2.283589028358903e-05,
+      "loss": 0.2699,
       "step": 6780
     },
     {
-      "epoch": 4.41,
-      "learning_rate": 1.863980940004332e-05,
-      "loss": 0.1626,
+      "epoch": 3.16,
+      "learning_rate": 2.2820393615372696e-05,
+      "loss": 0.2349,
       "step": 6790
     },
     {
-      "epoch": 4.42,
-      "learning_rate": 1.8618150314056748e-05,
-      "loss": 0.2632,
+      "epoch": 3.16,
+      "learning_rate": 2.2804896947156363e-05,
+      "loss": 0.1391,
       "step": 6800
     },
     {
-      "epoch": 4.42,
-      "learning_rate": 1.8596491228070176e-05,
-      "loss": 0.409,
+      "epoch": 3.17,
+      "learning_rate": 2.278940027894003e-05,
+      "loss": 0.277,
       "step": 6810
     },
     {
-      "epoch": 4.43,
-      "learning_rate": 1.85748321420836e-05,
-      "loss": 0.2442,
+      "epoch": 3.17,
+      "learning_rate": 2.2773903610723698e-05,
+      "loss": 0.2328,
       "step": 6820
     },
     {
-      "epoch": 4.44,
-      "learning_rate": 1.8553173056097034e-05,
-      "loss": 0.1129,
+      "epoch": 3.18,
+      "learning_rate": 2.275840694250736e-05,
+      "loss": 0.3107,
       "step": 6830
     },
     {
-      "epoch": 4.44,
-      "learning_rate": 1.8531513970110462e-05,
-      "loss": 0.2521,
+      "epoch": 3.18,
+      "learning_rate": 2.2742910274291026e-05,
+      "loss": 0.4042,
       "step": 6840
     },
     {
-      "epoch": 4.45,
-      "learning_rate": 1.850985488412389e-05,
-      "loss": 0.2092,
+      "epoch": 3.18,
+      "learning_rate": 2.2727413606074694e-05,
+      "loss": 0.368,
       "step": 6850
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 1.848819579813732e-05,
-      "loss": 0.1882,
+      "epoch": 3.19,
+      "learning_rate": 2.271191693785836e-05,
+      "loss": 0.2971,
       "step": 6860
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 1.846653671215075e-05,
-      "loss": 0.3425,
+      "epoch": 3.19,
+      "learning_rate": 2.2696420269642025e-05,
+      "loss": 0.393,
       "step": 6870
     },
     {
-      "epoch": 4.47,
-      "learning_rate": 1.8444877626164177e-05,
-      "loss": 0.2234,
+      "epoch": 3.2,
+      "learning_rate": 2.2680923601425693e-05,
+      "loss": 0.1365,
       "step": 6880
     },
     {
-      "epoch": 4.48,
-      "learning_rate": 1.8423218540177606e-05,
-      "loss": 0.0481,
+      "epoch": 3.2,
+      "learning_rate": 2.266542693320936e-05,
+      "loss": 0.1711,
       "step": 6890
     },
     {
-      "epoch": 4.48,
-      "learning_rate": 1.840155945419103e-05,
-      "loss": 0.0574,
+      "epoch": 3.21,
+      "learning_rate": 2.2649930264993028e-05,
+      "loss": 0.1392,
       "step": 6900
     },
     {
-      "epoch": 4.49,
-      "learning_rate": 1.837990036820446e-05,
-      "loss": 0.0049,
+      "epoch": 3.21,
+      "learning_rate": 2.2634433596776692e-05,
+      "loss": 0.3548,
       "step": 6910
     },
     {
-      "epoch": 4.49,
-      "learning_rate": 1.8358241282217892e-05,
-      "loss": 0.0997,
+      "epoch": 3.22,
+      "learning_rate": 2.261893692856036e-05,
+      "loss": 0.2115,
       "step": 6920
     },
     {
-      "epoch": 4.5,
-      "learning_rate": 1.833658219623132e-05,
-      "loss": 0.3253,
+      "epoch": 3.22,
+      "learning_rate": 2.2603440260344027e-05,
+      "loss": 0.1837,
       "step": 6930
     },
     {
-      "epoch": 4.51,
-      "learning_rate": 1.831492311024475e-05,
-      "loss": 0.1426,
+      "epoch": 3.23,
+      "learning_rate": 2.2587943592127694e-05,
+      "loss": 0.3216,
       "step": 6940
     },
     {
-      "epoch": 4.51,
-      "learning_rate": 1.8293264024258178e-05,
-      "loss": 0.0564,
+      "epoch": 3.23,
+      "learning_rate": 2.257244692391136e-05,
+      "loss": 0.2356,
       "step": 6950
     },
     {
-      "epoch": 4.52,
-      "learning_rate": 1.8271604938271607e-05,
-      "loss": 0.3849,
+      "epoch": 3.24,
+      "learning_rate": 2.2556950255695026e-05,
+      "loss": 0.3519,
       "step": 6960
     },
     {
-      "epoch": 4.53,
-      "learning_rate": 1.8249945852285035e-05,
-      "loss": 0.0362,
+      "epoch": 3.24,
+      "learning_rate": 2.2541453587478693e-05,
+      "loss": 0.2192,
       "step": 6970
     },
     {
-      "epoch": 4.53,
-      "learning_rate": 1.822828676629846e-05,
-      "loss": 0.1942,
+      "epoch": 3.25,
+      "learning_rate": 2.252595691926236e-05,
+      "loss": 0.2136,
       "step": 6980
     },
     {
-      "epoch": 4.54,
-      "learning_rate": 1.820662768031189e-05,
-      "loss": 0.2048,
+      "epoch": 3.25,
+      "learning_rate": 2.2510460251046025e-05,
+      "loss": 0.1826,
       "step": 6990
     },
     {
-      "epoch": 4.55,
-      "learning_rate": 1.8184968594325318e-05,
-      "loss": 0.1466,
+      "epoch": 3.25,
+      "learning_rate": 2.2494963582829693e-05,
+      "loss": 0.3629,
       "step": 7000
     },
     {
-      "epoch": 4.55,
-      "learning_rate": 1.816330950833875e-05,
-      "loss": 0.2379,
+      "epoch": 3.26,
+      "learning_rate": 2.247946691461336e-05,
+      "loss": 0.2271,
       "step": 7010
     },
     {
-      "epoch": 4.56,
-      "learning_rate": 1.814165042235218e-05,
-      "loss": 0.1948,
+      "epoch": 3.26,
+      "learning_rate": 2.2463970246397027e-05,
+      "loss": 0.0843,
       "step": 7020
     },
     {
-      "epoch": 4.57,
-      "learning_rate": 1.8119991336365607e-05,
-      "loss": 0.0612,
+      "epoch": 3.27,
+      "learning_rate": 2.244847357818069e-05,
+      "loss": 0.2753,
       "step": 7030
     },
     {
-      "epoch": 4.57,
-      "learning_rate": 1.8098332250379036e-05,
-      "loss": 0.1641,
+      "epoch": 3.27,
+      "learning_rate": 2.243297690996436e-05,
+      "loss": 0.4178,
       "step": 7040
     },
     {
-      "epoch": 4.58,
-      "learning_rate": 1.8076673164392465e-05,
-      "loss": 0.0618,
+      "epoch": 3.28,
+      "learning_rate": 2.2417480241748027e-05,
+      "loss": 0.0916,
       "step": 7050
     },
     {
-      "epoch": 4.59,
-      "learning_rate": 1.805501407840589e-05,
-      "loss": 0.1693,
+      "epoch": 3.28,
+      "learning_rate": 2.2401983573531694e-05,
+      "loss": 0.169,
       "step": 7060
     },
     {
-      "epoch": 4.59,
-      "learning_rate": 1.803335499241932e-05,
-      "loss": 0.134,
+      "epoch": 3.29,
+      "learning_rate": 2.2386486905315358e-05,
+      "loss": 0.2955,
       "step": 7070
     },
     {
-      "epoch": 4.6,
-      "learning_rate": 1.8011695906432747e-05,
-      "loss": 0.3439,
+      "epoch": 3.29,
+      "learning_rate": 2.2370990237099022e-05,
+      "loss": 0.1452,
       "step": 7080
     },
     {
-      "epoch": 4.61,
-      "learning_rate": 1.7990036820446176e-05,
-      "loss": 0.259,
+      "epoch": 3.3,
+      "learning_rate": 2.235549356888269e-05,
+      "loss": 0.2407,
       "step": 7090
     },
     {
-      "epoch": 4.61,
-      "learning_rate": 1.7968377734459608e-05,
-      "loss": 0.114,
+      "epoch": 3.3,
+      "learning_rate": 2.2339996900666357e-05,
+      "loss": 0.1063,
       "step": 7100
     },
     {
-      "epoch": 4.62,
-      "learning_rate": 1.7946718648473037e-05,
-      "loss": 0.3295,
+      "epoch": 3.31,
+      "learning_rate": 2.232450023245002e-05,
+      "loss": 0.3766,
       "step": 7110
     },
     {
-      "epoch": 4.62,
-      "learning_rate": 1.7925059562486466e-05,
-      "loss": 0.259,
+      "epoch": 3.31,
+      "learning_rate": 2.230900356423369e-05,
+      "loss": 0.4121,
       "step": 7120
     },
     {
-      "epoch": 4.63,
-      "learning_rate": 1.7903400476499894e-05,
-      "loss": 0.3889,
+      "epoch": 3.31,
+      "learning_rate": 2.2293506896017356e-05,
+      "loss": 0.1772,
       "step": 7130
     },
     {
-      "epoch": 4.64,
-      "learning_rate": 1.788174139051332e-05,
-      "loss": 0.1208,
+      "epoch": 3.32,
+      "learning_rate": 2.2278010227801024e-05,
+      "loss": 0.1767,
       "step": 7140
     },
     {
-      "epoch": 4.64,
-      "learning_rate": 1.7860082304526748e-05,
-      "loss": 0.1495,
+      "epoch": 3.32,
+      "learning_rate": 2.2262513559584688e-05,
+      "loss": 0.3964,
       "step": 7150
     },
     {
-      "epoch": 4.65,
-      "learning_rate": 1.7838423218540177e-05,
-      "loss": 0.2276,
+      "epoch": 3.33,
+      "learning_rate": 2.2247016891368355e-05,
+      "loss": 0.2643,
       "step": 7160
     },
     {
-      "epoch": 4.66,
-      "learning_rate": 1.7816764132553606e-05,
-      "loss": 0.4401,
+      "epoch": 3.33,
+      "learning_rate": 2.2231520223152023e-05,
+      "loss": 0.3653,
       "step": 7170
     },
     {
-      "epoch": 4.66,
-      "learning_rate": 1.7795105046567034e-05,
-      "loss": 0.18,
+      "epoch": 3.34,
+      "learning_rate": 2.221602355493569e-05,
+      "loss": 0.1428,
       "step": 7180
     },
     {
-      "epoch": 4.67,
-      "learning_rate": 1.7773445960580466e-05,
-      "loss": 0.1499,
+      "epoch": 3.34,
+      "learning_rate": 2.2200526886719354e-05,
+      "loss": 0.4344,
       "step": 7190
     },
     {
-      "epoch": 4.68,
-      "learning_rate": 1.7751786874593895e-05,
-      "loss": 0.1996,
+      "epoch": 3.35,
+      "learning_rate": 2.2185030218503022e-05,
+      "loss": 0.2614,
       "step": 7200
     },
     {
-      "epoch": 4.68,
-      "learning_rate": 1.7730127788607324e-05,
-      "loss": 0.3012,
+      "epoch": 3.35,
+      "learning_rate": 2.216953355028669e-05,
+      "loss": 0.3546,
       "step": 7210
     },
     {
-      "epoch": 4.69,
-      "learning_rate": 1.770846870262075e-05,
-      "loss": 0.1885,
+      "epoch": 3.36,
+      "learning_rate": 2.2154036882070357e-05,
+      "loss": 0.295,
       "step": 7220
     },
     {
-      "epoch": 4.7,
-      "learning_rate": 1.7686809616634178e-05,
-      "loss": 0.119,
+      "epoch": 3.36,
+      "learning_rate": 2.213854021385402e-05,
+      "loss": 0.1342,
       "step": 7230
     },
     {
-      "epoch": 4.7,
-      "learning_rate": 1.7665150530647606e-05,
-      "loss": 0.2105,
+      "epoch": 3.37,
+      "learning_rate": 2.212304354563769e-05,
+      "loss": 0.2365,
       "step": 7240
     },
     {
-      "epoch": 4.71,
-      "learning_rate": 1.7643491444661035e-05,
-      "loss": 0.1406,
+      "epoch": 3.37,
+      "learning_rate": 2.2107546877421356e-05,
+      "loss": 0.0508,
       "step": 7250
     },
     {
-      "epoch": 4.72,
-      "learning_rate": 1.7621832358674464e-05,
-      "loss": 0.358,
+      "epoch": 3.38,
+      "learning_rate": 2.2092050209205023e-05,
+      "loss": 0.1253,
       "step": 7260
     },
     {
-      "epoch": 4.72,
-      "learning_rate": 1.7600173272687892e-05,
-      "loss": 0.1969,
+      "epoch": 3.38,
+      "learning_rate": 2.2076553540988687e-05,
+      "loss": 0.1935,
       "step": 7270
     },
     {
-      "epoch": 4.73,
-      "learning_rate": 1.7578514186701325e-05,
-      "loss": 0.0513,
+      "epoch": 3.38,
+      "learning_rate": 2.2061056872772355e-05,
+      "loss": 0.1225,
       "step": 7280
     },
     {
-      "epoch": 4.74,
-      "learning_rate": 1.755685510071475e-05,
-      "loss": 0.1709,
+      "epoch": 3.39,
+      "learning_rate": 2.2045560204556022e-05,
+      "loss": 0.3626,
       "step": 7290
     },
     {
-      "epoch": 4.74,
-      "learning_rate": 1.753519601472818e-05,
-      "loss": 0.316,
+      "epoch": 3.39,
+      "learning_rate": 2.203006353633969e-05,
+      "loss": 0.1439,
       "step": 7300
     },
     {
-      "epoch": 4.75,
-      "learning_rate": 1.7513536928741607e-05,
-      "loss": 0.0858,
+      "epoch": 3.4,
+      "learning_rate": 2.2014566868123354e-05,
+      "loss": 0.2237,
       "step": 7310
     },
     {
-      "epoch": 4.75,
-      "learning_rate": 1.7491877842755036e-05,
-      "loss": 0.1269,
+      "epoch": 3.4,
+      "learning_rate": 2.199907019990702e-05,
+      "loss": 0.2407,
       "step": 7320
     },
     {
-      "epoch": 4.76,
-      "learning_rate": 1.7470218756768465e-05,
-      "loss": 0.1272,
+      "epoch": 3.41,
+      "learning_rate": 2.1983573531690686e-05,
+      "loss": 0.2577,
       "step": 7330
     },
     {
-      "epoch": 4.77,
-      "learning_rate": 1.7448559670781893e-05,
-      "loss": 0.2098,
+      "epoch": 3.41,
+      "learning_rate": 2.1968076863474353e-05,
+      "loss": 0.4271,
       "step": 7340
     },
     {
-      "epoch": 4.77,
-      "learning_rate": 1.7426900584795322e-05,
-      "loss": 0.1316,
+      "epoch": 3.42,
+      "learning_rate": 2.1952580195258017e-05,
+      "loss": 0.3962,
       "step": 7350
     },
     {
-      "epoch": 4.78,
-      "learning_rate": 1.740524149880875e-05,
-      "loss": 0.1895,
+      "epoch": 3.42,
+      "learning_rate": 2.1937083527041685e-05,
+      "loss": 0.1657,
       "step": 7360
     },
     {
-      "epoch": 4.79,
-      "learning_rate": 1.738358241282218e-05,
-      "loss": 0.1917,
+      "epoch": 3.43,
+      "learning_rate": 2.1921586858825352e-05,
+      "loss": 0.1609,
       "step": 7370
     },
     {
-      "epoch": 4.79,
-      "learning_rate": 1.7361923326835608e-05,
-      "loss": 0.2667,
+      "epoch": 3.43,
+      "learning_rate": 2.190609019060902e-05,
+      "loss": 0.3079,
       "step": 7380
     },
     {
-      "epoch": 4.8,
-      "learning_rate": 1.7340264240849037e-05,
-      "loss": 0.0624,
+      "epoch": 3.44,
+      "learning_rate": 2.1890593522392684e-05,
+      "loss": 0.1627,
       "step": 7390
     },
     {
-      "epoch": 4.81,
-      "learning_rate": 1.7318605154862465e-05,
-      "loss": 0.1343,
+      "epoch": 3.44,
+      "learning_rate": 2.187509685417635e-05,
+      "loss": 0.1931,
       "step": 7400
     },
     {
-      "epoch": 4.81,
-      "learning_rate": 1.7296946068875894e-05,
-      "loss": 0.1578,
+      "epoch": 3.44,
+      "learning_rate": 2.185960018596002e-05,
+      "loss": 0.376,
       "step": 7410
     },
     {
-      "epoch": 4.82,
-      "learning_rate": 1.7275286982889323e-05,
-      "loss": 0.0942,
+      "epoch": 3.45,
+      "learning_rate": 2.1844103517743686e-05,
+      "loss": 0.3854,
       "step": 7420
     },
     {
-      "epoch": 4.83,
-      "learning_rate": 1.725362789690275e-05,
-      "loss": 0.1821,
+      "epoch": 3.45,
+      "learning_rate": 2.182860684952735e-05,
+      "loss": 0.6919,
       "step": 7430
     },
     {
-      "epoch": 4.83,
-      "learning_rate": 1.723196881091618e-05,
-      "loss": 0.1143,
+      "epoch": 3.46,
+      "learning_rate": 2.1813110181311018e-05,
+      "loss": 0.4219,
       "step": 7440
     },
     {
-      "epoch": 4.84,
-      "learning_rate": 1.7210309724929605e-05,
-      "loss": 0.2072,
+      "epoch": 3.46,
+      "learning_rate": 2.1797613513094685e-05,
+      "loss": 0.2413,
       "step": 7450
     },
     {
-      "epoch": 4.85,
-      "learning_rate": 1.7188650638943038e-05,
-      "loss": 0.3436,
+      "epoch": 3.47,
+      "learning_rate": 2.1782116844878353e-05,
+      "loss": 0.1533,
       "step": 7460
     },
     {
-      "epoch": 4.85,
-      "learning_rate": 1.7166991552956466e-05,
-      "loss": 0.2031,
+      "epoch": 3.47,
+      "learning_rate": 2.176662017666202e-05,
+      "loss": 0.0551,
       "step": 7470
     },
     {
-      "epoch": 4.86,
-      "learning_rate": 1.7145332466969895e-05,
-      "loss": 0.1608,
+      "epoch": 3.48,
+      "learning_rate": 2.1751123508445684e-05,
+      "loss": 0.1819,
       "step": 7480
     },
     {
-      "epoch": 4.87,
-      "learning_rate": 1.7123673380983324e-05,
-      "loss": 0.1898,
+      "epoch": 3.48,
+      "learning_rate": 2.1735626840229352e-05,
+      "loss": 0.433,
       "step": 7490
     },
     {
-      "epoch": 4.87,
-      "learning_rate": 1.7102014294996752e-05,
-      "loss": 0.2166,
+      "epoch": 3.49,
+      "learning_rate": 2.172013017201302e-05,
+      "loss": 0.3456,
       "step": 7500
     },
     {
-      "epoch": 4.88,
-      "learning_rate": 1.708035520901018e-05,
-      "loss": 0.1145,
+      "epoch": 3.49,
+      "learning_rate": 2.1704633503796687e-05,
+      "loss": 0.0656,
       "step": 7510
     },
     {
-      "epoch": 4.88,
-      "learning_rate": 1.705869612302361e-05,
-      "loss": 0.3105,
+      "epoch": 3.5,
+      "learning_rate": 2.168913683558035e-05,
+      "loss": 0.2767,
       "step": 7520
     },
     {
-      "epoch": 4.89,
-      "learning_rate": 1.7037037037037035e-05,
-      "loss": 0.2545,
+      "epoch": 3.5,
+      "learning_rate": 2.167364016736402e-05,
+      "loss": 0.3476,
       "step": 7530
     },
     {
-      "epoch": 4.9,
-      "learning_rate": 1.7015377951050464e-05,
-      "loss": 0.016,
+      "epoch": 3.51,
+      "learning_rate": 2.1658143499147686e-05,
+      "loss": 0.1522,
       "step": 7540
     },
     {
-      "epoch": 4.9,
-      "learning_rate": 1.6993718865063896e-05,
-      "loss": 0.3899,
+      "epoch": 3.51,
+      "learning_rate": 2.1642646830931353e-05,
+      "loss": 0.3101,
       "step": 7550
     },
     {
-      "epoch": 4.91,
-      "learning_rate": 1.6972059779077324e-05,
-      "loss": 0.1974,
+      "epoch": 3.51,
+      "learning_rate": 2.1627150162715017e-05,
+      "loss": 0.1511,
       "step": 7560
     },
     {
-      "epoch": 4.92,
-      "learning_rate": 1.6950400693090753e-05,
-      "loss": 0.0903,
+      "epoch": 3.52,
+      "learning_rate": 2.1611653494498685e-05,
+      "loss": 0.3927,
       "step": 7570
     },
     {
-      "epoch": 4.92,
-      "learning_rate": 1.6928741607104182e-05,
-      "loss": 0.1623,
+      "epoch": 3.52,
+      "learning_rate": 2.159615682628235e-05,
+      "loss": 0.1585,
       "step": 7580
     },
     {
-      "epoch": 4.93,
-      "learning_rate": 1.690708252111761e-05,
-      "loss": 0.2203,
+      "epoch": 3.53,
+      "learning_rate": 2.1580660158066016e-05,
+      "loss": 0.2809,
       "step": 7590
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 1.688542343513104e-05,
-      "loss": 0.2525,
+      "epoch": 3.53,
+      "learning_rate": 2.156516348984968e-05,
+      "loss": 0.2061,
       "step": 7600
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 1.6863764349144464e-05,
-      "loss": 0.3311,
+      "epoch": 3.54,
+      "learning_rate": 2.1549666821633348e-05,
+      "loss": 0.1537,
       "step": 7610
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 1.6842105263157893e-05,
-      "loss": 0.0228,
+      "epoch": 3.54,
+      "learning_rate": 2.1534170153417016e-05,
+      "loss": 0.1988,
       "step": 7620
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 1.6820446177171322e-05,
-      "loss": 0.2533,
+      "epoch": 3.55,
+      "learning_rate": 2.1518673485200683e-05,
+      "loss": 0.2971,
       "step": 7630
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 1.6798787091184754e-05,
-      "loss": 0.1275,
+      "epoch": 3.55,
+      "learning_rate": 2.1503176816984347e-05,
+      "loss": 0.0679,
       "step": 7640
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 1.6777128005198183e-05,
-      "loss": 0.0841,
+      "epoch": 3.56,
+      "learning_rate": 2.1487680148768015e-05,
+      "loss": 0.2485,
       "step": 7650
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.675546891921161e-05,
-      "loss": 0.2046,
+      "epoch": 3.56,
+      "learning_rate": 2.1472183480551682e-05,
+      "loss": 0.2727,
       "step": 7660
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 1.673380983322504e-05,
-      "loss": 0.2394,
+      "epoch": 3.57,
+      "learning_rate": 2.145668681233535e-05,
+      "loss": 0.1308,
       "step": 7670
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 1.671215074723847e-05,
-      "loss": 0.1007,
+      "epoch": 3.57,
+      "learning_rate": 2.1441190144119014e-05,
+      "loss": 0.2575,
       "step": 7680
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 1.6690491661251894e-05,
-      "loss": 0.0272,
+      "epoch": 3.58,
+      "learning_rate": 2.142569347590268e-05,
+      "loss": 0.371,
       "step": 7690
     },
     {
-      "epoch": 5.0,
-      "eval_accuracy": 0.9829806807727691,
-      "eval_loss": 0.08533850312232971,
-      "eval_runtime": 12.8473,
-      "eval_samples_per_second": 169.218,
-      "eval_steps_per_second": 42.343,
-      "step": 7697
-    },
-    {
-      "epoch": 5.0,
-      "learning_rate": 1.6668832575265323e-05,
-      "loss": 0.1804,
+      "epoch": 3.58,
+      "learning_rate": 2.141019680768635e-05,
+      "loss": 0.2831,
       "step": 7700
     },
     {
-      "epoch": 5.01,
-      "learning_rate": 1.664717348927875e-05,
-      "loss": 0.0881,
+      "epoch": 3.58,
+      "learning_rate": 2.1394700139470016e-05,
+      "loss": 0.1156,
       "step": 7710
     },
     {
-      "epoch": 5.01,
-      "learning_rate": 1.662551440329218e-05,
-      "loss": 0.3207,
+      "epoch": 3.59,
+      "learning_rate": 2.137920347125368e-05,
+      "loss": 0.2441,
       "step": 7720
     },
     {
-      "epoch": 5.02,
-      "learning_rate": 1.660385531730561e-05,
-      "loss": 0.046,
+      "epoch": 3.59,
+      "learning_rate": 2.1363706803037348e-05,
+      "loss": 0.351,
       "step": 7730
     },
     {
-      "epoch": 5.03,
-      "learning_rate": 1.658219623131904e-05,
-      "loss": 0.0632,
+      "epoch": 3.6,
+      "learning_rate": 2.1348210134821015e-05,
+      "loss": 0.2607,
       "step": 7740
     },
     {
-      "epoch": 5.03,
-      "learning_rate": 1.656053714533247e-05,
-      "loss": 0.492,
+      "epoch": 3.6,
+      "learning_rate": 2.1332713466604683e-05,
+      "loss": 0.2863,
       "step": 7750
     },
     {
-      "epoch": 5.04,
-      "learning_rate": 1.6538878059345898e-05,
-      "loss": 0.1749,
+      "epoch": 3.61,
+      "learning_rate": 2.1317216798388347e-05,
+      "loss": 0.3064,
       "step": 7760
     },
     {
-      "epoch": 5.05,
-      "learning_rate": 1.6517218973359323e-05,
-      "loss": 0.0785,
+      "epoch": 3.61,
+      "learning_rate": 2.1301720130172014e-05,
+      "loss": 0.3971,
       "step": 7770
     },
     {
-      "epoch": 5.05,
-      "learning_rate": 1.6495559887372752e-05,
-      "loss": 0.0355,
+      "epoch": 3.62,
+      "learning_rate": 2.1286223461955682e-05,
+      "loss": 0.2833,
       "step": 7780
     },
     {
-      "epoch": 5.06,
-      "learning_rate": 1.647390080138618e-05,
-      "loss": 0.2957,
+      "epoch": 3.62,
+      "learning_rate": 2.127072679373935e-05,
+      "loss": 0.1994,
       "step": 7790
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 1.645224171539961e-05,
-      "loss": 0.3177,
+      "epoch": 3.63,
+      "learning_rate": 2.1255230125523013e-05,
+      "loss": 0.3087,
       "step": 7800
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 1.6430582629413038e-05,
-      "loss": 0.1878,
+      "epoch": 3.63,
+      "learning_rate": 2.123973345730668e-05,
+      "loss": 0.3272,
       "step": 7810
     },
     {
-      "epoch": 5.08,
-      "learning_rate": 1.6408923543426467e-05,
-      "loss": 0.1447,
+      "epoch": 3.64,
+      "learning_rate": 2.1224236789090345e-05,
+      "loss": 0.1855,
       "step": 7820
     },
     {
-      "epoch": 5.09,
-      "learning_rate": 1.63872644574399e-05,
-      "loss": 0.2844,
+      "epoch": 3.64,
+      "learning_rate": 2.1208740120874012e-05,
+      "loss": 0.1629,
       "step": 7830
     },
     {
-      "epoch": 5.09,
-      "learning_rate": 1.6365605371453328e-05,
-      "loss": 0.0256,
+      "epoch": 3.64,
+      "learning_rate": 2.1193243452657676e-05,
+      "loss": 0.2977,
       "step": 7840
     },
     {
-      "epoch": 5.1,
-      "learning_rate": 1.6343946285466753e-05,
-      "loss": 0.0758,
+      "epoch": 3.65,
+      "learning_rate": 2.1177746784441344e-05,
+      "loss": 0.2975,
       "step": 7850
     },
     {
-      "epoch": 5.11,
-      "learning_rate": 1.632228719948018e-05,
-      "loss": 0.1041,
+      "epoch": 3.65,
+      "learning_rate": 2.116225011622501e-05,
+      "loss": 0.1847,
       "step": 7860
     },
     {
-      "epoch": 5.11,
-      "learning_rate": 1.630062811349361e-05,
-      "loss": 0.1667,
+      "epoch": 3.66,
+      "learning_rate": 2.114675344800868e-05,
+      "loss": 0.2838,
       "step": 7870
     },
     {
-      "epoch": 5.12,
-      "learning_rate": 1.627896902750704e-05,
-      "loss": 0.2194,
+      "epoch": 3.66,
+      "learning_rate": 2.1131256779792343e-05,
+      "loss": 0.2486,
       "step": 7880
     },
     {
-      "epoch": 5.13,
-      "learning_rate": 1.6257309941520468e-05,
-      "loss": 0.2497,
+      "epoch": 3.67,
+      "learning_rate": 2.111576011157601e-05,
+      "loss": 0.2301,
       "step": 7890
     },
     {
-      "epoch": 5.13,
-      "learning_rate": 1.6235650855533896e-05,
-      "loss": 0.2496,
+      "epoch": 3.67,
+      "learning_rate": 2.1100263443359678e-05,
+      "loss": 0.3353,
       "step": 7900
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 1.6213991769547325e-05,
-      "loss": 0.2154,
+      "epoch": 3.68,
+      "learning_rate": 2.1084766775143345e-05,
+      "loss": 0.2155,
       "step": 7910
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 1.6192332683560757e-05,
-      "loss": 0.219,
+      "epoch": 3.68,
+      "learning_rate": 2.106927010692701e-05,
+      "loss": 0.0853,
       "step": 7920
     },
     {
-      "epoch": 5.15,
-      "learning_rate": 1.6170673597574182e-05,
-      "loss": 0.0798,
+      "epoch": 3.69,
+      "learning_rate": 2.1053773438710677e-05,
+      "loss": 0.279,
       "step": 7930
     },
     {
-      "epoch": 5.16,
-      "learning_rate": 1.614901451158761e-05,
-      "loss": 0.2998,
+      "epoch": 3.69,
+      "learning_rate": 2.1038276770494345e-05,
+      "loss": 0.2432,
       "step": 7940
     },
     {
-      "epoch": 5.16,
-      "learning_rate": 1.612735542560104e-05,
-      "loss": 0.0871,
+      "epoch": 3.7,
+      "learning_rate": 2.1022780102278012e-05,
+      "loss": 0.2808,
       "step": 7950
     },
     {
-      "epoch": 5.17,
-      "learning_rate": 1.610569633961447e-05,
-      "loss": 0.1863,
+      "epoch": 3.7,
+      "learning_rate": 2.1007283434061676e-05,
+      "loss": 0.4477,
       "step": 7960
     },
     {
-      "epoch": 5.18,
-      "learning_rate": 1.6084037253627897e-05,
-      "loss": 0.1953,
+      "epoch": 3.71,
+      "learning_rate": 2.0991786765845344e-05,
+      "loss": 0.2767,
       "step": 7970
     },
     {
-      "epoch": 5.18,
-      "learning_rate": 1.6062378167641326e-05,
-      "loss": 0.2771,
+      "epoch": 3.71,
+      "learning_rate": 2.097629009762901e-05,
+      "loss": 0.2519,
       "step": 7980
     },
     {
-      "epoch": 5.19,
-      "learning_rate": 1.6040719081654754e-05,
-      "loss": 0.1713,
+      "epoch": 3.71,
+      "learning_rate": 2.096079342941268e-05,
+      "loss": 0.1456,
       "step": 7990
     },
     {
-      "epoch": 5.2,
-      "learning_rate": 1.602122590426684e-05,
-      "loss": 0.4941,
+      "epoch": 3.72,
+      "learning_rate": 2.0945296761196343e-05,
+      "loss": 0.081,
       "step": 8000
     },
     {
-      "epoch": 5.2,
-      "learning_rate": 1.599956681828027e-05,
-      "loss": 0.0853,
+      "epoch": 3.72,
+      "learning_rate": 2.092980009298001e-05,
+      "loss": 0.2688,
       "step": 8010
     },
     {
-      "epoch": 5.21,
-      "learning_rate": 1.5977907732293696e-05,
-      "loss": 0.1868,
+      "epoch": 3.73,
+      "learning_rate": 2.0914303424763678e-05,
+      "loss": 0.4485,
       "step": 8020
     },
     {
-      "epoch": 5.22,
-      "learning_rate": 1.5956248646307124e-05,
-      "loss": 0.2171,
+      "epoch": 3.73,
+      "learning_rate": 2.0898806756547345e-05,
+      "loss": 0.1985,
       "step": 8030
     },
     {
-      "epoch": 5.22,
-      "learning_rate": 1.5934589560320553e-05,
-      "loss": 0.0884,
+      "epoch": 3.74,
+      "learning_rate": 2.088331008833101e-05,
+      "loss": 0.181,
       "step": 8040
     },
     {
-      "epoch": 5.23,
-      "learning_rate": 1.591293047433398e-05,
-      "loss": 0.2246,
+      "epoch": 3.74,
+      "learning_rate": 2.0867813420114677e-05,
+      "loss": 0.3206,
       "step": 8050
     },
     {
-      "epoch": 5.24,
-      "learning_rate": 1.5891271388347414e-05,
-      "loss": 0.1167,
+      "epoch": 3.75,
+      "learning_rate": 2.0852316751898344e-05,
+      "loss": 0.266,
       "step": 8060
     },
     {
-      "epoch": 5.24,
-      "learning_rate": 1.5869612302360842e-05,
-      "loss": 0.2971,
+      "epoch": 3.75,
+      "learning_rate": 2.0836820083682008e-05,
+      "loss": 0.386,
       "step": 8070
     },
     {
-      "epoch": 5.25,
-      "learning_rate": 1.584795321637427e-05,
-      "loss": 0.1353,
+      "epoch": 3.76,
+      "learning_rate": 2.0821323415465672e-05,
+      "loss": 0.4394,
       "step": 8080
     },
     {
-      "epoch": 5.25,
-      "learning_rate": 1.58262941303877e-05,
-      "loss": 0.1755,
+      "epoch": 3.76,
+      "learning_rate": 2.080582674724934e-05,
+      "loss": 0.2419,
       "step": 8090
     },
     {
-      "epoch": 5.26,
-      "learning_rate": 1.5804635044401125e-05,
-      "loss": 0.1807,
+      "epoch": 3.77,
+      "learning_rate": 2.0790330079033007e-05,
+      "loss": 0.3,
       "step": 8100
     },
     {
-      "epoch": 5.27,
-      "learning_rate": 1.5782975958414554e-05,
-      "loss": 0.1284,
+      "epoch": 3.77,
+      "learning_rate": 2.0774833410816675e-05,
+      "loss": 0.4203,
       "step": 8110
     },
     {
-      "epoch": 5.27,
-      "learning_rate": 1.5761316872427982e-05,
-      "loss": 0.113,
+      "epoch": 3.77,
+      "learning_rate": 2.075933674260034e-05,
+      "loss": 0.0996,
       "step": 8120
     },
     {
-      "epoch": 5.28,
-      "learning_rate": 1.573965778644141e-05,
-      "loss": 0.163,
+      "epoch": 3.78,
+      "learning_rate": 2.0743840074384006e-05,
+      "loss": 0.392,
       "step": 8130
     },
     {
-      "epoch": 5.29,
-      "learning_rate": 1.571799870045484e-05,
-      "loss": 0.2105,
+      "epoch": 3.78,
+      "learning_rate": 2.0728343406167674e-05,
+      "loss": 0.0694,
       "step": 8140
     },
     {
-      "epoch": 5.29,
-      "learning_rate": 1.5696339614468272e-05,
-      "loss": 0.2947,
+      "epoch": 3.79,
+      "learning_rate": 2.071284673795134e-05,
+      "loss": 0.27,
       "step": 8150
     },
     {
-      "epoch": 5.3,
-      "learning_rate": 1.56746805284817e-05,
-      "loss": 0.0778,
+      "epoch": 3.79,
+      "learning_rate": 2.0697350069735005e-05,
+      "loss": 0.1992,
       "step": 8160
     },
     {
-      "epoch": 5.31,
-      "learning_rate": 1.565302144249513e-05,
-      "loss": 0.2409,
+      "epoch": 3.8,
+      "learning_rate": 2.0681853401518673e-05,
+      "loss": 0.2648,
       "step": 8170
     },
     {
-      "epoch": 5.31,
-      "learning_rate": 1.5631362356508555e-05,
-      "loss": 0.2679,
+      "epoch": 3.8,
+      "learning_rate": 2.066635673330234e-05,
+      "loss": 0.2639,
       "step": 8180
     },
     {
-      "epoch": 5.32,
-      "learning_rate": 1.5609703270521983e-05,
-      "loss": 0.1457,
+      "epoch": 3.81,
+      "learning_rate": 2.0650860065086008e-05,
+      "loss": 0.1874,
       "step": 8190
     },
     {
-      "epoch": 5.33,
-      "learning_rate": 1.5588044184535412e-05,
-      "loss": 0.2009,
+      "epoch": 3.81,
+      "learning_rate": 2.0635363396869672e-05,
+      "loss": 0.2049,
       "step": 8200
     },
     {
-      "epoch": 5.33,
-      "learning_rate": 1.556638509854884e-05,
-      "loss": 0.1151,
+      "epoch": 3.82,
+      "learning_rate": 2.061986672865334e-05,
+      "loss": 0.1137,
       "step": 8210
     },
     {
-      "epoch": 5.34,
-      "learning_rate": 1.554472601256227e-05,
-      "loss": 0.1441,
+      "epoch": 3.82,
+      "learning_rate": 2.0604370060437007e-05,
+      "loss": 0.2709,
       "step": 8220
     },
     {
-      "epoch": 5.35,
-      "learning_rate": 1.5523066926575698e-05,
-      "loss": 0.043,
+      "epoch": 3.83,
+      "learning_rate": 2.0588873392220674e-05,
+      "loss": 0.0748,
       "step": 8230
     },
     {
-      "epoch": 5.35,
-      "learning_rate": 1.550140784058913e-05,
-      "loss": 0.344,
+      "epoch": 3.83,
+      "learning_rate": 2.057337672400434e-05,
+      "loss": 0.2456,
       "step": 8240
     },
     {
-      "epoch": 5.36,
-      "learning_rate": 1.547974875460256e-05,
-      "loss": 0.1488,
+      "epoch": 3.84,
+      "learning_rate": 2.0557880055788006e-05,
+      "loss": 0.2471,
       "step": 8250
     },
     {
-      "epoch": 5.37,
-      "learning_rate": 1.5458089668615984e-05,
-      "loss": 0.3524,
+      "epoch": 3.84,
+      "learning_rate": 2.0542383387571674e-05,
+      "loss": 0.2883,
       "step": 8260
     },
     {
-      "epoch": 5.37,
-      "learning_rate": 1.5436430582629413e-05,
-      "loss": 0.1053,
+      "epoch": 3.84,
+      "learning_rate": 2.052688671935534e-05,
+      "loss": 0.1922,
       "step": 8270
     },
     {
-      "epoch": 5.38,
-      "learning_rate": 1.541477149664284e-05,
-      "loss": 0.3589,
+      "epoch": 3.85,
+      "learning_rate": 2.0511390051139005e-05,
+      "loss": 0.2462,
       "step": 8280
     },
     {
-      "epoch": 5.38,
-      "learning_rate": 1.539311241065627e-05,
-      "loss": 0.1829,
+      "epoch": 3.85,
+      "learning_rate": 2.0495893382922673e-05,
+      "loss": 0.4388,
       "step": 8290
     },
     {
-      "epoch": 5.39,
-      "learning_rate": 1.53714533246697e-05,
-      "loss": 0.2131,
+      "epoch": 3.86,
+      "learning_rate": 2.048039671470634e-05,
+      "loss": 0.146,
       "step": 8300
     },
     {
-      "epoch": 5.4,
-      "learning_rate": 1.5349794238683127e-05,
-      "loss": 0.0702,
+      "epoch": 3.86,
+      "learning_rate": 2.0464900046490008e-05,
+      "loss": 0.3036,
       "step": 8310
     },
     {
-      "epoch": 5.4,
-      "learning_rate": 1.5328135152696556e-05,
-      "loss": 0.0808,
+      "epoch": 3.87,
+      "learning_rate": 2.044940337827367e-05,
+      "loss": 0.1348,
       "step": 8320
     },
     {
-      "epoch": 5.41,
-      "learning_rate": 1.5306476066709988e-05,
-      "loss": 0.184,
+      "epoch": 3.87,
+      "learning_rate": 2.0433906710057336e-05,
+      "loss": 0.1958,
       "step": 8330
     },
     {
-      "epoch": 5.42,
-      "learning_rate": 1.5284816980723414e-05,
-      "loss": 0.3782,
+      "epoch": 3.88,
+      "learning_rate": 2.0418410041841003e-05,
+      "loss": 0.3578,
       "step": 8340
     },
     {
-      "epoch": 5.42,
-      "learning_rate": 1.5263157894736842e-05,
-      "loss": 0.021,
+      "epoch": 3.88,
+      "learning_rate": 2.040291337362467e-05,
+      "loss": 0.2913,
       "step": 8350
     },
     {
-      "epoch": 5.43,
-      "learning_rate": 1.5241498808750271e-05,
-      "loss": 0.1233,
+      "epoch": 3.89,
+      "learning_rate": 2.0387416705408338e-05,
+      "loss": 0.3328,
       "step": 8360
     },
     {
-      "epoch": 5.44,
-      "learning_rate": 1.52198397227637e-05,
-      "loss": 0.2307,
+      "epoch": 3.89,
+      "learning_rate": 2.0371920037192002e-05,
+      "loss": 0.2062,
       "step": 8370
     },
     {
-      "epoch": 5.44,
-      "learning_rate": 1.5198180636777128e-05,
-      "loss": 0.3145,
+      "epoch": 3.9,
+      "learning_rate": 2.035642336897567e-05,
+      "loss": 0.1622,
       "step": 8380
     },
     {
-      "epoch": 5.45,
-      "learning_rate": 1.5176521550790557e-05,
-      "loss": 0.1034,
+      "epoch": 3.9,
+      "learning_rate": 2.0340926700759337e-05,
+      "loss": 0.0684,
       "step": 8390
     },
     {
-      "epoch": 5.46,
-      "learning_rate": 1.5154862464803984e-05,
-      "loss": 0.0123,
+      "epoch": 3.91,
+      "learning_rate": 2.0325430032543005e-05,
+      "loss": 0.1468,
       "step": 8400
     },
     {
-      "epoch": 5.46,
-      "learning_rate": 1.5133203378817413e-05,
-      "loss": 0.1678,
+      "epoch": 3.91,
+      "learning_rate": 2.030993336432667e-05,
+      "loss": 0.2684,
       "step": 8410
     },
     {
-      "epoch": 5.47,
-      "learning_rate": 1.5111544292830845e-05,
-      "loss": 0.1051,
+      "epoch": 3.91,
+      "learning_rate": 2.0294436696110336e-05,
+      "loss": 0.1253,
       "step": 8420
     },
     {
-      "epoch": 5.48,
-      "learning_rate": 1.5089885206844273e-05,
-      "loss": 0.0742,
+      "epoch": 3.92,
+      "learning_rate": 2.0278940027894004e-05,
+      "loss": 0.0528,
       "step": 8430
     },
     {
-      "epoch": 5.48,
-      "learning_rate": 1.50682261208577e-05,
-      "loss": 0.2037,
+      "epoch": 3.92,
+      "learning_rate": 2.026344335967767e-05,
+      "loss": 0.3669,
       "step": 8440
     },
     {
-      "epoch": 5.49,
-      "learning_rate": 1.5046567034871129e-05,
-      "loss": 0.3068,
+      "epoch": 3.93,
+      "learning_rate": 2.0247946691461335e-05,
+      "loss": 0.41,
       "step": 8450
     },
     {
-      "epoch": 5.5,
-      "learning_rate": 1.5024907948884558e-05,
-      "loss": 0.2932,
+      "epoch": 3.93,
+      "learning_rate": 2.0232450023245003e-05,
+      "loss": 0.2348,
       "step": 8460
     },
     {
-      "epoch": 5.5,
-      "learning_rate": 1.5003248862897986e-05,
-      "loss": 0.0899,
+      "epoch": 3.94,
+      "learning_rate": 2.021695335502867e-05,
+      "loss": 0.2706,
       "step": 8470
     },
     {
-      "epoch": 5.51,
-      "learning_rate": 1.4981589776911413e-05,
-      "loss": 0.1688,
+      "epoch": 3.94,
+      "learning_rate": 2.0201456686812338e-05,
+      "loss": 0.3618,
       "step": 8480
     },
     {
-      "epoch": 5.51,
-      "learning_rate": 1.4959930690924844e-05,
-      "loss": 0.2213,
+      "epoch": 3.95,
+      "learning_rate": 2.0185960018596002e-05,
+      "loss": 0.2847,
       "step": 8490
     },
     {
-      "epoch": 5.52,
-      "learning_rate": 1.4938271604938272e-05,
-      "loss": 0.1739,
+      "epoch": 3.95,
+      "learning_rate": 2.017046335037967e-05,
+      "loss": 0.3707,
       "step": 8500
     },
     {
-      "epoch": 5.53,
-      "learning_rate": 1.4916612518951701e-05,
-      "loss": 0.1328,
+      "epoch": 3.96,
+      "learning_rate": 2.0154966682163337e-05,
+      "loss": 0.1629,
       "step": 8510
     },
     {
-      "epoch": 5.53,
-      "learning_rate": 1.4894953432965128e-05,
-      "loss": 0.0799,
+      "epoch": 3.96,
+      "learning_rate": 2.0139470013947004e-05,
+      "loss": 0.1829,
       "step": 8520
     },
     {
-      "epoch": 5.54,
-      "learning_rate": 1.4873294346978557e-05,
-      "loss": 0.0648,
+      "epoch": 3.97,
+      "learning_rate": 2.012397334573067e-05,
+      "loss": 0.0678,
       "step": 8530
     },
     {
-      "epoch": 5.55,
-      "learning_rate": 1.4851635260991987e-05,
-      "loss": 0.1861,
+      "epoch": 3.97,
+      "learning_rate": 2.0108476677514336e-05,
+      "loss": 0.1555,
       "step": 8540
     },
     {
-      "epoch": 5.55,
-      "learning_rate": 1.4829976175005416e-05,
-      "loss": 0.1486,
+      "epoch": 3.97,
+      "learning_rate": 2.0092980009298004e-05,
+      "loss": 0.167,
       "step": 8550
     },
     {
-      "epoch": 5.56,
-      "learning_rate": 1.4808317089018843e-05,
-      "loss": 0.3231,
+      "epoch": 3.98,
+      "learning_rate": 2.007748334108167e-05,
+      "loss": 0.2657,
       "step": 8560
     },
     {
-      "epoch": 5.57,
-      "learning_rate": 1.4786658003032272e-05,
-      "loss": 0.0632,
+      "epoch": 3.98,
+      "learning_rate": 2.0061986672865332e-05,
+      "loss": 0.2729,
       "step": 8570
     },
     {
-      "epoch": 5.57,
-      "learning_rate": 1.4764998917045702e-05,
-      "loss": 0.0212,
+      "epoch": 3.99,
+      "learning_rate": 2.0046490004649e-05,
+      "loss": 0.3074,
       "step": 8580
     },
     {
-      "epoch": 5.58,
-      "learning_rate": 1.474333983105913e-05,
-      "loss": 0.0212,
+      "epoch": 3.99,
+      "learning_rate": 2.0030993336432667e-05,
+      "loss": 0.3672,
       "step": 8590
     },
     {
-      "epoch": 5.59,
-      "learning_rate": 1.4721680745072558e-05,
-      "loss": 0.0259,
+      "epoch": 4.0,
+      "learning_rate": 2.0015496668216334e-05,
+      "loss": 0.0351,
       "step": 8600
     },
     {
-      "epoch": 5.59,
-      "learning_rate": 1.4700021659085986e-05,
-      "loss": 0.2181,
+      "epoch": 4.0,
+      "eval_accuracy": 0.9677313137965097,
+      "eval_loss": 0.14472757279872894,
+      "eval_runtime": 17.3686,
+      "eval_samples_per_second": 174.856,
+      "eval_steps_per_second": 43.757,
+      "step": 8604
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 0.1173,
       "step": 8610
     },
     {
-      "epoch": 5.6,
-      "learning_rate": 1.4678362573099415e-05,
-      "loss": 0.1272,
+      "epoch": 4.01,
+      "learning_rate": 1.9984503331783666e-05,
+      "loss": 0.3015,
       "step": 8620
     },
     {
-      "epoch": 5.61,
-      "learning_rate": 1.4656703487112845e-05,
-      "loss": 0.3194,
+      "epoch": 4.01,
+      "learning_rate": 1.9969006663567333e-05,
+      "loss": 0.3571,
       "step": 8630
     },
     {
-      "epoch": 5.61,
-      "learning_rate": 1.4635044401126272e-05,
-      "loss": 0.1792,
+      "epoch": 4.02,
+      "learning_rate": 1.9953509995351e-05,
+      "loss": 0.2678,
       "step": 8640
     },
     {
-      "epoch": 5.62,
-      "learning_rate": 1.4613385315139701e-05,
-      "loss": 0.1701,
+      "epoch": 4.02,
+      "learning_rate": 1.9938013327134665e-05,
+      "loss": 0.3516,
       "step": 8650
     },
     {
-      "epoch": 5.63,
-      "learning_rate": 1.459172622915313e-05,
-      "loss": 0.2487,
+      "epoch": 4.03,
+      "learning_rate": 1.9922516658918332e-05,
+      "loss": 0.2841,
       "step": 8660
     },
     {
-      "epoch": 5.63,
-      "learning_rate": 1.457006714316656e-05,
-      "loss": 0.0454,
+      "epoch": 4.03,
+      "learning_rate": 1.9907019990702e-05,
+      "loss": 0.2212,
       "step": 8670
     },
     {
-      "epoch": 5.64,
-      "learning_rate": 1.4548408057179987e-05,
-      "loss": 0.1839,
+      "epoch": 4.04,
+      "learning_rate": 1.9891523322485667e-05,
+      "loss": 0.2416,
       "step": 8680
     },
     {
-      "epoch": 5.64,
-      "learning_rate": 1.4526748971193416e-05,
-      "loss": 0.2466,
+      "epoch": 4.04,
+      "learning_rate": 1.987602665426933e-05,
+      "loss": 0.1888,
       "step": 8690
     },
     {
-      "epoch": 5.65,
-      "learning_rate": 1.4505089885206844e-05,
-      "loss": 0.2125,
+      "epoch": 4.04,
+      "learning_rate": 1.9860529986053e-05,
+      "loss": 0.3491,
       "step": 8700
     },
     {
-      "epoch": 5.66,
-      "learning_rate": 1.4483430799220273e-05,
-      "loss": 0.1713,
+      "epoch": 4.05,
+      "learning_rate": 1.9845033317836666e-05,
+      "loss": 0.1847,
       "step": 8710
     },
     {
-      "epoch": 5.66,
-      "learning_rate": 1.4461771713233702e-05,
-      "loss": 0.1824,
+      "epoch": 4.05,
+      "learning_rate": 1.9829536649620334e-05,
+      "loss": 0.2327,
       "step": 8720
     },
     {
-      "epoch": 5.67,
-      "learning_rate": 1.444011262724713e-05,
-      "loss": 0.1248,
+      "epoch": 4.06,
+      "learning_rate": 1.9814039981403998e-05,
+      "loss": 0.368,
       "step": 8730
     },
     {
-      "epoch": 5.68,
-      "learning_rate": 1.441845354126056e-05,
-      "loss": 0.0942,
+      "epoch": 4.06,
+      "learning_rate": 1.9798543313187665e-05,
+      "loss": 0.1673,
       "step": 8740
     },
     {
-      "epoch": 5.68,
-      "learning_rate": 1.4396794455273988e-05,
-      "loss": 0.2349,
+      "epoch": 4.07,
+      "learning_rate": 1.9783046644971333e-05,
+      "loss": 0.366,
       "step": 8750
     },
     {
-      "epoch": 5.69,
-      "learning_rate": 1.4375135369287417e-05,
-      "loss": 0.1787,
+      "epoch": 4.07,
+      "learning_rate": 1.9767549976755e-05,
+      "loss": 0.2762,
       "step": 8760
     },
     {
-      "epoch": 5.7,
-      "learning_rate": 1.4353476283300845e-05,
-      "loss": 0.3789,
+      "epoch": 4.08,
+      "learning_rate": 1.9752053308538664e-05,
+      "loss": 0.0122,
       "step": 8770
     },
     {
-      "epoch": 5.7,
-      "learning_rate": 1.4331817197314274e-05,
-      "loss": 0.166,
+      "epoch": 4.08,
+      "learning_rate": 1.9736556640322332e-05,
+      "loss": 0.3858,
       "step": 8780
     },
     {
-      "epoch": 5.71,
-      "learning_rate": 1.4310158111327703e-05,
-      "loss": 0.2508,
+      "epoch": 4.09,
+      "learning_rate": 1.9721059972106e-05,
+      "loss": 0.2058,
       "step": 8790
     },
     {
-      "epoch": 5.72,
-      "learning_rate": 1.428849902534113e-05,
-      "loss": 0.3274,
+      "epoch": 4.09,
+      "learning_rate": 1.9705563303889667e-05,
+      "loss": 0.1502,
       "step": 8800
     },
     {
-      "epoch": 5.72,
-      "learning_rate": 1.426683993935456e-05,
-      "loss": 0.0026,
+      "epoch": 4.1,
+      "learning_rate": 1.969006663567333e-05,
+      "loss": 0.3947,
       "step": 8810
     },
     {
-      "epoch": 5.73,
-      "learning_rate": 1.4245180853367989e-05,
-      "loss": 0.3321,
+      "epoch": 4.1,
+      "learning_rate": 1.9674569967456995e-05,
+      "loss": 0.1435,
       "step": 8820
     },
     {
-      "epoch": 5.74,
-      "learning_rate": 1.4223521767381417e-05,
-      "loss": 0.4053,
+      "epoch": 4.11,
+      "learning_rate": 1.9659073299240663e-05,
+      "loss": 0.366,
       "step": 8830
     },
     {
-      "epoch": 5.74,
-      "learning_rate": 1.4201862681394844e-05,
-      "loss": 0.2369,
+      "epoch": 4.11,
+      "learning_rate": 1.964357663102433e-05,
+      "loss": 0.2266,
       "step": 8840
     },
     {
-      "epoch": 5.75,
-      "learning_rate": 1.4180203595408275e-05,
-      "loss": 0.0657,
+      "epoch": 4.11,
+      "learning_rate": 1.9628079962807994e-05,
+      "loss": 0.1253,
       "step": 8850
     },
     {
-      "epoch": 5.76,
-      "learning_rate": 1.4158544509421703e-05,
-      "loss": 0.1151,
+      "epoch": 4.12,
+      "learning_rate": 1.961258329459166e-05,
+      "loss": 0.1908,
       "step": 8860
     },
     {
-      "epoch": 5.76,
-      "learning_rate": 1.4136885423435132e-05,
-      "loss": 0.2244,
+      "epoch": 4.12,
+      "learning_rate": 1.959708662637533e-05,
+      "loss": 0.1677,
       "step": 8870
     },
     {
-      "epoch": 5.77,
-      "learning_rate": 1.4115226337448559e-05,
-      "loss": 0.0992,
+      "epoch": 4.13,
+      "learning_rate": 1.9581589958158997e-05,
+      "loss": 0.3631,
       "step": 8880
     },
     {
-      "epoch": 5.77,
-      "learning_rate": 1.4093567251461988e-05,
-      "loss": 0.0408,
+      "epoch": 4.13,
+      "learning_rate": 1.956609328994266e-05,
+      "loss": 0.2653,
       "step": 8890
     },
     {
-      "epoch": 5.78,
-      "learning_rate": 1.4071908165475418e-05,
-      "loss": 0.1377,
+      "epoch": 4.14,
+      "learning_rate": 1.9550596621726328e-05,
+      "loss": 0.1417,
       "step": 8900
     },
     {
-      "epoch": 5.79,
-      "learning_rate": 1.4050249079488847e-05,
-      "loss": 0.2829,
+      "epoch": 4.14,
+      "learning_rate": 1.9535099953509996e-05,
+      "loss": 0.1626,
       "step": 8910
     },
     {
-      "epoch": 5.79,
-      "learning_rate": 1.4028589993502274e-05,
-      "loss": 0.2788,
+      "epoch": 4.15,
+      "learning_rate": 1.9519603285293663e-05,
+      "loss": 0.269,
       "step": 8920
     },
     {
-      "epoch": 5.8,
-      "learning_rate": 1.4006930907515703e-05,
-      "loss": 0.1138,
+      "epoch": 4.15,
+      "learning_rate": 1.9504106617077327e-05,
+      "loss": 0.4557,
       "step": 8930
     },
     {
-      "epoch": 5.81,
-      "learning_rate": 1.3985271821529133e-05,
-      "loss": 0.0994,
+      "epoch": 4.16,
+      "learning_rate": 1.9488609948860995e-05,
+      "loss": 0.2997,
       "step": 8940
     },
     {
-      "epoch": 5.81,
-      "learning_rate": 1.3963612735542562e-05,
-      "loss": 0.1967,
+      "epoch": 4.16,
+      "learning_rate": 1.9473113280644662e-05,
+      "loss": 0.4433,
       "step": 8950
     },
     {
-      "epoch": 5.82,
-      "learning_rate": 1.3941953649555989e-05,
-      "loss": 0.0688,
+      "epoch": 4.17,
+      "learning_rate": 1.945761661242833e-05,
+      "loss": 0.446,
       "step": 8960
     },
     {
-      "epoch": 5.83,
-      "learning_rate": 1.3920294563569417e-05,
-      "loss": 0.1298,
+      "epoch": 4.17,
+      "learning_rate": 1.9442119944211994e-05,
+      "loss": 0.3395,
       "step": 8970
     },
     {
-      "epoch": 5.83,
-      "learning_rate": 1.3898635477582846e-05,
-      "loss": 0.1009,
+      "epoch": 4.17,
+      "learning_rate": 1.942662327599566e-05,
+      "loss": 0.1092,
       "step": 8980
     },
     {
-      "epoch": 5.84,
-      "learning_rate": 1.3876976391596276e-05,
-      "loss": 0.2893,
+      "epoch": 4.18,
+      "learning_rate": 1.941112660777933e-05,
+      "loss": 0.1988,
       "step": 8990
     },
     {
-      "epoch": 5.85,
-      "learning_rate": 1.3855317305609703e-05,
-      "loss": 0.2114,
+      "epoch": 4.18,
+      "learning_rate": 1.9395629939562996e-05,
+      "loss": 0.0444,
       "step": 9000
     },
     {
-      "epoch": 5.85,
-      "learning_rate": 1.3833658219623132e-05,
-      "loss": 0.306,
+      "epoch": 4.19,
+      "learning_rate": 1.938013327134666e-05,
+      "loss": 0.0706,
       "step": 9010
     },
     {
-      "epoch": 5.86,
-      "learning_rate": 1.381199913363656e-05,
-      "loss": 0.0636,
+      "epoch": 4.19,
+      "learning_rate": 1.9364636603130328e-05,
+      "loss": 0.1558,
       "step": 9020
     },
     {
-      "epoch": 5.87,
-      "learning_rate": 1.379034004764999e-05,
-      "loss": 0.1042,
+      "epoch": 4.2,
+      "learning_rate": 1.9349139934913995e-05,
+      "loss": 0.3305,
       "step": 9030
     },
     {
-      "epoch": 5.87,
-      "learning_rate": 1.3768680961663418e-05,
-      "loss": 0.0434,
+      "epoch": 4.2,
+      "learning_rate": 1.9333643266697663e-05,
+      "loss": 0.3989,
       "step": 9040
     },
     {
-      "epoch": 5.88,
-      "learning_rate": 1.3747021875676847e-05,
-      "loss": 0.2535,
+      "epoch": 4.21,
+      "learning_rate": 1.9318146598481327e-05,
+      "loss": 0.2291,
       "step": 9050
     },
     {
-      "epoch": 5.89,
-      "learning_rate": 1.3725362789690275e-05,
-      "loss": 0.2026,
+      "epoch": 4.21,
+      "learning_rate": 1.9302649930264994e-05,
+      "loss": 0.392,
       "step": 9060
     },
     {
-      "epoch": 5.89,
-      "learning_rate": 1.3703703703703704e-05,
-      "loss": 0.2205,
+      "epoch": 4.22,
+      "learning_rate": 1.928715326204866e-05,
+      "loss": 0.2284,
       "step": 9070
     },
     {
-      "epoch": 5.9,
-      "learning_rate": 1.3682044617717133e-05,
-      "loss": 0.0956,
+      "epoch": 4.22,
+      "learning_rate": 1.9271656593832326e-05,
+      "loss": 0.3651,
       "step": 9080
     },
     {
-      "epoch": 5.9,
-      "learning_rate": 1.3660385531730562e-05,
-      "loss": 0.2372,
+      "epoch": 4.23,
+      "learning_rate": 1.925615992561599e-05,
+      "loss": 0.0411,
       "step": 9090
     },
     {
-      "epoch": 5.91,
-      "learning_rate": 1.363872644574399e-05,
-      "loss": 0.0798,
+      "epoch": 4.23,
+      "learning_rate": 1.9240663257399658e-05,
+      "loss": 0.3452,
       "step": 9100
     },
     {
-      "epoch": 5.92,
-      "learning_rate": 1.3617067359757417e-05,
-      "loss": 0.0851,
+      "epoch": 4.24,
+      "learning_rate": 1.9225166589183325e-05,
+      "loss": 0.1801,
       "step": 9110
     },
     {
-      "epoch": 5.92,
-      "learning_rate": 1.3595408273770848e-05,
-      "loss": 0.1327,
+      "epoch": 4.24,
+      "learning_rate": 1.9209669920966993e-05,
+      "loss": 0.1883,
       "step": 9120
     },
     {
-      "epoch": 5.93,
-      "learning_rate": 1.3573749187784276e-05,
-      "loss": 0.4017,
+      "epoch": 4.24,
+      "learning_rate": 1.9194173252750657e-05,
+      "loss": 0.3017,
       "step": 9130
     },
     {
-      "epoch": 5.94,
-      "learning_rate": 1.3552090101797705e-05,
-      "loss": 0.0811,
+      "epoch": 4.25,
+      "learning_rate": 1.9178676584534324e-05,
+      "loss": 0.1604,
       "step": 9140
     },
     {
-      "epoch": 5.94,
-      "learning_rate": 1.3530431015811132e-05,
-      "loss": 0.4282,
+      "epoch": 4.25,
+      "learning_rate": 1.916317991631799e-05,
+      "loss": 0.4053,
       "step": 9150
     },
     {
-      "epoch": 5.95,
-      "learning_rate": 1.350877192982456e-05,
-      "loss": 0.1739,
+      "epoch": 4.26,
+      "learning_rate": 1.914768324810166e-05,
+      "loss": 0.3137,
       "step": 9160
     },
     {
-      "epoch": 5.96,
-      "learning_rate": 1.3487112843837991e-05,
-      "loss": 0.0615,
+      "epoch": 4.26,
+      "learning_rate": 1.9132186579885323e-05,
+      "loss": 0.2369,
       "step": 9170
     },
     {
-      "epoch": 5.96,
-      "learning_rate": 1.346545375785142e-05,
-      "loss": 0.2251,
+      "epoch": 4.27,
+      "learning_rate": 1.911668991166899e-05,
+      "loss": 0.227,
       "step": 9180
     },
     {
-      "epoch": 5.97,
-      "learning_rate": 1.3443794671864847e-05,
-      "loss": 0.0924,
+      "epoch": 4.27,
+      "learning_rate": 1.9101193243452658e-05,
+      "loss": 0.2225,
       "step": 9190
     },
     {
-      "epoch": 5.98,
-      "learning_rate": 1.3422135585878275e-05,
-      "loss": 0.1031,
+      "epoch": 4.28,
+      "learning_rate": 1.9085696575236326e-05,
+      "loss": 0.0715,
       "step": 9200
     },
     {
-      "epoch": 5.98,
-      "learning_rate": 1.3400476499891706e-05,
-      "loss": 0.1001,
+      "epoch": 4.28,
+      "learning_rate": 1.9070199907019993e-05,
+      "loss": 0.229,
       "step": 9210
     },
     {
-      "epoch": 5.99,
-      "learning_rate": 1.3378817413905134e-05,
-      "loss": 0.2866,
+      "epoch": 4.29,
+      "learning_rate": 1.9054703238803657e-05,
+      "loss": 0.1222,
       "step": 9220
     },
     {
-      "epoch": 6.0,
-      "learning_rate": 1.3357158327918561e-05,
-      "loss": 0.1269,
+      "epoch": 4.29,
+      "learning_rate": 1.9039206570587325e-05,
+      "loss": 0.1746,
       "step": 9230
     },
     {
-      "epoch": 6.0,
-      "eval_accuracy": 0.9806807727690893,
-      "eval_loss": 0.10807611048221588,
-      "eval_runtime": 13.2387,
-      "eval_samples_per_second": 164.215,
-      "eval_steps_per_second": 41.092,
-      "step": 9237
-    },
-    {
-      "epoch": 6.0,
-      "learning_rate": 1.333549924193199e-05,
-      "loss": 0.1506,
+      "epoch": 4.3,
+      "learning_rate": 1.9023709902370992e-05,
+      "loss": 0.2233,
       "step": 9240
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 1.3313840155945419e-05,
-      "loss": 0.3085,
+      "epoch": 4.3,
+      "learning_rate": 1.900821323415466e-05,
+      "loss": 0.1788,
       "step": 9250
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 1.3292181069958849e-05,
-      "loss": 0.1471,
+      "epoch": 4.3,
+      "learning_rate": 1.8992716565938324e-05,
+      "loss": 0.212,
       "step": 9260
     },
     {
-      "epoch": 6.02,
-      "learning_rate": 1.3270521983972276e-05,
-      "loss": 0.0518,
+      "epoch": 4.31,
+      "learning_rate": 1.897721989772199e-05,
+      "loss": 0.4573,
       "step": 9270
     },
     {
-      "epoch": 6.03,
-      "learning_rate": 1.3248862897985705e-05,
-      "loss": 0.1339,
+      "epoch": 4.31,
+      "learning_rate": 1.896172322950566e-05,
+      "loss": 0.2735,
       "step": 9280
     },
     {
-      "epoch": 6.03,
-      "learning_rate": 1.3227203811999134e-05,
-      "loss": 0.0966,
+      "epoch": 4.32,
+      "learning_rate": 1.8946226561289326e-05,
+      "loss": 0.0787,
       "step": 9290
     },
     {
-      "epoch": 6.04,
-      "learning_rate": 1.3205544726012564e-05,
-      "loss": 0.2445,
+      "epoch": 4.32,
+      "learning_rate": 1.893072989307299e-05,
+      "loss": 0.0995,
       "step": 9300
     },
     {
-      "epoch": 6.05,
-      "learning_rate": 1.3183885640025991e-05,
-      "loss": 0.3489,
+      "epoch": 4.33,
+      "learning_rate": 1.8915233224856658e-05,
+      "loss": 0.2497,
       "step": 9310
     },
     {
-      "epoch": 6.05,
-      "learning_rate": 1.316222655403942e-05,
-      "loss": 0.127,
+      "epoch": 4.33,
+      "learning_rate": 1.8899736556640322e-05,
+      "loss": 0.1778,
       "step": 9320
     },
     {
-      "epoch": 6.06,
-      "learning_rate": 1.3140567468052848e-05,
-      "loss": 0.1226,
+      "epoch": 4.34,
+      "learning_rate": 1.888423988842399e-05,
+      "loss": 0.1694,
       "step": 9330
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 1.3118908382066277e-05,
-      "loss": 0.2297,
+      "epoch": 4.34,
+      "learning_rate": 1.8868743220207653e-05,
+      "loss": 0.2727,
       "step": 9340
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 1.3097249296079706e-05,
-      "loss": 0.0035,
+      "epoch": 4.35,
+      "learning_rate": 1.885324655199132e-05,
+      "loss": 0.4632,
       "step": 9350
     },
     {
-      "epoch": 6.08,
-      "learning_rate": 1.3075590210093134e-05,
-      "loss": 0.1268,
+      "epoch": 4.35,
+      "learning_rate": 1.883774988377499e-05,
+      "loss": 0.3306,
       "step": 9360
     },
     {
-      "epoch": 6.09,
-      "learning_rate": 1.3053931124106563e-05,
-      "loss": 0.1522,
+      "epoch": 4.36,
+      "learning_rate": 1.8822253215558656e-05,
+      "loss": 0.3646,
       "step": 9370
     },
     {
-      "epoch": 6.09,
-      "learning_rate": 1.3032272038119992e-05,
-      "loss": 0.193,
+      "epoch": 4.36,
+      "learning_rate": 1.880675654734232e-05,
+      "loss": 0.1806,
       "step": 9380
     },
     {
-      "epoch": 6.1,
-      "learning_rate": 1.301061295213342e-05,
-      "loss": 0.0452,
+      "epoch": 4.37,
+      "learning_rate": 1.8791259879125987e-05,
+      "loss": 0.2507,
       "step": 9390
     },
     {
-      "epoch": 6.11,
-      "learning_rate": 1.2988953866146849e-05,
-      "loss": 0.0601,
+      "epoch": 4.37,
+      "learning_rate": 1.8775763210909655e-05,
+      "loss": 0.2634,
       "step": 9400
     },
     {
-      "epoch": 6.11,
-      "learning_rate": 1.2967294780160278e-05,
-      "loss": 0.1202,
+      "epoch": 4.37,
+      "learning_rate": 1.8760266542693322e-05,
+      "loss": 0.1891,
       "step": 9410
     },
     {
-      "epoch": 6.12,
-      "learning_rate": 1.2945635694173706e-05,
-      "loss": 0.0262,
+      "epoch": 4.38,
+      "learning_rate": 1.8744769874476987e-05,
+      "loss": 0.189,
       "step": 9420
     },
     {
-      "epoch": 6.13,
-      "learning_rate": 1.2923976608187133e-05,
-      "loss": 0.1947,
+      "epoch": 4.38,
+      "learning_rate": 1.8729273206260654e-05,
+      "loss": 0.2814,
       "step": 9430
     },
     {
-      "epoch": 6.13,
-      "learning_rate": 1.2902317522200564e-05,
-      "loss": 0.0498,
+      "epoch": 4.39,
+      "learning_rate": 1.871377653804432e-05,
+      "loss": 0.4847,
       "step": 9440
     },
     {
-      "epoch": 6.14,
-      "learning_rate": 1.2880658436213992e-05,
-      "loss": 0.1465,
+      "epoch": 4.39,
+      "learning_rate": 1.869827986982799e-05,
+      "loss": 0.3275,
       "step": 9450
     },
     {
-      "epoch": 6.14,
-      "learning_rate": 1.2858999350227421e-05,
-      "loss": 0.148,
+      "epoch": 4.4,
+      "learning_rate": 1.8682783201611653e-05,
+      "loss": 0.3991,
       "step": 9460
     },
     {
-      "epoch": 6.15,
-      "learning_rate": 1.2837340264240848e-05,
-      "loss": 0.2404,
+      "epoch": 4.4,
+      "learning_rate": 1.866728653339532e-05,
+      "loss": 0.359,
       "step": 9470
     },
     {
-      "epoch": 6.16,
-      "learning_rate": 1.2815681178254279e-05,
-      "loss": 0.4131,
+      "epoch": 4.41,
+      "learning_rate": 1.8651789865178988e-05,
+      "loss": 0.1361,
       "step": 9480
     },
     {
-      "epoch": 6.16,
-      "learning_rate": 1.2794022092267707e-05,
-      "loss": 0.0116,
+      "epoch": 4.41,
+      "learning_rate": 1.8636293196962656e-05,
+      "loss": 0.4369,
       "step": 9490
     },
     {
-      "epoch": 6.17,
-      "learning_rate": 1.2772363006281136e-05,
-      "loss": 0.2536,
+      "epoch": 4.42,
+      "learning_rate": 1.862079652874632e-05,
+      "loss": 0.1532,
       "step": 9500
     },
     {
-      "epoch": 6.18,
-      "learning_rate": 1.2750703920294563e-05,
-      "loss": 0.275,
+      "epoch": 4.42,
+      "learning_rate": 1.8605299860529987e-05,
+      "loss": 0.1969,
       "step": 9510
     },
     {
-      "epoch": 6.18,
-      "learning_rate": 1.2729044834307992e-05,
-      "loss": 0.1243,
+      "epoch": 4.43,
+      "learning_rate": 1.8589803192313655e-05,
+      "loss": 0.0733,
       "step": 9520
     },
     {
-      "epoch": 6.19,
-      "learning_rate": 1.2707385748321422e-05,
-      "loss": 0.0021,
+      "epoch": 4.43,
+      "learning_rate": 1.8574306524097322e-05,
+      "loss": 0.0648,
       "step": 9530
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 1.268572666233485e-05,
-      "loss": 0.3057,
+      "epoch": 4.44,
+      "learning_rate": 1.8558809855880986e-05,
+      "loss": 0.12,
       "step": 9540
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 1.2664067576348278e-05,
-      "loss": 0.163,
+      "epoch": 4.44,
+      "learning_rate": 1.8543313187664654e-05,
+      "loss": 0.2015,
       "step": 9550
     },
     {
-      "epoch": 6.21,
-      "learning_rate": 1.2642408490361706e-05,
-      "loss": 0.0507,
+      "epoch": 4.44,
+      "learning_rate": 1.852781651944832e-05,
+      "loss": 0.145,
       "step": 9560
     },
     {
-      "epoch": 6.22,
-      "learning_rate": 1.2620749404375137e-05,
-      "loss": 0.2131,
+      "epoch": 4.45,
+      "learning_rate": 1.8512319851231985e-05,
+      "loss": 0.2925,
       "step": 9570
     },
     {
-      "epoch": 6.22,
-      "learning_rate": 1.2599090318388565e-05,
-      "loss": 0.0124,
+      "epoch": 4.45,
+      "learning_rate": 1.849682318301565e-05,
+      "loss": 0.0852,
       "step": 9580
     },
     {
-      "epoch": 6.23,
-      "learning_rate": 1.2577431232401992e-05,
-      "loss": 0.0315,
+      "epoch": 4.46,
+      "learning_rate": 1.8481326514799317e-05,
+      "loss": 0.6667,
       "step": 9590
     },
     {
-      "epoch": 6.24,
-      "learning_rate": 1.2555772146415421e-05,
-      "loss": 0.3318,
+      "epoch": 4.46,
+      "learning_rate": 1.8465829846582984e-05,
+      "loss": 0.3139,
       "step": 9600
     },
     {
-      "epoch": 6.24,
-      "learning_rate": 1.253411306042885e-05,
-      "loss": 0.0796,
+      "epoch": 4.47,
+      "learning_rate": 1.8450333178366652e-05,
+      "loss": 0.1729,
       "step": 9610
     },
     {
-      "epoch": 6.25,
-      "learning_rate": 1.251245397444228e-05,
-      "loss": 0.108,
+      "epoch": 4.47,
+      "learning_rate": 1.8434836510150316e-05,
+      "loss": 0.2929,
       "step": 9620
     },
     {
-      "epoch": 6.26,
-      "learning_rate": 1.2490794888455707e-05,
-      "loss": 0.1974,
+      "epoch": 4.48,
+      "learning_rate": 1.8419339841933983e-05,
+      "loss": 0.0262,
       "step": 9630
     },
     {
-      "epoch": 6.26,
-      "learning_rate": 1.2469135802469136e-05,
-      "loss": 0.2185,
+      "epoch": 4.48,
+      "learning_rate": 1.840384317371765e-05,
+      "loss": 0.1509,
       "step": 9640
     },
     {
-      "epoch": 6.27,
-      "learning_rate": 1.2447476716482564e-05,
-      "loss": 0.1898,
+      "epoch": 4.49,
+      "learning_rate": 1.838834650550132e-05,
+      "loss": 0.1843,
       "step": 9650
     },
     {
-      "epoch": 6.27,
-      "learning_rate": 1.2425817630495995e-05,
-      "loss": 0.0438,
+      "epoch": 4.49,
+      "learning_rate": 1.8372849837284982e-05,
+      "loss": 0.3832,
       "step": 9660
     },
     {
-      "epoch": 6.28,
-      "learning_rate": 1.2404158544509422e-05,
-      "loss": 0.0117,
+      "epoch": 4.5,
+      "learning_rate": 1.835735316906865e-05,
+      "loss": 0.1706,
       "step": 9670
     },
     {
-      "epoch": 6.29,
-      "learning_rate": 1.238249945852285e-05,
-      "loss": 0.2061,
+      "epoch": 4.5,
+      "learning_rate": 1.8341856500852317e-05,
+      "loss": 0.1408,
       "step": 9680
     },
     {
-      "epoch": 6.29,
-      "learning_rate": 1.236084037253628e-05,
-      "loss": 0.0047,
+      "epoch": 4.5,
+      "learning_rate": 1.8326359832635985e-05,
+      "loss": 0.1749,
       "step": 9690
     },
     {
-      "epoch": 6.3,
-      "learning_rate": 1.2339181286549708e-05,
-      "loss": 0.0662,
+      "epoch": 4.51,
+      "learning_rate": 1.831086316441965e-05,
+      "loss": 0.3377,
       "step": 9700
     },
     {
-      "epoch": 6.31,
-      "learning_rate": 1.2317522200563137e-05,
-      "loss": 0.0885,
+      "epoch": 4.51,
+      "learning_rate": 1.8295366496203316e-05,
+      "loss": 0.0584,
       "step": 9710
     },
     {
-      "epoch": 6.31,
-      "learning_rate": 1.2295863114576565e-05,
-      "loss": 0.0855,
+      "epoch": 4.52,
+      "learning_rate": 1.8279869827986984e-05,
+      "loss": 0.0664,
       "step": 9720
     },
     {
-      "epoch": 6.32,
-      "learning_rate": 1.2274204028589994e-05,
-      "loss": 0.1832,
+      "epoch": 4.52,
+      "learning_rate": 1.826437315977065e-05,
+      "loss": 0.2053,
       "step": 9730
     },
     {
-      "epoch": 6.33,
-      "learning_rate": 1.2252544942603423e-05,
-      "loss": 0.0117,
+      "epoch": 4.53,
+      "learning_rate": 1.8248876491554316e-05,
+      "loss": 0.0655,
       "step": 9740
     },
     {
-      "epoch": 6.33,
-      "learning_rate": 1.2230885856616851e-05,
-      "loss": 0.1825,
+      "epoch": 4.53,
+      "learning_rate": 1.8233379823337983e-05,
+      "loss": 0.2331,
       "step": 9750
     },
     {
-      "epoch": 6.34,
-      "learning_rate": 1.220922677063028e-05,
-      "loss": 0.0776,
+      "epoch": 4.54,
+      "learning_rate": 1.821788315512165e-05,
+      "loss": 0.227,
       "step": 9760
     },
     {
-      "epoch": 6.35,
-      "learning_rate": 1.2187567684643709e-05,
-      "loss": 0.276,
+      "epoch": 4.54,
+      "learning_rate": 1.8202386486905318e-05,
+      "loss": 0.2126,
       "step": 9770
     },
     {
-      "epoch": 6.35,
-      "learning_rate": 1.2165908598657137e-05,
-      "loss": 0.0053,
+      "epoch": 4.55,
+      "learning_rate": 1.8186889818688982e-05,
+      "loss": 0.1872,
       "step": 9780
     },
     {
-      "epoch": 6.36,
-      "learning_rate": 1.2144249512670564e-05,
-      "loss": 0.2128,
+      "epoch": 4.55,
+      "learning_rate": 1.817139315047265e-05,
+      "loss": 0.3639,
       "step": 9790
     },
     {
-      "epoch": 6.37,
-      "learning_rate": 1.2122590426683995e-05,
-      "loss": 0.2099,
+      "epoch": 4.56,
+      "learning_rate": 1.8155896482256317e-05,
+      "loss": 0.2239,
       "step": 9800
     },
     {
-      "epoch": 6.37,
-      "learning_rate": 1.2100931340697423e-05,
-      "loss": 0.0055,
+      "epoch": 4.56,
+      "learning_rate": 1.8140399814039985e-05,
+      "loss": 0.1346,
       "step": 9810
     },
     {
-      "epoch": 6.38,
-      "learning_rate": 1.2079272254710852e-05,
-      "loss": 0.122,
+      "epoch": 4.57,
+      "learning_rate": 1.8124903145823645e-05,
+      "loss": 0.1937,
       "step": 9820
     },
     {
-      "epoch": 6.39,
-      "learning_rate": 1.2057613168724279e-05,
-      "loss": 0.0016,
+      "epoch": 4.57,
+      "learning_rate": 1.8109406477607313e-05,
+      "loss": 0.1042,
       "step": 9830
     },
     {
-      "epoch": 6.39,
-      "learning_rate": 1.203595408273771e-05,
-      "loss": 0.0019,
+      "epoch": 4.57,
+      "learning_rate": 1.809390980939098e-05,
+      "loss": 0.3263,
       "step": 9840
     },
     {
-      "epoch": 6.4,
-      "learning_rate": 1.2014294996751138e-05,
-      "loss": 0.0679,
+      "epoch": 4.58,
+      "learning_rate": 1.8078413141174648e-05,
+      "loss": 0.2908,
       "step": 9850
     },
     {
-      "epoch": 6.4,
-      "learning_rate": 1.1992635910764567e-05,
-      "loss": 0.2834,
+      "epoch": 4.58,
+      "learning_rate": 1.8062916472958312e-05,
+      "loss": 0.1898,
       "step": 9860
     },
     {
-      "epoch": 6.41,
-      "learning_rate": 1.1970976824777994e-05,
-      "loss": 0.3688,
+      "epoch": 4.59,
+      "learning_rate": 1.804741980474198e-05,
+      "loss": 0.0576,
       "step": 9870
     },
     {
-      "epoch": 6.42,
-      "learning_rate": 1.1949317738791423e-05,
-      "loss": 0.1276,
+      "epoch": 4.59,
+      "learning_rate": 1.8031923136525647e-05,
+      "loss": 0.2679,
       "step": 9880
     },
     {
-      "epoch": 6.42,
-      "learning_rate": 1.1927658652804853e-05,
-      "loss": 0.1572,
+      "epoch": 4.6,
+      "learning_rate": 1.8016426468309314e-05,
+      "loss": 0.219,
       "step": 9890
     },
     {
-      "epoch": 6.43,
-      "learning_rate": 1.1905999566818282e-05,
-      "loss": 0.2999,
+      "epoch": 4.6,
+      "learning_rate": 1.800092980009298e-05,
+      "loss": 0.1807,
       "step": 9900
     },
     {
-      "epoch": 6.44,
-      "learning_rate": 1.1884340480831709e-05,
-      "loss": 0.0549,
+      "epoch": 4.61,
+      "learning_rate": 1.7985433131876646e-05,
+      "loss": 0.1964,
       "step": 9910
     },
     {
-      "epoch": 6.44,
-      "learning_rate": 1.1862681394845137e-05,
-      "loss": 0.0244,
+      "epoch": 4.61,
+      "learning_rate": 1.7969936463660313e-05,
+      "loss": 0.192,
       "step": 9920
     },
     {
-      "epoch": 6.45,
-      "learning_rate": 1.1841022308858568e-05,
-      "loss": 0.0144,
+      "epoch": 4.62,
+      "learning_rate": 1.795443979544398e-05,
+      "loss": 0.35,
       "step": 9930
     },
     {
-      "epoch": 6.46,
-      "learning_rate": 1.1819363222871996e-05,
-      "loss": 0.0892,
+      "epoch": 4.62,
+      "learning_rate": 1.7938943127227645e-05,
+      "loss": 0.2089,
       "step": 9940
     },
     {
-      "epoch": 6.46,
-      "learning_rate": 1.1797704136885423e-05,
-      "loss": 0.0441,
+      "epoch": 4.63,
+      "learning_rate": 1.7923446459011312e-05,
+      "loss": 0.3019,
       "step": 9950
     },
     {
-      "epoch": 6.47,
-      "learning_rate": 1.1776045050898852e-05,
-      "loss": 0.1115,
+      "epoch": 4.63,
+      "learning_rate": 1.790794979079498e-05,
+      "loss": 0.197,
       "step": 9960
     },
     {
-      "epoch": 6.48,
-      "learning_rate": 1.175438596491228e-05,
-      "loss": 0.1608,
+      "epoch": 4.64,
+      "learning_rate": 1.7892453122578647e-05,
+      "loss": 0.0417,
       "step": 9970
     },
     {
-      "epoch": 6.48,
-      "learning_rate": 1.1734892787524366e-05,
-      "loss": 0.2287,
+      "epoch": 4.64,
+      "learning_rate": 1.787695645436231e-05,
+      "loss": 0.3343,
       "step": 9980
     },
     {
-      "epoch": 6.49,
-      "learning_rate": 1.1713233701537797e-05,
-      "loss": 0.0351,
+      "epoch": 4.64,
+      "learning_rate": 1.786145978614598e-05,
+      "loss": 0.2121,
       "step": 9990
     },
     {
-      "epoch": 6.5,
-      "learning_rate": 1.1691574615551224e-05,
-      "loss": 0.0526,
+      "epoch": 4.65,
+      "learning_rate": 1.7845963117929646e-05,
+      "loss": 0.2491,
       "step": 10000
     },
     {
-      "epoch": 6.5,
-      "learning_rate": 1.1669915529564652e-05,
-      "loss": 0.0966,
+      "epoch": 4.65,
+      "learning_rate": 1.7830466449713314e-05,
+      "loss": 0.4517,
       "step": 10010
     },
     {
-      "epoch": 6.51,
-      "learning_rate": 1.1648256443578081e-05,
-      "loss": 0.2692,
+      "epoch": 4.66,
+      "learning_rate": 1.7814969781496978e-05,
+      "loss": 0.007,
       "step": 10020
     },
     {
-      "epoch": 6.52,
-      "learning_rate": 1.162659735759151e-05,
-      "loss": 0.0803,
+      "epoch": 4.66,
+      "learning_rate": 1.7799473113280646e-05,
+      "loss": 0.1944,
       "step": 10030
     },
     {
-      "epoch": 6.52,
-      "learning_rate": 1.1604938271604938e-05,
-      "loss": 0.3985,
+      "epoch": 4.67,
+      "learning_rate": 1.7783976445064313e-05,
+      "loss": 0.2048,
       "step": 10040
     },
     {
-      "epoch": 6.53,
-      "learning_rate": 1.1583279185618367e-05,
-      "loss": 0.1525,
+      "epoch": 4.67,
+      "learning_rate": 1.776847977684798e-05,
+      "loss": 0.2641,
       "step": 10050
     },
     {
-      "epoch": 6.53,
-      "learning_rate": 1.1561620099631796e-05,
-      "loss": 0.1275,
+      "epoch": 4.68,
+      "learning_rate": 1.7752983108631648e-05,
+      "loss": 0.1171,
       "step": 10060
     },
     {
-      "epoch": 6.54,
-      "learning_rate": 1.1539961013645224e-05,
-      "loss": 0.0422,
+      "epoch": 4.68,
+      "learning_rate": 1.773748644041531e-05,
+      "loss": 0.1774,
       "step": 10070
     },
     {
-      "epoch": 6.55,
-      "learning_rate": 1.1518301927658653e-05,
-      "loss": 0.1179,
+      "epoch": 4.69,
+      "learning_rate": 1.7721989772198976e-05,
+      "loss": 0.2047,
       "step": 10080
     },
     {
-      "epoch": 6.55,
-      "learning_rate": 1.1496642841672082e-05,
-      "loss": 0.2389,
+      "epoch": 4.69,
+      "learning_rate": 1.7706493103982644e-05,
+      "loss": 0.2553,
       "step": 10090
     },
     {
-      "epoch": 6.56,
-      "learning_rate": 1.147498375568551e-05,
-      "loss": 0.0017,
+      "epoch": 4.7,
+      "learning_rate": 1.769099643576631e-05,
+      "loss": 0.2305,
       "step": 10100
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 1.1453324669698939e-05,
-      "loss": 0.5718,
+      "epoch": 4.7,
+      "learning_rate": 1.7675499767549975e-05,
+      "loss": 0.1751,
       "step": 10110
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 1.1431665583712366e-05,
-      "loss": 0.1237,
+      "epoch": 4.7,
+      "learning_rate": 1.7660003099333643e-05,
+      "loss": 0.2144,
       "step": 10120
     },
     {
-      "epoch": 6.58,
-      "learning_rate": 1.1410006497725796e-05,
-      "loss": 0.1591,
+      "epoch": 4.71,
+      "learning_rate": 1.764450643111731e-05,
+      "loss": 0.1307,
       "step": 10130
     },
     {
-      "epoch": 6.59,
-      "learning_rate": 1.1388347411739225e-05,
-      "loss": 0.0555,
+      "epoch": 4.71,
+      "learning_rate": 1.7629009762900978e-05,
+      "loss": 0.2372,
       "step": 10140
     },
     {
-      "epoch": 6.59,
-      "learning_rate": 1.1366688325752654e-05,
-      "loss": 0.0478,
+      "epoch": 4.72,
+      "learning_rate": 1.7613513094684642e-05,
+      "loss": 0.1058,
       "step": 10150
     },
     {
-      "epoch": 6.6,
-      "learning_rate": 1.134502923976608e-05,
-      "loss": 0.1472,
+      "epoch": 4.72,
+      "learning_rate": 1.759801642646831e-05,
+      "loss": 0.3989,
       "step": 10160
     },
     {
-      "epoch": 6.61,
-      "learning_rate": 1.1323370153779511e-05,
-      "loss": 0.1414,
+      "epoch": 4.73,
+      "learning_rate": 1.7582519758251977e-05,
+      "loss": 0.1809,
       "step": 10170
     },
     {
-      "epoch": 6.61,
-      "learning_rate": 1.130171106779294e-05,
-      "loss": 0.1624,
+      "epoch": 4.73,
+      "learning_rate": 1.7567023090035644e-05,
+      "loss": 0.1928,
       "step": 10180
     },
     {
-      "epoch": 6.62,
-      "learning_rate": 1.1280051981806369e-05,
-      "loss": 0.4795,
+      "epoch": 4.74,
+      "learning_rate": 1.755152642181931e-05,
+      "loss": 0.0986,
       "step": 10190
     },
     {
-      "epoch": 6.63,
-      "learning_rate": 1.1258392895819796e-05,
-      "loss": 0.2595,
+      "epoch": 4.74,
+      "learning_rate": 1.7536029753602976e-05,
+      "loss": 0.3917,
       "step": 10200
     },
     {
-      "epoch": 6.63,
-      "learning_rate": 1.1236733809833224e-05,
-      "loss": 0.0409,
+      "epoch": 4.75,
+      "learning_rate": 1.7520533085386643e-05,
+      "loss": 0.4439,
       "step": 10210
     },
     {
-      "epoch": 6.64,
-      "learning_rate": 1.1215074723846655e-05,
-      "loss": 0.0421,
+      "epoch": 4.75,
+      "learning_rate": 1.750503641717031e-05,
+      "loss": 0.4091,
       "step": 10220
     },
     {
-      "epoch": 6.65,
-      "learning_rate": 1.1193415637860083e-05,
-      "loss": 0.1452,
+      "epoch": 4.76,
+      "learning_rate": 1.7489539748953975e-05,
+      "loss": 0.0528,
       "step": 10230
     },
     {
-      "epoch": 6.65,
-      "learning_rate": 1.117175655187351e-05,
-      "loss": 0.1621,
+      "epoch": 4.76,
+      "learning_rate": 1.7474043080737642e-05,
+      "loss": 0.3893,
       "step": 10240
     },
     {
-      "epoch": 6.66,
-      "learning_rate": 1.1150097465886939e-05,
-      "loss": 0.1155,
+      "epoch": 4.77,
+      "learning_rate": 1.745854641252131e-05,
+      "loss": 0.3417,
       "step": 10250
     },
     {
-      "epoch": 6.66,
-      "learning_rate": 1.112843837990037e-05,
-      "loss": 0.1035,
+      "epoch": 4.77,
+      "learning_rate": 1.7443049744304977e-05,
+      "loss": 0.3438,
       "step": 10260
     },
     {
-      "epoch": 6.67,
-      "learning_rate": 1.1106779293913798e-05,
-      "loss": 0.1507,
+      "epoch": 4.77,
+      "learning_rate": 1.742755307608864e-05,
+      "loss": 0.2202,
       "step": 10270
     },
     {
-      "epoch": 6.68,
-      "learning_rate": 1.1085120207927225e-05,
-      "loss": 0.1554,
+      "epoch": 4.78,
+      "learning_rate": 1.741205640787231e-05,
+      "loss": 0.1588,
       "step": 10280
     },
     {
-      "epoch": 6.68,
-      "learning_rate": 1.1063461121940654e-05,
-      "loss": 0.0084,
+      "epoch": 4.78,
+      "learning_rate": 1.7396559739655976e-05,
+      "loss": 0.2943,
       "step": 10290
     },
     {
-      "epoch": 6.69,
-      "learning_rate": 1.1041802035954082e-05,
-      "loss": 0.2307,
+      "epoch": 4.79,
+      "learning_rate": 1.7381063071439644e-05,
+      "loss": 0.3582,
       "step": 10300
     },
     {
-      "epoch": 6.7,
-      "learning_rate": 1.1020142949967513e-05,
-      "loss": 0.197,
+      "epoch": 4.79,
+      "learning_rate": 1.7365566403223308e-05,
+      "loss": 0.063,
       "step": 10310
     },
     {
-      "epoch": 6.7,
-      "learning_rate": 1.099848386398094e-05,
-      "loss": 0.2854,
+      "epoch": 4.8,
+      "learning_rate": 1.7350069735006972e-05,
+      "loss": 0.2663,
       "step": 10320
     },
     {
-      "epoch": 6.71,
-      "learning_rate": 1.0976824777994368e-05,
-      "loss": 0.1157,
+      "epoch": 4.8,
+      "learning_rate": 1.733457306679064e-05,
+      "loss": 0.2395,
       "step": 10330
     },
     {
-      "epoch": 6.72,
-      "learning_rate": 1.0955165692007797e-05,
-      "loss": 0.1403,
+      "epoch": 4.81,
+      "learning_rate": 1.7319076398574307e-05,
+      "loss": 0.1856,
       "step": 10340
     },
     {
-      "epoch": 6.72,
-      "learning_rate": 1.0933506606021228e-05,
-      "loss": 0.1204,
+      "epoch": 4.81,
+      "learning_rate": 1.730357973035797e-05,
+      "loss": 0.4515,
       "step": 10350
     },
     {
-      "epoch": 6.73,
-      "learning_rate": 1.0911847520034655e-05,
-      "loss": 0.2161,
+      "epoch": 4.82,
+      "learning_rate": 1.728808306214164e-05,
+      "loss": 0.1686,
       "step": 10360
     },
     {
-      "epoch": 6.74,
-      "learning_rate": 1.0890188434048083e-05,
-      "loss": 0.2184,
+      "epoch": 4.82,
+      "learning_rate": 1.7272586393925306e-05,
+      "loss": 0.2511,
       "step": 10370
     },
     {
-      "epoch": 6.74,
-      "learning_rate": 1.0868529348061512e-05,
-      "loss": 0.2235,
+      "epoch": 4.83,
+      "learning_rate": 1.7257089725708974e-05,
+      "loss": 0.0803,
       "step": 10380
     },
     {
-      "epoch": 6.75,
-      "learning_rate": 1.084687026207494e-05,
-      "loss": 0.0236,
+      "epoch": 4.83,
+      "learning_rate": 1.7241593057492638e-05,
+      "loss": 0.0816,
       "step": 10390
     },
     {
-      "epoch": 6.76,
-      "learning_rate": 1.082521117608837e-05,
-      "loss": 0.2451,
+      "epoch": 4.83,
+      "learning_rate": 1.7226096389276305e-05,
+      "loss": 0.2426,
       "step": 10400
     },
     {
-      "epoch": 6.76,
-      "learning_rate": 1.0803552090101798e-05,
-      "loss": 0.1368,
+      "epoch": 4.84,
+      "learning_rate": 1.7210599721059973e-05,
+      "loss": 0.302,
       "step": 10410
     },
     {
-      "epoch": 6.77,
-      "learning_rate": 1.0781893004115227e-05,
-      "loss": 0.166,
+      "epoch": 4.84,
+      "learning_rate": 1.719510305284364e-05,
+      "loss": 0.3127,
       "step": 10420
     },
     {
-      "epoch": 6.77,
-      "learning_rate": 1.0760233918128655e-05,
-      "loss": 0.0876,
+      "epoch": 4.85,
+      "learning_rate": 1.7179606384627304e-05,
+      "loss": 0.0458,
       "step": 10430
     },
     {
-      "epoch": 6.78,
-      "learning_rate": 1.0738574832142084e-05,
-      "loss": 0.1326,
+      "epoch": 4.85,
+      "learning_rate": 1.7164109716410972e-05,
+      "loss": 0.3047,
       "step": 10440
     },
     {
-      "epoch": 6.79,
-      "learning_rate": 1.0716915746155513e-05,
-      "loss": 0.1558,
+      "epoch": 4.86,
+      "learning_rate": 1.714861304819464e-05,
+      "loss": 0.1506,
       "step": 10450
     },
     {
-      "epoch": 6.79,
-      "learning_rate": 1.0695256660168941e-05,
-      "loss": 0.2767,
+      "epoch": 4.86,
+      "learning_rate": 1.7133116379978307e-05,
+      "loss": 0.1487,
       "step": 10460
     },
     {
-      "epoch": 6.8,
-      "learning_rate": 1.067359757418237e-05,
-      "loss": 0.2269,
+      "epoch": 4.87,
+      "learning_rate": 1.711761971176197e-05,
+      "loss": 0.0061,
       "step": 10470
     },
     {
-      "epoch": 6.81,
-      "learning_rate": 1.0651938488195797e-05,
-      "loss": 0.2727,
+      "epoch": 4.87,
+      "learning_rate": 1.7102123043545638e-05,
+      "loss": 0.2508,
       "step": 10480
     },
     {
-      "epoch": 6.81,
-      "learning_rate": 1.0630279402209227e-05,
-      "loss": 0.1883,
+      "epoch": 4.88,
+      "learning_rate": 1.7086626375329306e-05,
+      "loss": 0.1569,
       "step": 10490
     },
     {
-      "epoch": 6.82,
-      "learning_rate": 1.0608620316222656e-05,
-      "loss": 0.3771,
+      "epoch": 4.88,
+      "learning_rate": 1.7071129707112973e-05,
+      "loss": 0.0749,
       "step": 10500
     },
     {
-      "epoch": 6.83,
-      "learning_rate": 1.0586961230236085e-05,
-      "loss": 0.1681,
+      "epoch": 4.89,
+      "learning_rate": 1.7055633038896637e-05,
+      "loss": 0.0238,
       "step": 10510
     },
     {
-      "epoch": 6.83,
-      "learning_rate": 1.0565302144249512e-05,
-      "loss": 0.0599,
+      "epoch": 4.89,
+      "learning_rate": 1.7040136370680305e-05,
+      "loss": 0.0766,
       "step": 10520
     },
     {
-      "epoch": 6.84,
-      "learning_rate": 1.0543643058262942e-05,
-      "loss": 0.2127,
+      "epoch": 4.9,
+      "learning_rate": 1.7024639702463972e-05,
+      "loss": 0.2067,
       "step": 10530
     },
     {
-      "epoch": 6.85,
-      "learning_rate": 1.0521983972276371e-05,
-      "loss": 0.1691,
+      "epoch": 4.9,
+      "learning_rate": 1.700914303424764e-05,
+      "loss": 0.4927,
       "step": 10540
     },
     {
-      "epoch": 6.85,
-      "learning_rate": 1.05003248862898e-05,
-      "loss": 0.2349,
+      "epoch": 4.9,
+      "learning_rate": 1.6993646366031304e-05,
+      "loss": 0.43,
       "step": 10550
     },
     {
-      "epoch": 6.86,
-      "learning_rate": 1.0478665800303227e-05,
-      "loss": 0.2292,
+      "epoch": 4.91,
+      "learning_rate": 1.697814969781497e-05,
+      "loss": 0.0936,
       "step": 10560
     },
     {
-      "epoch": 6.87,
-      "learning_rate": 1.0457006714316655e-05,
-      "loss": 0.3338,
+      "epoch": 4.91,
+      "learning_rate": 1.6962653029598635e-05,
+      "loss": 0.1476,
       "step": 10570
     },
     {
-      "epoch": 6.87,
-      "learning_rate": 1.0435347628330086e-05,
-      "loss": 0.1032,
+      "epoch": 4.92,
+      "learning_rate": 1.6947156361382303e-05,
+      "loss": 0.1279,
       "step": 10580
     },
     {
-      "epoch": 6.88,
-      "learning_rate": 1.0413688542343514e-05,
-      "loss": 0.059,
+      "epoch": 4.92,
+      "learning_rate": 1.6931659693165967e-05,
+      "loss": 0.1488,
       "step": 10590
     },
     {
-      "epoch": 6.89,
-      "learning_rate": 1.0392029456356941e-05,
-      "loss": 0.272,
+      "epoch": 4.93,
+      "learning_rate": 1.6916163024949635e-05,
+      "loss": 0.1668,
       "step": 10600
     },
     {
-      "epoch": 6.89,
-      "learning_rate": 1.037037037037037e-05,
-      "loss": 0.0206,
+      "epoch": 4.93,
+      "learning_rate": 1.6900666356733302e-05,
+      "loss": 0.351,
       "step": 10610
     },
     {
-      "epoch": 6.9,
-      "learning_rate": 1.03487112843838e-05,
-      "loss": 0.2188,
+      "epoch": 4.94,
+      "learning_rate": 1.688516968851697e-05,
+      "loss": 0.2326,
       "step": 10620
     },
     {
-      "epoch": 6.9,
-      "learning_rate": 1.0327052198397229e-05,
-      "loss": 0.1537,
+      "epoch": 4.94,
+      "learning_rate": 1.6869673020300634e-05,
+      "loss": 0.184,
       "step": 10630
     },
     {
-      "epoch": 6.91,
-      "learning_rate": 1.0305393112410656e-05,
-      "loss": 0.2969,
+      "epoch": 4.95,
+      "learning_rate": 1.68541763520843e-05,
+      "loss": 0.4558,
       "step": 10640
     },
     {
-      "epoch": 6.92,
-      "learning_rate": 1.0283734026424085e-05,
-      "loss": 0.1091,
+      "epoch": 4.95,
+      "learning_rate": 1.683867968386797e-05,
+      "loss": 0.1829,
       "step": 10650
     },
     {
-      "epoch": 6.92,
-      "learning_rate": 1.0262074940437513e-05,
-      "loss": 0.1372,
+      "epoch": 4.96,
+      "learning_rate": 1.6823183015651636e-05,
+      "loss": 0.0798,
       "step": 10660
     },
     {
-      "epoch": 6.93,
-      "learning_rate": 1.0240415854450944e-05,
-      "loss": 0.1222,
+      "epoch": 4.96,
+      "learning_rate": 1.68076863474353e-05,
+      "loss": 0.1519,
       "step": 10670
     },
     {
-      "epoch": 6.94,
-      "learning_rate": 1.021875676846437e-05,
-      "loss": 0.0026,
+      "epoch": 4.97,
+      "learning_rate": 1.6793739346040604e-05,
+      "loss": 0.2262,
       "step": 10680
     },
     {
-      "epoch": 6.94,
-      "learning_rate": 1.01970976824778e-05,
-      "loss": 0.0525,
+      "epoch": 4.97,
+      "learning_rate": 1.677824267782427e-05,
+      "loss": 0.2732,
       "step": 10690
     },
     {
-      "epoch": 6.95,
-      "learning_rate": 1.0175438596491228e-05,
-      "loss": 0.2373,
+      "epoch": 4.97,
+      "learning_rate": 1.6762746009607936e-05,
+      "loss": 0.2281,
       "step": 10700
     },
     {
-      "epoch": 6.96,
-      "learning_rate": 1.0153779510504659e-05,
-      "loss": 0.1444,
+      "epoch": 4.98,
+      "learning_rate": 1.6747249341391603e-05,
+      "loss": 0.0631,
       "step": 10710
     },
     {
-      "epoch": 6.96,
-      "learning_rate": 1.0132120424518086e-05,
-      "loss": 0.1887,
+      "epoch": 4.98,
+      "learning_rate": 1.6731752673175267e-05,
+      "loss": 0.18,
       "step": 10720
     },
     {
-      "epoch": 6.97,
-      "learning_rate": 1.0110461338531514e-05,
-      "loss": 0.0863,
+      "epoch": 4.99,
+      "learning_rate": 1.671625600495893e-05,
+      "loss": 0.1363,
       "step": 10730
     },
     {
-      "epoch": 6.98,
-      "learning_rate": 1.0088802252544943e-05,
-      "loss": 0.1191,
+      "epoch": 4.99,
+      "learning_rate": 1.67007593367426e-05,
+      "loss": 0.4271,
       "step": 10740
     },
     {
-      "epoch": 6.98,
-      "learning_rate": 1.0067143166558372e-05,
-      "loss": 0.2556,
+      "epoch": 5.0,
+      "learning_rate": 1.6685262668526266e-05,
+      "loss": 0.2679,
       "step": 10750
     },
     {
-      "epoch": 6.99,
-      "learning_rate": 1.00454840805718e-05,
-      "loss": 0.2222,
-      "step": 10760
+      "epoch": 5.0,
+      "eval_accuracy": 0.9703654922621008,
+      "eval_loss": 0.13722212612628937,
+      "eval_runtime": 17.106,
+      "eval_samples_per_second": 177.54,
+      "eval_steps_per_second": 44.429,
+      "step": 10755
     },
     {
-      "epoch": 7.0,
-      "learning_rate": 1.0023824994585229e-05,
-      "loss": 0.031,
-      "step": 10770
+      "epoch": 5.0,
+      "learning_rate": 1.6669766000309934e-05,
+      "loss": 0.0703,
+      "step": 10760
     },
     {
-      "epoch": 7.0,
-      "eval_accuracy": 0.984820607175713,
-      "eval_loss": 0.08112048357725143,
-      "eval_runtime": 12.6945,
-      "eval_samples_per_second": 171.255,
-      "eval_steps_per_second": 42.853,
-      "step": 10776
+      "epoch": 5.01,
+      "learning_rate": 1.6654269332093598e-05,
+      "loss": 0.2159,
+      "step": 10770
     },
     {
-      "epoch": 7.0,
-      "learning_rate": 1.0002165908598658e-05,
-      "loss": 0.0887,
+      "epoch": 5.01,
+      "learning_rate": 1.6638772663877266e-05,
+      "loss": 0.3982,
       "step": 10780
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 9.980506822612086e-06,
-      "loss": 0.195,
+      "epoch": 5.02,
+      "learning_rate": 1.6623275995660933e-05,
+      "loss": 0.1444,
       "step": 10790
     },
     {
-      "epoch": 7.02,
-      "learning_rate": 9.958847736625515e-06,
-      "loss": 0.176,
+      "epoch": 5.02,
+      "learning_rate": 1.66077793274446e-05,
+      "loss": 0.3431,
       "step": 10800
     },
     {
-      "epoch": 7.02,
-      "learning_rate": 9.937188650638944e-06,
-      "loss": 0.1994,
+      "epoch": 5.03,
+      "learning_rate": 1.6592282659228265e-05,
+      "loss": 0.1391,
       "step": 10810
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 9.915529564652372e-06,
-      "loss": 0.0794,
+      "epoch": 5.03,
+      "learning_rate": 1.6576785991011932e-05,
+      "loss": 0.0821,
       "step": 10820
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 9.893870478665801e-06,
-      "loss": 0.0675,
+      "epoch": 5.03,
+      "learning_rate": 1.65612893227956e-05,
+      "loss": 0.0049,
       "step": 10830
     },
     {
-      "epoch": 7.04,
-      "learning_rate": 9.872211392679228e-06,
-      "loss": 0.1071,
+      "epoch": 5.04,
+      "learning_rate": 1.6545792654579267e-05,
+      "loss": 0.1551,
       "step": 10840
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 9.850552306692658e-06,
-      "loss": 0.033,
+      "epoch": 5.04,
+      "learning_rate": 1.653029598636293e-05,
+      "loss": 0.1058,
       "step": 10850
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 9.828893220706087e-06,
-      "loss": 0.0126,
+      "epoch": 5.05,
+      "learning_rate": 1.65147993181466e-05,
+      "loss": 0.4053,
       "step": 10860
     },
     {
-      "epoch": 7.06,
-      "learning_rate": 9.807234134719514e-06,
-      "loss": 0.2058,
+      "epoch": 5.05,
+      "learning_rate": 1.6499302649930266e-05,
+      "loss": 0.1804,
       "step": 10870
     },
     {
-      "epoch": 7.07,
-      "learning_rate": 9.785575048732943e-06,
-      "loss": 0.2196,
+      "epoch": 5.06,
+      "learning_rate": 1.6483805981713934e-05,
+      "loss": 0.311,
       "step": 10880
     },
     {
-      "epoch": 7.07,
-      "learning_rate": 9.763915962746373e-06,
-      "loss": 0.0015,
+      "epoch": 5.06,
+      "learning_rate": 1.6468309313497598e-05,
+      "loss": 0.3197,
       "step": 10890
     },
     {
-      "epoch": 7.08,
-      "learning_rate": 9.742256876759802e-06,
-      "loss": 0.0216,
+      "epoch": 5.07,
+      "learning_rate": 1.6452812645281265e-05,
+      "loss": 0.1787,
       "step": 10900
     },
     {
-      "epoch": 7.09,
-      "learning_rate": 9.720597790773229e-06,
-      "loss": 0.008,
+      "epoch": 5.07,
+      "learning_rate": 1.6437315977064933e-05,
+      "loss": 0.2355,
       "step": 10910
     },
     {
-      "epoch": 7.09,
-      "learning_rate": 9.698938704786658e-06,
-      "loss": 0.3726,
+      "epoch": 5.08,
+      "learning_rate": 1.64218193088486e-05,
+      "loss": 0.2442,
       "step": 10920
     },
     {
-      "epoch": 7.1,
-      "learning_rate": 9.677279618800086e-06,
-      "loss": 0.1296,
+      "epoch": 5.08,
+      "learning_rate": 1.6406322640632264e-05,
+      "loss": 0.097,
       "step": 10930
     },
     {
-      "epoch": 7.11,
-      "learning_rate": 9.655620532813517e-06,
-      "loss": 0.034,
+      "epoch": 5.09,
+      "learning_rate": 1.6390825972415932e-05,
+      "loss": 0.1611,
       "step": 10940
     },
     {
-      "epoch": 7.11,
-      "learning_rate": 9.633961446826944e-06,
-      "loss": 0.0329,
+      "epoch": 5.09,
+      "learning_rate": 1.63753293041996e-05,
+      "loss": 0.4332,
       "step": 10950
     },
     {
-      "epoch": 7.12,
-      "learning_rate": 9.612302360840372e-06,
-      "loss": 0.1104,
+      "epoch": 5.1,
+      "learning_rate": 1.6359832635983267e-05,
+      "loss": 0.1403,
       "step": 10960
     },
     {
-      "epoch": 7.13,
-      "learning_rate": 9.590643274853801e-06,
-      "loss": 0.0433,
+      "epoch": 5.1,
+      "learning_rate": 1.6344335967766927e-05,
+      "loss": 0.2731,
       "step": 10970
     },
     {
-      "epoch": 7.13,
-      "learning_rate": 9.568984188867231e-06,
-      "loss": 0.0929,
+      "epoch": 5.1,
+      "learning_rate": 1.6328839299550595e-05,
+      "loss": 0.1247,
       "step": 10980
     },
     {
-      "epoch": 7.14,
-      "learning_rate": 9.547325102880658e-06,
-      "loss": 0.0112,
+      "epoch": 5.11,
+      "learning_rate": 1.6313342631334262e-05,
+      "loss": 0.1658,
       "step": 10990
     },
     {
-      "epoch": 7.15,
-      "learning_rate": 9.525666016894087e-06,
-      "loss": 0.076,
+      "epoch": 5.11,
+      "learning_rate": 1.629784596311793e-05,
+      "loss": 0.267,
       "step": 11000
     },
     {
-      "epoch": 7.15,
-      "learning_rate": 9.504006930907516e-06,
-      "loss": 0.0653,
+      "epoch": 5.12,
+      "learning_rate": 1.6282349294901594e-05,
+      "loss": 0.2864,
       "step": 11010
     },
     {
-      "epoch": 7.16,
-      "learning_rate": 9.482347844920944e-06,
-      "loss": 0.3116,
+      "epoch": 5.12,
+      "learning_rate": 1.626685262668526e-05,
+      "loss": 0.056,
       "step": 11020
     },
     {
-      "epoch": 7.16,
-      "learning_rate": 9.460688758934373e-06,
-      "loss": 0.0533,
+      "epoch": 5.13,
+      "learning_rate": 1.625135595846893e-05,
+      "loss": 0.2699,
       "step": 11030
     },
     {
-      "epoch": 7.17,
-      "learning_rate": 9.439029672947802e-06,
-      "loss": 0.0627,
+      "epoch": 5.13,
+      "learning_rate": 1.6235859290252596e-05,
+      "loss": 0.0831,
       "step": 11040
     },
     {
-      "epoch": 7.18,
-      "learning_rate": 9.41737058696123e-06,
-      "loss": 0.0206,
+      "epoch": 5.14,
+      "learning_rate": 1.622036262203626e-05,
+      "loss": 0.1419,
       "step": 11050
     },
     {
-      "epoch": 7.18,
-      "learning_rate": 9.395711500974659e-06,
-      "loss": 0.1429,
+      "epoch": 5.14,
+      "learning_rate": 1.6204865953819928e-05,
+      "loss": 0.3128,
       "step": 11060
     },
     {
-      "epoch": 7.19,
-      "learning_rate": 9.374052414988088e-06,
-      "loss": 0.1702,
+      "epoch": 5.15,
+      "learning_rate": 1.6189369285603595e-05,
+      "loss": 0.3543,
       "step": 11070
     },
     {
-      "epoch": 7.2,
-      "learning_rate": 9.352393329001516e-06,
-      "loss": 0.0337,
+      "epoch": 5.15,
+      "learning_rate": 1.6173872617387263e-05,
+      "loss": 0.4131,
       "step": 11080
     },
     {
-      "epoch": 7.2,
-      "learning_rate": 9.330734243014945e-06,
-      "loss": 0.1393,
+      "epoch": 5.16,
+      "learning_rate": 1.6158375949170927e-05,
+      "loss": 0.1686,
       "step": 11090
     },
     {
-      "epoch": 7.21,
-      "learning_rate": 9.309075157028374e-06,
-      "loss": 0.1942,
+      "epoch": 5.16,
+      "learning_rate": 1.6142879280954595e-05,
+      "loss": 0.1019,
       "step": 11100
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 9.2874160710418e-06,
-      "loss": 0.2188,
+      "epoch": 5.17,
+      "learning_rate": 1.6127382612738262e-05,
+      "loss": 0.0761,
       "step": 11110
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 9.265756985055231e-06,
-      "loss": 0.0614,
+      "epoch": 5.17,
+      "learning_rate": 1.611188594452193e-05,
+      "loss": 0.047,
       "step": 11120
     },
     {
-      "epoch": 7.23,
-      "learning_rate": 9.24409789906866e-06,
-      "loss": 0.2601,
+      "epoch": 5.17,
+      "learning_rate": 1.6096389276305594e-05,
+      "loss": 0.1922,
       "step": 11130
     },
     {
-      "epoch": 7.24,
-      "learning_rate": 9.222438813082089e-06,
-      "loss": 0.2971,
+      "epoch": 5.18,
+      "learning_rate": 1.608089260808926e-05,
+      "loss": 0.2174,
       "step": 11140
     },
     {
-      "epoch": 7.24,
-      "learning_rate": 9.200779727095516e-06,
-      "loss": 0.1568,
+      "epoch": 5.18,
+      "learning_rate": 1.606539593987293e-05,
+      "loss": 0.0466,
       "step": 11150
     },
     {
-      "epoch": 7.25,
-      "learning_rate": 9.179120641108946e-06,
-      "loss": 0.2477,
+      "epoch": 5.19,
+      "learning_rate": 1.6049899271656596e-05,
+      "loss": 0.1752,
       "step": 11160
     },
     {
-      "epoch": 7.26,
-      "learning_rate": 9.157461555122375e-06,
-      "loss": 0.3047,
+      "epoch": 5.19,
+      "learning_rate": 1.603440260344026e-05,
+      "loss": 0.4495,
       "step": 11170
     },
     {
-      "epoch": 7.26,
-      "learning_rate": 9.135802469135803e-06,
-      "loss": 0.1162,
+      "epoch": 5.2,
+      "learning_rate": 1.6018905935223928e-05,
+      "loss": 0.3565,
       "step": 11180
     },
     {
-      "epoch": 7.27,
-      "learning_rate": 9.11414338314923e-06,
-      "loss": 0.112,
+      "epoch": 5.2,
+      "learning_rate": 1.6003409267007595e-05,
+      "loss": 0.1406,
       "step": 11190
     },
     {
-      "epoch": 7.28,
-      "learning_rate": 9.092484297162659e-06,
-      "loss": 0.1143,
+      "epoch": 5.21,
+      "learning_rate": 1.5987912598791263e-05,
+      "loss": 0.0822,
       "step": 11200
     },
     {
-      "epoch": 7.28,
-      "learning_rate": 9.07082521117609e-06,
-      "loss": 0.0668,
+      "epoch": 5.21,
+      "learning_rate": 1.597241593057493e-05,
+      "loss": 0.0941,
       "step": 11210
     },
     {
-      "epoch": 7.29,
-      "learning_rate": 9.049166125189518e-06,
-      "loss": 0.2082,
+      "epoch": 5.22,
+      "learning_rate": 1.595691926235859e-05,
+      "loss": 0.3932,
       "step": 11220
     },
     {
-      "epoch": 7.29,
-      "learning_rate": 9.027507039202945e-06,
-      "loss": 0.1127,
+      "epoch": 5.22,
+      "learning_rate": 1.5941422594142258e-05,
+      "loss": 0.0531,
       "step": 11230
     },
     {
-      "epoch": 7.3,
-      "learning_rate": 9.005847953216374e-06,
-      "loss": 0.0592,
+      "epoch": 5.23,
+      "learning_rate": 1.5925925925925926e-05,
+      "loss": 0.1834,
       "step": 11240
     },
     {
-      "epoch": 7.31,
-      "learning_rate": 8.984188867229804e-06,
-      "loss": 0.1539,
+      "epoch": 5.23,
+      "learning_rate": 1.5910429257709593e-05,
+      "loss": 0.0858,
       "step": 11250
     },
     {
-      "epoch": 7.31,
-      "learning_rate": 8.962529781243233e-06,
-      "loss": 0.1297,
+      "epoch": 5.23,
+      "learning_rate": 1.5894932589493257e-05,
+      "loss": 0.1469,
       "step": 11260
     },
     {
-      "epoch": 7.32,
-      "learning_rate": 8.94087069525666e-06,
-      "loss": 0.1659,
+      "epoch": 5.24,
+      "learning_rate": 1.5879435921276925e-05,
+      "loss": 0.0554,
       "step": 11270
     },
     {
-      "epoch": 7.33,
-      "learning_rate": 8.919211609270088e-06,
-      "loss": 0.1437,
+      "epoch": 5.24,
+      "learning_rate": 1.5863939253060592e-05,
+      "loss": 0.1853,
       "step": 11280
     },
     {
-      "epoch": 7.33,
-      "learning_rate": 8.897552523283517e-06,
-      "loss": 0.0978,
+      "epoch": 5.25,
+      "learning_rate": 1.584844258484426e-05,
+      "loss": 0.136,
       "step": 11290
     },
     {
-      "epoch": 7.34,
-      "learning_rate": 8.875893437296948e-06,
-      "loss": 0.228,
+      "epoch": 5.25,
+      "learning_rate": 1.5832945916627924e-05,
+      "loss": 0.2439,
       "step": 11300
     },
     {
-      "epoch": 7.35,
-      "learning_rate": 8.854234351310375e-06,
-      "loss": 0.0906,
+      "epoch": 5.26,
+      "learning_rate": 1.581744924841159e-05,
+      "loss": 0.317,
       "step": 11310
     },
     {
-      "epoch": 7.35,
-      "learning_rate": 8.832575265323803e-06,
-      "loss": 0.173,
+      "epoch": 5.26,
+      "learning_rate": 1.580195258019526e-05,
+      "loss": 0.2662,
       "step": 11320
     },
     {
-      "epoch": 7.36,
-      "learning_rate": 8.810916179337232e-06,
-      "loss": 0.0306,
+      "epoch": 5.27,
+      "learning_rate": 1.5786455911978926e-05,
+      "loss": 0.3763,
       "step": 11330
     },
     {
-      "epoch": 7.37,
-      "learning_rate": 8.789257093350662e-06,
-      "loss": 0.1072,
+      "epoch": 5.27,
+      "learning_rate": 1.577095924376259e-05,
+      "loss": 0.3602,
       "step": 11340
     },
     {
-      "epoch": 7.37,
-      "learning_rate": 8.76759800736409e-06,
-      "loss": 0.0704,
+      "epoch": 5.28,
+      "learning_rate": 1.5755462575546258e-05,
+      "loss": 0.1409,
       "step": 11350
     },
     {
-      "epoch": 7.38,
-      "learning_rate": 8.745938921377518e-06,
-      "loss": 0.0096,
+      "epoch": 5.28,
+      "learning_rate": 1.5739965907329925e-05,
+      "loss": 0.2269,
       "step": 11360
     },
     {
-      "epoch": 7.39,
-      "learning_rate": 8.724279835390947e-06,
-      "loss": 0.4016,
+      "epoch": 5.29,
+      "learning_rate": 1.5724469239113593e-05,
+      "loss": 0.2376,
       "step": 11370
     },
     {
-      "epoch": 7.39,
-      "learning_rate": 8.702620749404375e-06,
-      "loss": 0.2172,
+      "epoch": 5.29,
+      "learning_rate": 1.5708972570897257e-05,
+      "loss": 0.0921,
       "step": 11380
     },
     {
-      "epoch": 7.4,
-      "learning_rate": 8.680961663417804e-06,
-      "loss": 0.2804,
+      "epoch": 5.3,
+      "learning_rate": 1.5693475902680924e-05,
+      "loss": 0.1173,
       "step": 11390
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 8.659302577431233e-06,
-      "loss": 0.1273,
+      "epoch": 5.3,
+      "learning_rate": 1.5677979234464592e-05,
+      "loss": 0.4219,
       "step": 11400
     },
     {
-      "epoch": 7.41,
-      "learning_rate": 8.637643491444661e-06,
-      "loss": 0.0394,
+      "epoch": 5.3,
+      "learning_rate": 1.566248256624826e-05,
+      "loss": 0.0347,
       "step": 11410
     },
     {
-      "epoch": 7.42,
-      "learning_rate": 8.61598440545809e-06,
-      "loss": 0.0711,
+      "epoch": 5.31,
+      "learning_rate": 1.5646985898031924e-05,
+      "loss": 0.3427,
       "step": 11420
     },
     {
-      "epoch": 7.42,
-      "learning_rate": 8.594325319471519e-06,
-      "loss": 0.188,
+      "epoch": 5.31,
+      "learning_rate": 1.563148922981559e-05,
+      "loss": 0.1206,
       "step": 11430
     },
     {
-      "epoch": 7.43,
-      "learning_rate": 8.572666233484947e-06,
-      "loss": 0.046,
+      "epoch": 5.32,
+      "learning_rate": 1.561599256159926e-05,
+      "loss": 0.2318,
       "step": 11440
     },
     {
-      "epoch": 7.44,
-      "learning_rate": 8.551007147498376e-06,
-      "loss": 0.1146,
+      "epoch": 5.32,
+      "learning_rate": 1.5600495893382926e-05,
+      "loss": 0.123,
       "step": 11450
     },
     {
-      "epoch": 7.44,
-      "learning_rate": 8.529348061511805e-06,
-      "loss": 0.039,
+      "epoch": 5.33,
+      "learning_rate": 1.558499922516659e-05,
+      "loss": 0.2684,
       "step": 11460
     },
     {
-      "epoch": 7.45,
-      "learning_rate": 8.507688975525232e-06,
-      "loss": 0.1031,
+      "epoch": 5.33,
+      "learning_rate": 1.5569502556950254e-05,
+      "loss": 0.4133,
       "step": 11470
     },
     {
-      "epoch": 7.46,
-      "learning_rate": 8.486029889538662e-06,
-      "loss": 0.003,
+      "epoch": 5.34,
+      "learning_rate": 1.5554005888733922e-05,
+      "loss": 0.4216,
       "step": 11480
     },
     {
-      "epoch": 7.46,
-      "learning_rate": 8.464370803552091e-06,
-      "loss": 0.018,
+      "epoch": 5.34,
+      "learning_rate": 1.553850922051759e-05,
+      "loss": 0.1955,
       "step": 11490
     },
     {
-      "epoch": 7.47,
-      "learning_rate": 8.44271171756552e-06,
-      "loss": 0.2962,
+      "epoch": 5.35,
+      "learning_rate": 1.5523012552301253e-05,
+      "loss": 0.1309,
       "step": 11500
     },
     {
-      "epoch": 7.48,
-      "learning_rate": 8.421052631578947e-06,
-      "loss": 0.053,
+      "epoch": 5.35,
+      "learning_rate": 1.550751588408492e-05,
+      "loss": 0.2917,
       "step": 11510
     },
     {
-      "epoch": 7.48,
-      "learning_rate": 8.399393545592377e-06,
-      "loss": 0.0209,
+      "epoch": 5.36,
+      "learning_rate": 1.5492019215868588e-05,
+      "loss": 0.2985,
       "step": 11520
     },
     {
-      "epoch": 7.49,
-      "learning_rate": 8.377734459605806e-06,
-      "loss": 0.0926,
+      "epoch": 5.36,
+      "learning_rate": 1.5476522547652256e-05,
+      "loss": 0.1159,
       "step": 11530
     },
     {
-      "epoch": 7.5,
-      "learning_rate": 8.356075373619234e-06,
-      "loss": 0.1764,
+      "epoch": 5.36,
+      "learning_rate": 1.546102587943592e-05,
+      "loss": 0.0084,
       "step": 11540
     },
     {
-      "epoch": 7.5,
-      "learning_rate": 8.334416287632661e-06,
-      "loss": 0.0722,
+      "epoch": 5.37,
+      "learning_rate": 1.5445529211219587e-05,
+      "loss": 0.1886,
       "step": 11550
     },
     {
-      "epoch": 7.51,
-      "learning_rate": 8.31275720164609e-06,
-      "loss": 0.1181,
+      "epoch": 5.37,
+      "learning_rate": 1.5430032543003255e-05,
+      "loss": 0.1164,
       "step": 11560
     },
     {
-      "epoch": 7.52,
-      "learning_rate": 8.29109811565952e-06,
-      "loss": 0.1074,
+      "epoch": 5.38,
+      "learning_rate": 1.5414535874786922e-05,
+      "loss": 0.2624,
       "step": 11570
     },
     {
-      "epoch": 7.52,
-      "learning_rate": 8.269439029672949e-06,
-      "loss": 0.1395,
+      "epoch": 5.38,
+      "learning_rate": 1.5399039206570586e-05,
+      "loss": 0.1023,
       "step": 11580
     },
     {
-      "epoch": 7.53,
-      "learning_rate": 8.247779943686376e-06,
-      "loss": 0.095,
+      "epoch": 5.39,
+      "learning_rate": 1.538509220517589e-05,
+      "loss": 0.3386,
       "step": 11590
     },
     {
-      "epoch": 7.53,
-      "learning_rate": 8.226120857699805e-06,
-      "loss": 0.0765,
+      "epoch": 5.39,
+      "learning_rate": 1.5369595536959555e-05,
+      "loss": 0.3027,
       "step": 11600
     },
     {
-      "epoch": 7.54,
-      "learning_rate": 8.204461771713233e-06,
-      "loss": 0.4006,
+      "epoch": 5.4,
+      "learning_rate": 1.5354098868743222e-05,
+      "loss": 0.2126,
       "step": 11610
     },
     {
-      "epoch": 7.55,
-      "learning_rate": 8.182802685726664e-06,
-      "loss": 0.0065,
+      "epoch": 5.4,
+      "learning_rate": 1.5338602200526886e-05,
+      "loss": 0.2166,
       "step": 11620
     },
     {
-      "epoch": 7.55,
-      "learning_rate": 8.16114359974009e-06,
-      "loss": 0.1837,
+      "epoch": 5.41,
+      "learning_rate": 1.5323105532310554e-05,
+      "loss": 0.1497,
       "step": 11630
     },
     {
-      "epoch": 7.56,
-      "learning_rate": 8.13948451375352e-06,
-      "loss": 0.1463,
+      "epoch": 5.41,
+      "learning_rate": 1.5307608864094218e-05,
+      "loss": 0.1867,
       "step": 11640
     },
     {
-      "epoch": 7.57,
-      "learning_rate": 8.117825427766948e-06,
-      "loss": 0.2499,
+      "epoch": 5.42,
+      "learning_rate": 1.5292112195877885e-05,
+      "loss": 0.1521,
       "step": 11650
     },
     {
-      "epoch": 7.57,
-      "learning_rate": 8.096166341780379e-06,
-      "loss": 0.1759,
+      "epoch": 5.42,
+      "learning_rate": 1.5276615527661553e-05,
+      "loss": 0.1259,
       "step": 11660
     },
     {
-      "epoch": 7.58,
-      "learning_rate": 8.074507255793806e-06,
-      "loss": 0.0632,
+      "epoch": 5.43,
+      "learning_rate": 1.526111885944522e-05,
+      "loss": 0.2591,
       "step": 11670
     },
     {
-      "epoch": 7.59,
-      "learning_rate": 8.052848169807234e-06,
-      "loss": 0.046,
+      "epoch": 5.43,
+      "learning_rate": 1.5245622191228884e-05,
+      "loss": 0.0995,
       "step": 11680
     },
     {
-      "epoch": 7.59,
-      "learning_rate": 8.031189083820663e-06,
-      "loss": 0.1882,
+      "epoch": 5.43,
+      "learning_rate": 1.5230125523012552e-05,
+      "loss": 0.2385,
       "step": 11690
     },
     {
-      "epoch": 7.6,
-      "learning_rate": 8.009529997834092e-06,
-      "loss": 0.2443,
+      "epoch": 5.44,
+      "learning_rate": 1.521462885479622e-05,
+      "loss": 0.1062,
       "step": 11700
     },
     {
-      "epoch": 7.61,
-      "learning_rate": 7.98787091184752e-06,
-      "loss": 0.0786,
+      "epoch": 5.44,
+      "learning_rate": 1.5199132186579887e-05,
+      "loss": 0.302,
       "step": 11710
     },
     {
-      "epoch": 7.61,
-      "learning_rate": 7.966211825860949e-06,
-      "loss": 0.0354,
+      "epoch": 5.45,
+      "learning_rate": 1.5183635518363551e-05,
+      "loss": 0.1853,
       "step": 11720
     },
     {
-      "epoch": 7.62,
-      "learning_rate": 7.944552739874378e-06,
-      "loss": 0.1392,
+      "epoch": 5.45,
+      "learning_rate": 1.5168138850147218e-05,
+      "loss": 0.2605,
       "step": 11730
     },
     {
-      "epoch": 7.63,
-      "learning_rate": 7.922893653887806e-06,
-      "loss": 0.1614,
+      "epoch": 5.46,
+      "learning_rate": 1.5152642181930886e-05,
+      "loss": 0.2215,
       "step": 11740
     },
     {
-      "epoch": 7.63,
-      "learning_rate": 7.901234567901235e-06,
-      "loss": 0.0229,
+      "epoch": 5.46,
+      "learning_rate": 1.5137145513714553e-05,
+      "loss": 0.2838,
       "step": 11750
     },
     {
-      "epoch": 7.64,
-      "learning_rate": 7.879575481914664e-06,
-      "loss": 0.0742,
+      "epoch": 5.47,
+      "learning_rate": 1.5121648845498217e-05,
+      "loss": 0.1278,
       "step": 11760
     },
     {
-      "epoch": 7.65,
-      "learning_rate": 7.857916395928092e-06,
-      "loss": 0.0247,
+      "epoch": 5.47,
+      "learning_rate": 1.5106152177281885e-05,
+      "loss": 0.0232,
       "step": 11770
     },
     {
-      "epoch": 7.65,
-      "learning_rate": 7.836257309941521e-06,
-      "loss": 0.0734,
+      "epoch": 5.48,
+      "learning_rate": 1.5090655509065552e-05,
+      "loss": 0.3235,
       "step": 11780
     },
     {
-      "epoch": 7.66,
-      "learning_rate": 7.814598223954948e-06,
-      "loss": 0.161,
+      "epoch": 5.48,
+      "learning_rate": 1.507515884084922e-05,
+      "loss": 0.2472,
       "step": 11790
     },
     {
-      "epoch": 7.66,
-      "learning_rate": 7.792939137968378e-06,
-      "loss": 0.1736,
+      "epoch": 5.49,
+      "learning_rate": 1.5059662172632884e-05,
+      "loss": 0.2624,
       "step": 11800
     },
     {
-      "epoch": 7.67,
-      "learning_rate": 7.771280051981807e-06,
-      "loss": 0.1407,
+      "epoch": 5.49,
+      "learning_rate": 1.504416550441655e-05,
+      "loss": 0.2014,
       "step": 11810
     },
     {
-      "epoch": 7.68,
-      "learning_rate": 7.749620965995236e-06,
-      "loss": 0.104,
+      "epoch": 5.5,
+      "learning_rate": 1.5028668836200217e-05,
+      "loss": 0.229,
       "step": 11820
     },
     {
-      "epoch": 7.68,
-      "learning_rate": 7.727961880008663e-06,
-      "loss": 0.1501,
+      "epoch": 5.5,
+      "learning_rate": 1.5013172167983885e-05,
+      "loss": 0.0053,
       "step": 11830
     },
     {
-      "epoch": 7.69,
-      "learning_rate": 7.706302794022093e-06,
-      "loss": 0.1137,
+      "epoch": 5.5,
+      "learning_rate": 1.499767549976755e-05,
+      "loss": 0.063,
       "step": 11840
     },
     {
-      "epoch": 7.7,
-      "learning_rate": 7.684643708035522e-06,
-      "loss": 0.1669,
+      "epoch": 5.51,
+      "learning_rate": 1.4982178831551216e-05,
+      "loss": 0.0395,
       "step": 11850
     },
     {
-      "epoch": 7.7,
-      "learning_rate": 7.662984622048949e-06,
-      "loss": 0.0013,
+      "epoch": 5.51,
+      "learning_rate": 1.4966682163334884e-05,
+      "loss": 0.0545,
       "step": 11860
     },
     {
-      "epoch": 7.71,
-      "learning_rate": 7.641325536062378e-06,
-      "loss": 0.1373,
+      "epoch": 5.52,
+      "learning_rate": 1.495118549511855e-05,
+      "loss": 0.2367,
       "step": 11870
     },
     {
-      "epoch": 7.72,
-      "learning_rate": 7.619666450075806e-06,
-      "loss": 0.1723,
+      "epoch": 5.52,
+      "learning_rate": 1.4935688826902217e-05,
+      "loss": 0.2555,
       "step": 11880
     },
     {
-      "epoch": 7.72,
-      "learning_rate": 7.598007364089236e-06,
-      "loss": 0.0015,
+      "epoch": 5.53,
+      "learning_rate": 1.4920192158685883e-05,
+      "loss": 0.216,
       "step": 11890
     },
     {
-      "epoch": 7.73,
-      "learning_rate": 7.576348278102664e-06,
-      "loss": 0.0992,
+      "epoch": 5.53,
+      "learning_rate": 1.490469549046955e-05,
+      "loss": 0.0395,
       "step": 11900
     },
     {
-      "epoch": 7.74,
-      "learning_rate": 7.554689192116092e-06,
-      "loss": 0.0976,
+      "epoch": 5.54,
+      "learning_rate": 1.4889198822253216e-05,
+      "loss": 0.1207,
       "step": 11910
     },
     {
-      "epoch": 7.74,
-      "learning_rate": 7.533030106129521e-06,
-      "loss": 0.0918,
+      "epoch": 5.54,
+      "learning_rate": 1.4873702154036884e-05,
+      "loss": 0.0864,
       "step": 11920
     },
     {
-      "epoch": 7.75,
-      "learning_rate": 7.5113710201429505e-06,
-      "loss": 0.0709,
+      "epoch": 5.55,
+      "learning_rate": 1.4858205485820548e-05,
+      "loss": 0.0692,
       "step": 11930
     },
     {
-      "epoch": 7.76,
-      "learning_rate": 7.489711934156379e-06,
-      "loss": 0.2004,
+      "epoch": 5.55,
+      "learning_rate": 1.4842708817604215e-05,
+      "loss": 0.0309,
       "step": 11940
     },
     {
-      "epoch": 7.76,
-      "learning_rate": 7.468052848169807e-06,
-      "loss": 0.0302,
+      "epoch": 5.56,
+      "learning_rate": 1.4827212149387881e-05,
+      "loss": 0.2822,
       "step": 11950
     },
     {
-      "epoch": 7.77,
-      "learning_rate": 7.446393762183236e-06,
-      "loss": 0.3603,
+      "epoch": 5.56,
+      "learning_rate": 1.4811715481171548e-05,
+      "loss": 0.0721,
       "step": 11960
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 7.424734676196664e-06,
-      "loss": 0.0168,
+      "epoch": 5.56,
+      "learning_rate": 1.4796218812955214e-05,
+      "loss": 0.2282,
       "step": 11970
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 7.403075590210093e-06,
-      "loss": 0.0821,
+      "epoch": 5.57,
+      "learning_rate": 1.4780722144738882e-05,
+      "loss": 0.5436,
       "step": 11980
     },
     {
-      "epoch": 7.79,
-      "learning_rate": 7.381416504223522e-06,
-      "loss": 0.1954,
+      "epoch": 5.57,
+      "learning_rate": 1.4765225476522548e-05,
+      "loss": 0.3879,
       "step": 11990
     },
     {
-      "epoch": 7.79,
-      "learning_rate": 7.35975741823695e-06,
-      "loss": 0.1345,
+      "epoch": 5.58,
+      "learning_rate": 1.4749728808306215e-05,
+      "loss": 0.2802,
       "step": 12000
     },
     {
-      "epoch": 7.8,
-      "learning_rate": 7.338098332250379e-06,
-      "loss": 0.0208,
+      "epoch": 5.58,
+      "learning_rate": 1.473423214008988e-05,
+      "loss": 0.0243,
       "step": 12010
     },
     {
-      "epoch": 7.81,
-      "learning_rate": 7.316439246263808e-06,
-      "loss": 0.1959,
+      "epoch": 5.59,
+      "learning_rate": 1.4718735471873548e-05,
+      "loss": 0.1327,
       "step": 12020
     },
     {
-      "epoch": 7.81,
-      "learning_rate": 7.2947801602772365e-06,
-      "loss": 0.0016,
+      "epoch": 5.59,
+      "learning_rate": 1.4703238803657214e-05,
+      "loss": 0.056,
       "step": 12030
     },
     {
-      "epoch": 7.82,
-      "learning_rate": 7.273121074290665e-06,
-      "loss": 0.2701,
+      "epoch": 5.6,
+      "learning_rate": 1.4687742135440882e-05,
+      "loss": 0.284,
       "step": 12040
     },
     {
-      "epoch": 7.83,
-      "learning_rate": 7.251461988304093e-06,
-      "loss": 0.0076,
+      "epoch": 5.6,
+      "learning_rate": 1.4672245467224547e-05,
+      "loss": 0.3406,
       "step": 12050
     },
     {
-      "epoch": 7.83,
-      "learning_rate": 7.2298029023175225e-06,
-      "loss": 0.0192,
+      "epoch": 5.61,
+      "learning_rate": 1.4656748799008213e-05,
+      "loss": 0.2064,
       "step": 12060
     },
     {
-      "epoch": 7.84,
-      "learning_rate": 7.20814381633095e-06,
-      "loss": 0.0733,
+      "epoch": 5.61,
+      "learning_rate": 1.4641252130791879e-05,
+      "loss": 0.2296,
       "step": 12070
     },
     {
-      "epoch": 7.85,
-      "learning_rate": 7.18648473034438e-06,
-      "loss": 0.0395,
+      "epoch": 5.62,
+      "learning_rate": 1.4625755462575546e-05,
+      "loss": 0.1643,
       "step": 12080
     },
     {
-      "epoch": 7.85,
-      "learning_rate": 7.164825644357808e-06,
-      "loss": 0.044,
+      "epoch": 5.62,
+      "learning_rate": 1.4610258794359212e-05,
+      "loss": 0.2724,
       "step": 12090
     },
     {
-      "epoch": 7.86,
-      "learning_rate": 7.143166558371237e-06,
-      "loss": 0.1878,
+      "epoch": 5.63,
+      "learning_rate": 1.459476212614288e-05,
+      "loss": 0.1311,
       "step": 12100
     },
     {
-      "epoch": 7.87,
-      "learning_rate": 7.121507472384665e-06,
-      "loss": 0.233,
+      "epoch": 5.63,
+      "learning_rate": 1.4579265457926545e-05,
+      "loss": 0.1393,
       "step": 12110
     },
     {
-      "epoch": 7.87,
-      "learning_rate": 7.099848386398095e-06,
-      "loss": 0.1157,
+      "epoch": 5.63,
+      "learning_rate": 1.4563768789710213e-05,
+      "loss": 0.144,
       "step": 12120
     },
     {
-      "epoch": 7.88,
-      "learning_rate": 7.0781893004115225e-06,
-      "loss": 0.098,
+      "epoch": 5.64,
+      "learning_rate": 1.4548272121493879e-05,
+      "loss": 0.1708,
       "step": 12130
     },
     {
-      "epoch": 7.89,
-      "learning_rate": 7.056530214424951e-06,
-      "loss": 0.1827,
+      "epoch": 5.64,
+      "learning_rate": 1.4532775453277546e-05,
+      "loss": 0.1765,
       "step": 12140
     },
     {
-      "epoch": 7.89,
-      "learning_rate": 7.03487112843838e-06,
-      "loss": 0.2982,
+      "epoch": 5.65,
+      "learning_rate": 1.4517278785061212e-05,
+      "loss": 0.0997,
       "step": 12150
     },
     {
-      "epoch": 7.9,
-      "learning_rate": 7.0132120424518085e-06,
-      "loss": 0.1493,
+      "epoch": 5.65,
+      "learning_rate": 1.450178211684488e-05,
+      "loss": 0.2217,
       "step": 12160
     },
     {
-      "epoch": 7.91,
-      "learning_rate": 6.991552956465237e-06,
-      "loss": 0.1547,
+      "epoch": 5.66,
+      "learning_rate": 1.4486285448628545e-05,
+      "loss": 0.1893,
       "step": 12170
     },
     {
-      "epoch": 7.91,
-      "learning_rate": 6.969893870478666e-06,
-      "loss": 0.0282,
+      "epoch": 5.66,
+      "learning_rate": 1.4470788780412211e-05,
+      "loss": 0.129,
       "step": 12180
     },
     {
-      "epoch": 7.92,
-      "learning_rate": 6.948234784492095e-06,
-      "loss": 0.2962,
+      "epoch": 5.67,
+      "learning_rate": 1.4455292112195877e-05,
+      "loss": 0.3474,
       "step": 12190
     },
     {
-      "epoch": 7.92,
-      "learning_rate": 6.926575698505523e-06,
-      "loss": 0.0203,
+      "epoch": 5.67,
+      "learning_rate": 1.4439795443979544e-05,
+      "loss": 0.2942,
       "step": 12200
     },
     {
-      "epoch": 7.93,
-      "learning_rate": 6.904916612518952e-06,
-      "loss": 0.3606,
+      "epoch": 5.68,
+      "learning_rate": 1.442429877576321e-05,
+      "loss": 0.147,
       "step": 12210
     },
     {
-      "epoch": 7.94,
-      "learning_rate": 6.883257526532381e-06,
-      "loss": 0.2018,
+      "epoch": 5.68,
+      "learning_rate": 1.4408802107546878e-05,
+      "loss": 0.206,
       "step": 12220
     },
     {
-      "epoch": 7.94,
-      "learning_rate": 6.8615984405458085e-06,
-      "loss": 0.1647,
+      "epoch": 5.69,
+      "learning_rate": 1.4393305439330543e-05,
+      "loss": 0.0984,
       "step": 12230
     },
     {
-      "epoch": 7.95,
-      "learning_rate": 6.839939354559238e-06,
-      "loss": 0.0782,
+      "epoch": 5.69,
+      "learning_rate": 1.4377808771114211e-05,
+      "loss": 0.5019,
       "step": 12240
     },
     {
-      "epoch": 7.96,
-      "learning_rate": 6.818280268572666e-06,
-      "loss": 0.0471,
+      "epoch": 5.7,
+      "learning_rate": 1.4362312102897877e-05,
+      "loss": 0.1503,
       "step": 12250
     },
     {
-      "epoch": 7.96,
-      "learning_rate": 6.796621182586095e-06,
-      "loss": 0.0793,
+      "epoch": 5.7,
+      "learning_rate": 1.4346815434681544e-05,
+      "loss": 0.4062,
       "step": 12260
     },
     {
-      "epoch": 7.97,
-      "learning_rate": 6.774962096599523e-06,
-      "loss": 0.1777,
+      "epoch": 5.7,
+      "learning_rate": 1.433131876646521e-05,
+      "loss": 0.0735,
       "step": 12270
     },
     {
-      "epoch": 7.98,
-      "learning_rate": 6.753303010612953e-06,
-      "loss": 0.1029,
+      "epoch": 5.71,
+      "learning_rate": 1.4315822098248877e-05,
+      "loss": 0.3409,
       "step": 12280
     },
     {
-      "epoch": 7.98,
-      "learning_rate": 6.731643924626381e-06,
-      "loss": 0.0013,
+      "epoch": 5.71,
+      "learning_rate": 1.4300325430032543e-05,
+      "loss": 0.1813,
       "step": 12290
     },
     {
-      "epoch": 7.99,
-      "learning_rate": 6.70998483863981e-06,
-      "loss": 0.1963,
+      "epoch": 5.72,
+      "learning_rate": 1.428482876181621e-05,
+      "loss": 0.2068,
       "step": 12300
     },
     {
-      "epoch": 8.0,
-      "learning_rate": 6.688325752653238e-06,
-      "loss": 0.0781,
+      "epoch": 5.72,
+      "learning_rate": 1.4269332093599875e-05,
+      "loss": 0.2818,
       "step": 12310
     },
     {
-      "epoch": 8.0,
-      "eval_accuracy": 0.984820607175713,
-      "eval_loss": 0.07072959840297699,
-      "eval_runtime": 13.0034,
-      "eval_samples_per_second": 167.187,
-      "eval_steps_per_second": 41.835,
-      "step": 12316
-    },
-    {
-      "epoch": 8.0,
-      "learning_rate": 6.666666666666667e-06,
-      "loss": 0.1161,
+      "epoch": 5.73,
+      "learning_rate": 1.4253835425383542e-05,
+      "loss": 0.2827,
       "step": 12320
     },
     {
-      "epoch": 8.01,
-      "learning_rate": 6.645007580680095e-06,
-      "loss": 0.2413,
+      "epoch": 5.73,
+      "learning_rate": 1.423833875716721e-05,
+      "loss": 0.2609,
       "step": 12330
     },
     {
-      "epoch": 8.02,
-      "learning_rate": 6.623348494693524e-06,
-      "loss": 0.1146,
+      "epoch": 5.74,
+      "learning_rate": 1.4222842088950876e-05,
+      "loss": 0.1641,
       "step": 12340
     },
     {
-      "epoch": 8.02,
-      "learning_rate": 6.601689408706953e-06,
-      "loss": 0.0088,
+      "epoch": 5.74,
+      "learning_rate": 1.4207345420734543e-05,
+      "loss": 0.1817,
       "step": 12350
     },
     {
-      "epoch": 8.03,
-      "learning_rate": 6.580030322720381e-06,
-      "loss": 0.116,
+      "epoch": 5.75,
+      "learning_rate": 1.4191848752518209e-05,
+      "loss": 0.3664,
       "step": 12360
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 6.55837123673381e-06,
-      "loss": 0.0927,
+      "epoch": 5.75,
+      "learning_rate": 1.4176352084301876e-05,
+      "loss": 0.0603,
       "step": 12370
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 6.536712150747239e-06,
-      "loss": 0.3197,
+      "epoch": 5.76,
+      "learning_rate": 1.4160855416085542e-05,
+      "loss": 0.216,
       "step": 12380
     },
     {
-      "epoch": 8.05,
-      "learning_rate": 6.5150530647606674e-06,
-      "loss": 0.1136,
+      "epoch": 5.76,
+      "learning_rate": 1.414535874786921e-05,
+      "loss": 0.1701,
       "step": 12390
     },
     {
-      "epoch": 8.05,
-      "learning_rate": 6.493393978774096e-06,
-      "loss": 0.2585,
+      "epoch": 5.76,
+      "learning_rate": 1.4129862079652875e-05,
+      "loss": 0.2231,
       "step": 12400
     },
     {
-      "epoch": 8.06,
-      "learning_rate": 6.471734892787524e-06,
-      "loss": 0.127,
+      "epoch": 5.77,
+      "learning_rate": 1.4114365411436543e-05,
+      "loss": 0.2279,
       "step": 12410
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 6.4500758068009535e-06,
-      "loss": 0.0578,
+      "epoch": 5.77,
+      "learning_rate": 1.4098868743220209e-05,
+      "loss": 0.0225,
       "step": 12420
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 6.428416720814381e-06,
-      "loss": 0.0706,
+      "epoch": 5.78,
+      "learning_rate": 1.4083372075003874e-05,
+      "loss": 0.2637,
       "step": 12430
     },
     {
-      "epoch": 8.08,
-      "learning_rate": 6.406757634827811e-06,
-      "loss": 0.1134,
+      "epoch": 5.78,
+      "learning_rate": 1.406787540678754e-05,
+      "loss": 0.1246,
       "step": 12440
     },
     {
-      "epoch": 8.09,
-      "learning_rate": 6.385098548841239e-06,
-      "loss": 0.0335,
+      "epoch": 5.79,
+      "learning_rate": 1.4052378738571208e-05,
+      "loss": 0.0886,
       "step": 12450
     },
     {
-      "epoch": 8.09,
-      "learning_rate": 6.363439462854668e-06,
-      "loss": 0.0277,
+      "epoch": 5.79,
+      "learning_rate": 1.4036882070354874e-05,
+      "loss": 0.0999,
       "step": 12460
     },
     {
-      "epoch": 8.1,
-      "learning_rate": 6.341780376868096e-06,
-      "loss": 0.1412,
+      "epoch": 5.8,
+      "learning_rate": 1.4021385402138541e-05,
+      "loss": 0.0849,
       "step": 12470
     },
     {
-      "epoch": 8.11,
-      "learning_rate": 6.320121290881526e-06,
-      "loss": 0.07,
+      "epoch": 5.8,
+      "learning_rate": 1.4005888733922207e-05,
+      "loss": 0.0611,
       "step": 12480
     },
     {
-      "epoch": 8.11,
-      "learning_rate": 6.2984622048949534e-06,
-      "loss": 0.1496,
+      "epoch": 5.81,
+      "learning_rate": 1.3990392065705874e-05,
+      "loss": 0.3946,
       "step": 12490
     },
     {
-      "epoch": 8.12,
-      "learning_rate": 6.276803118908382e-06,
-      "loss": 0.0603,
+      "epoch": 5.81,
+      "learning_rate": 1.397489539748954e-05,
+      "loss": 0.1693,
       "step": 12500
     },
     {
-      "epoch": 8.13,
-      "learning_rate": 6.255144032921811e-06,
-      "loss": 0.2813,
+      "epoch": 5.82,
+      "learning_rate": 1.3959398729273208e-05,
+      "loss": 0.0813,
       "step": 12510
     },
     {
-      "epoch": 8.13,
-      "learning_rate": 6.2334849469352395e-06,
-      "loss": 0.1433,
+      "epoch": 5.82,
+      "learning_rate": 1.3943902061056873e-05,
+      "loss": 0.2583,
       "step": 12520
     },
     {
-      "epoch": 8.14,
-      "learning_rate": 6.211825860948668e-06,
-      "loss": 0.3679,
+      "epoch": 5.83,
+      "learning_rate": 1.3928405392840541e-05,
+      "loss": 0.083,
       "step": 12530
     },
     {
-      "epoch": 8.15,
-      "learning_rate": 6.190166774962097e-06,
-      "loss": 0.0328,
+      "epoch": 5.83,
+      "learning_rate": 1.3912908724624207e-05,
+      "loss": 0.3325,
       "step": 12540
     },
     {
-      "epoch": 8.15,
-      "learning_rate": 6.1685076889755256e-06,
-      "loss": 0.1398,
+      "epoch": 5.83,
+      "learning_rate": 1.3897412056407872e-05,
+      "loss": 0.2105,
       "step": 12550
     },
     {
-      "epoch": 8.16,
-      "learning_rate": 6.146848602988954e-06,
-      "loss": 0.1047,
+      "epoch": 5.84,
+      "learning_rate": 1.3881915388191538e-05,
+      "loss": 0.2928,
       "step": 12560
     },
     {
-      "epoch": 8.16,
-      "learning_rate": 6.125189517002383e-06,
-      "loss": 0.1391,
+      "epoch": 5.84,
+      "learning_rate": 1.3866418719975206e-05,
+      "loss": 0.2176,
       "step": 12570
     },
     {
-      "epoch": 8.17,
-      "learning_rate": 6.103530431015812e-06,
-      "loss": 0.0941,
+      "epoch": 5.85,
+      "learning_rate": 1.3850922051758872e-05,
+      "loss": 0.1962,
       "step": 12580
     },
     {
-      "epoch": 8.18,
-      "learning_rate": 6.0818713450292395e-06,
-      "loss": 0.0813,
+      "epoch": 5.85,
+      "learning_rate": 1.3835425383542539e-05,
+      "loss": 0.0099,
       "step": 12590
     },
     {
-      "epoch": 8.18,
-      "learning_rate": 6.060212259042669e-06,
-      "loss": 0.0015,
+      "epoch": 5.86,
+      "learning_rate": 1.3819928715326205e-05,
+      "loss": 0.5605,
       "step": 12600
     },
     {
-      "epoch": 8.19,
-      "learning_rate": 6.038553173056097e-06,
-      "loss": 0.1206,
+      "epoch": 5.86,
+      "learning_rate": 1.3804432047109872e-05,
+      "loss": 0.1368,
       "step": 12610
     },
     {
-      "epoch": 8.2,
-      "learning_rate": 6.016894087069526e-06,
-      "loss": 0.158,
+      "epoch": 5.87,
+      "learning_rate": 1.3788935378893538e-05,
+      "loss": 0.3033,
       "step": 12620
     },
     {
-      "epoch": 8.2,
-      "learning_rate": 5.995235001082954e-06,
-      "loss": 0.0406,
+      "epoch": 5.87,
+      "learning_rate": 1.3773438710677206e-05,
+      "loss": 0.2857,
       "step": 12630
     },
     {
-      "epoch": 8.21,
-      "learning_rate": 5.973575915096384e-06,
-      "loss": 0.1629,
+      "epoch": 5.88,
+      "learning_rate": 1.3757942042460871e-05,
+      "loss": 0.0981,
       "step": 12640
     },
     {
-      "epoch": 8.22,
-      "learning_rate": 5.9519168291098116e-06,
-      "loss": 0.0426,
+      "epoch": 5.88,
+      "learning_rate": 1.3742445374244539e-05,
+      "loss": 0.0775,
       "step": 12650
     },
     {
-      "epoch": 8.22,
-      "learning_rate": 5.930257743123241e-06,
-      "loss": 0.0673,
+      "epoch": 5.89,
+      "learning_rate": 1.3726948706028205e-05,
+      "loss": 0.2109,
       "step": 12660
     },
     {
-      "epoch": 8.23,
-      "learning_rate": 5.908598657136669e-06,
-      "loss": 0.037,
+      "epoch": 5.89,
+      "learning_rate": 1.3711452037811872e-05,
+      "loss": 0.3579,
       "step": 12670
     },
     {
-      "epoch": 8.24,
-      "learning_rate": 5.886939571150098e-06,
-      "loss": 0.0649,
+      "epoch": 5.89,
+      "learning_rate": 1.3695955369595536e-05,
+      "loss": 0.0032,
       "step": 12680
     },
     {
-      "epoch": 8.24,
-      "learning_rate": 5.865280485163526e-06,
-      "loss": 0.2742,
+      "epoch": 5.9,
+      "learning_rate": 1.3680458701379204e-05,
+      "loss": 0.1483,
       "step": 12690
     },
     {
-      "epoch": 8.25,
-      "learning_rate": 5.843621399176955e-06,
-      "loss": 0.2166,
+      "epoch": 5.9,
+      "learning_rate": 1.366496203316287e-05,
+      "loss": 0.133,
       "step": 12700
     },
     {
-      "epoch": 8.26,
-      "learning_rate": 5.821962313190384e-06,
-      "loss": 0.0853,
+      "epoch": 5.91,
+      "learning_rate": 1.3649465364946537e-05,
+      "loss": 0.3291,
       "step": 12710
     },
     {
-      "epoch": 8.26,
-      "learning_rate": 5.800303227203812e-06,
-      "loss": 0.0083,
+      "epoch": 5.91,
+      "learning_rate": 1.3633968696730203e-05,
+      "loss": 0.0629,
       "step": 12720
     },
     {
-      "epoch": 8.27,
-      "learning_rate": 5.778644141217241e-06,
-      "loss": 0.1324,
+      "epoch": 5.92,
+      "learning_rate": 1.361847202851387e-05,
+      "loss": 0.1384,
       "step": 12730
     },
     {
-      "epoch": 8.28,
-      "learning_rate": 5.75698505523067e-06,
-      "loss": 0.0543,
+      "epoch": 5.92,
+      "learning_rate": 1.3602975360297536e-05,
+      "loss": 0.2241,
       "step": 12740
     },
     {
-      "epoch": 8.28,
-      "learning_rate": 5.7353259692440976e-06,
-      "loss": 0.1978,
+      "epoch": 5.93,
+      "learning_rate": 1.3587478692081203e-05,
+      "loss": 0.1742,
       "step": 12750
     },
     {
-      "epoch": 8.29,
-      "learning_rate": 5.713666883257527e-06,
-      "loss": 0.112,
+      "epoch": 5.93,
+      "learning_rate": 1.357198202386487e-05,
+      "loss": 0.3987,
       "step": 12760
     },
     {
-      "epoch": 8.29,
-      "learning_rate": 5.692007797270955e-06,
-      "loss": 0.1806,
+      "epoch": 5.94,
+      "learning_rate": 1.3556485355648537e-05,
+      "loss": 0.1664,
       "step": 12770
     },
     {
-      "epoch": 8.3,
-      "learning_rate": 5.6703487112843845e-06,
-      "loss": 0.0336,
+      "epoch": 5.94,
+      "learning_rate": 1.3540988687432203e-05,
+      "loss": 0.1917,
       "step": 12780
     },
     {
-      "epoch": 8.31,
-      "learning_rate": 5.648689625297812e-06,
-      "loss": 0.2045,
+      "epoch": 5.95,
+      "learning_rate": 1.352549201921587e-05,
+      "loss": 0.2952,
       "step": 12790
     },
     {
-      "epoch": 8.31,
-      "learning_rate": 5.627030539311242e-06,
-      "loss": 0.0968,
+      "epoch": 5.95,
+      "learning_rate": 1.3509995350999534e-05,
+      "loss": 0.1671,
       "step": 12800
     },
     {
-      "epoch": 8.32,
-      "learning_rate": 5.60537145332467e-06,
-      "loss": 0.2564,
+      "epoch": 5.96,
+      "learning_rate": 1.3494498682783202e-05,
+      "loss": 0.1415,
       "step": 12810
     },
     {
-      "epoch": 8.33,
-      "learning_rate": 5.583712367338099e-06,
-      "loss": 0.078,
+      "epoch": 5.96,
+      "learning_rate": 1.3479002014566867e-05,
+      "loss": 0.0288,
       "step": 12820
     },
     {
-      "epoch": 8.33,
-      "learning_rate": 5.562053281351527e-06,
-      "loss": 0.2211,
+      "epoch": 5.96,
+      "learning_rate": 1.3463505346350535e-05,
+      "loss": 0.114,
       "step": 12830
     },
     {
-      "epoch": 8.34,
-      "learning_rate": 5.540394195364956e-06,
-      "loss": 0.0235,
+      "epoch": 5.97,
+      "learning_rate": 1.34480086781342e-05,
+      "loss": 0.2097,
       "step": 12840
     },
     {
-      "epoch": 8.35,
-      "learning_rate": 5.518735109378384e-06,
-      "loss": 0.0086,
+      "epoch": 5.97,
+      "learning_rate": 1.3432512009917868e-05,
+      "loss": 0.443,
       "step": 12850
     },
     {
-      "epoch": 8.35,
-      "learning_rate": 5.497076023391813e-06,
-      "loss": 0.0448,
+      "epoch": 5.98,
+      "learning_rate": 1.3417015341701534e-05,
+      "loss": 0.0946,
       "step": 12860
     },
     {
-      "epoch": 8.36,
-      "learning_rate": 5.475416937405242e-06,
-      "loss": 0.0874,
+      "epoch": 5.98,
+      "learning_rate": 1.3401518673485201e-05,
+      "loss": 0.0602,
       "step": 12870
     },
     {
-      "epoch": 8.37,
-      "learning_rate": 5.4537578514186705e-06,
-      "loss": 0.1144,
+      "epoch": 5.99,
+      "learning_rate": 1.3386022005268867e-05,
+      "loss": 0.2549,
       "step": 12880
     },
     {
-      "epoch": 8.37,
-      "learning_rate": 5.432098765432099e-06,
-      "loss": 0.0012,
+      "epoch": 5.99,
+      "learning_rate": 1.3370525337052535e-05,
+      "loss": 0.279,
       "step": 12890
     },
     {
-      "epoch": 8.38,
-      "learning_rate": 5.410439679445528e-06,
-      "loss": 0.1392,
+      "epoch": 6.0,
+      "learning_rate": 1.33550286688362e-05,
+      "loss": 0.3252,
       "step": 12900
     },
     {
-      "epoch": 8.39,
-      "learning_rate": 5.3887805934589565e-06,
-      "loss": 0.0546,
+      "epoch": 6.0,
+      "eval_accuracy": 0.9720118538030952,
+      "eval_loss": 0.12808209657669067,
+      "eval_runtime": 17.5753,
+      "eval_samples_per_second": 172.8,
+      "eval_steps_per_second": 43.243,
+      "step": 12906
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3339532000619868e-05,
+      "loss": 0.1309,
       "step": 12910
     },
     {
-      "epoch": 8.39,
-      "learning_rate": 5.367121507472384e-06,
-      "loss": 0.0058,
+      "epoch": 6.01,
+      "learning_rate": 1.3324035332403534e-05,
+      "loss": 0.2778,
       "step": 12920
     },
     {
-      "epoch": 8.4,
-      "learning_rate": 5.345462421485813e-06,
-      "loss": 0.1225,
+      "epoch": 6.01,
+      "learning_rate": 1.33085386641872e-05,
+      "loss": 0.2875,
       "step": 12930
     },
     {
-      "epoch": 8.41,
-      "learning_rate": 5.323803335499242e-06,
-      "loss": 0.0963,
+      "epoch": 6.02,
+      "learning_rate": 1.3293041995970865e-05,
+      "loss": 0.2078,
       "step": 12940
     },
     {
-      "epoch": 8.41,
-      "learning_rate": 5.30214424951267e-06,
-      "loss": 0.1001,
+      "epoch": 6.02,
+      "learning_rate": 1.3277545327754533e-05,
+      "loss": 0.2939,
       "step": 12950
     },
     {
-      "epoch": 8.42,
-      "learning_rate": 5.280485163526099e-06,
-      "loss": 0.1365,
+      "epoch": 6.03,
+      "learning_rate": 1.3262048659538199e-05,
+      "loss": 0.1103,
       "step": 12960
     },
     {
-      "epoch": 8.42,
-      "learning_rate": 5.258826077539528e-06,
-      "loss": 0.1296,
+      "epoch": 6.03,
+      "learning_rate": 1.3246551991321866e-05,
+      "loss": 0.1293,
       "step": 12970
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 5.2371669915529565e-06,
-      "loss": 0.0279,
+      "epoch": 6.03,
+      "learning_rate": 1.3231055323105532e-05,
+      "loss": 0.166,
       "step": 12980
     },
     {
-      "epoch": 8.44,
-      "learning_rate": 5.215507905566385e-06,
-      "loss": 0.0445,
+      "epoch": 6.04,
+      "learning_rate": 1.32155586548892e-05,
+      "loss": 0.0301,
       "step": 12990
     },
     {
-      "epoch": 8.44,
-      "learning_rate": 5.193848819579814e-06,
-      "loss": 0.064,
+      "epoch": 6.04,
+      "learning_rate": 1.3200061986672865e-05,
+      "loss": 0.1238,
       "step": 13000
     },
     {
-      "epoch": 8.45,
-      "learning_rate": 5.1721897335932425e-06,
-      "loss": 0.0262,
+      "epoch": 6.05,
+      "learning_rate": 1.3184565318456533e-05,
+      "loss": 0.2215,
       "step": 13010
     },
     {
-      "epoch": 8.46,
-      "learning_rate": 5.15053064760667e-06,
-      "loss": 0.2856,
+      "epoch": 6.05,
+      "learning_rate": 1.3169068650240198e-05,
+      "loss": 0.1132,
       "step": 13020
     },
     {
-      "epoch": 8.46,
-      "learning_rate": 5.1288715616201e-06,
-      "loss": 0.1308,
+      "epoch": 6.06,
+      "learning_rate": 1.3153571982023866e-05,
+      "loss": 0.2113,
       "step": 13030
     },
     {
-      "epoch": 8.47,
-      "learning_rate": 5.107212475633528e-06,
-      "loss": 0.1229,
+      "epoch": 6.06,
+      "learning_rate": 1.3138075313807532e-05,
+      "loss": 0.2348,
       "step": 13040
     },
     {
-      "epoch": 8.48,
-      "learning_rate": 5.085553389646957e-06,
-      "loss": 0.1929,
+      "epoch": 6.07,
+      "learning_rate": 1.3122578645591198e-05,
+      "loss": 0.1862,
       "step": 13050
     },
     {
-      "epoch": 8.48,
-      "learning_rate": 5.063894303660385e-06,
-      "loss": 0.1193,
+      "epoch": 6.07,
+      "learning_rate": 1.3107081977374863e-05,
+      "loss": 0.1097,
       "step": 13060
     },
     {
-      "epoch": 8.49,
-      "learning_rate": 5.042235217673815e-06,
-      "loss": 0.059,
+      "epoch": 6.08,
+      "learning_rate": 1.309158530915853e-05,
+      "loss": 0.2081,
       "step": 13070
     },
     {
-      "epoch": 8.5,
-      "learning_rate": 5.0205761316872425e-06,
-      "loss": 0.1365,
+      "epoch": 6.08,
+      "learning_rate": 1.3076088640942197e-05,
+      "loss": 0.1265,
       "step": 13080
     },
     {
-      "epoch": 8.5,
-      "learning_rate": 4.998917045700672e-06,
-      "loss": 0.052,
+      "epoch": 6.09,
+      "learning_rate": 1.3060591972725864e-05,
+      "loss": 0.1855,
       "step": 13090
     },
     {
-      "epoch": 8.51,
-      "learning_rate": 4.9772579597141e-06,
-      "loss": 0.0587,
+      "epoch": 6.09,
+      "learning_rate": 1.304509530450953e-05,
+      "loss": 0.0711,
       "step": 13100
     },
     {
-      "epoch": 8.52,
-      "learning_rate": 4.9555988737275285e-06,
-      "loss": 0.1808,
+      "epoch": 6.09,
+      "learning_rate": 1.3029598636293197e-05,
+      "loss": 0.0704,
       "step": 13110
     },
     {
-      "epoch": 8.52,
-      "learning_rate": 4.933939787740957e-06,
-      "loss": 0.071,
+      "epoch": 6.1,
+      "learning_rate": 1.3014101968076863e-05,
+      "loss": 0.2293,
       "step": 13120
     },
     {
-      "epoch": 8.53,
-      "learning_rate": 4.912280701754386e-06,
-      "loss": 0.0762,
+      "epoch": 6.1,
+      "learning_rate": 1.299860529986053e-05,
+      "loss": 0.1046,
       "step": 13130
     },
     {
-      "epoch": 8.54,
-      "learning_rate": 4.890621615767815e-06,
-      "loss": 0.0501,
+      "epoch": 6.11,
+      "learning_rate": 1.2983108631644196e-05,
+      "loss": 0.2733,
       "step": 13140
     },
     {
-      "epoch": 8.54,
-      "learning_rate": 4.868962529781243e-06,
-      "loss": 0.1167,
+      "epoch": 6.11,
+      "learning_rate": 1.2967611963427864e-05,
+      "loss": 0.2281,
       "step": 13150
     },
     {
-      "epoch": 8.55,
-      "learning_rate": 4.847303443794672e-06,
-      "loss": 0.2363,
+      "epoch": 6.12,
+      "learning_rate": 1.295211529521153e-05,
+      "loss": 0.0939,
       "step": 13160
     },
     {
-      "epoch": 8.55,
-      "learning_rate": 4.825644357808101e-06,
-      "loss": 0.1063,
+      "epoch": 6.12,
+      "learning_rate": 1.2936618626995197e-05,
+      "loss": 0.1873,
       "step": 13170
     },
     {
-      "epoch": 8.56,
-      "learning_rate": 4.803985271821529e-06,
-      "loss": 0.0592,
+      "epoch": 6.13,
+      "learning_rate": 1.2921121958778861e-05,
+      "loss": 0.141,
       "step": 13180
     },
     {
-      "epoch": 8.57,
-      "learning_rate": 4.782326185834958e-06,
-      "loss": 0.1423,
+      "epoch": 6.13,
+      "learning_rate": 1.2905625290562529e-05,
+      "loss": 0.1002,
       "step": 13190
     },
     {
-      "epoch": 8.57,
-      "learning_rate": 4.760667099848386e-06,
-      "loss": 0.1531,
+      "epoch": 6.14,
+      "learning_rate": 1.2890128622346196e-05,
+      "loss": 0.1087,
       "step": 13200
     },
     {
-      "epoch": 8.58,
-      "learning_rate": 4.739008013861815e-06,
-      "loss": 0.0018,
+      "epoch": 6.14,
+      "learning_rate": 1.2874631954129862e-05,
+      "loss": 0.1113,
       "step": 13210
     },
     {
-      "epoch": 8.59,
-      "learning_rate": 4.717348927875243e-06,
-      "loss": 0.1259,
+      "epoch": 6.15,
+      "learning_rate": 1.285913528591353e-05,
+      "loss": 0.1038,
       "step": 13220
     },
     {
-      "epoch": 8.59,
-      "learning_rate": 4.695689841888673e-06,
-      "loss": 0.0096,
+      "epoch": 6.15,
+      "learning_rate": 1.2843638617697195e-05,
+      "loss": 0.2177,
       "step": 13230
     },
     {
-      "epoch": 8.6,
-      "learning_rate": 4.674030755902101e-06,
-      "loss": 0.1893,
+      "epoch": 6.16,
+      "learning_rate": 1.2828141949480863e-05,
+      "loss": 0.0688,
       "step": 13240
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 4.65237166991553e-06,
-      "loss": 0.1007,
+      "epoch": 6.16,
+      "learning_rate": 1.2812645281264529e-05,
+      "loss": 0.037,
       "step": 13250
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 4.630712583928958e-06,
-      "loss": 0.0186,
+      "epoch": 6.16,
+      "learning_rate": 1.2797148613048196e-05,
+      "loss": 0.1282,
       "step": 13260
     },
     {
-      "epoch": 8.62,
-      "learning_rate": 4.6090534979423875e-06,
-      "loss": 0.1633,
+      "epoch": 6.17,
+      "learning_rate": 1.2781651944831862e-05,
+      "loss": 0.2578,
       "step": 13270
     },
     {
-      "epoch": 8.63,
-      "learning_rate": 4.587394411955815e-06,
-      "loss": 0.1736,
+      "epoch": 6.17,
+      "learning_rate": 1.276615527661553e-05,
+      "loss": 0.4221,
       "step": 13280
     },
     {
-      "epoch": 8.63,
-      "learning_rate": 4.565735325969244e-06,
-      "loss": 0.1066,
+      "epoch": 6.18,
+      "learning_rate": 1.2750658608399195e-05,
+      "loss": 0.0246,
       "step": 13290
     },
     {
-      "epoch": 8.64,
-      "learning_rate": 4.544076239982673e-06,
-      "loss": 0.2152,
+      "epoch": 6.18,
+      "learning_rate": 1.2735161940182861e-05,
+      "loss": 0.1784,
       "step": 13300
     },
     {
-      "epoch": 8.65,
-      "learning_rate": 4.522417153996101e-06,
-      "loss": 0.1314,
+      "epoch": 6.19,
+      "learning_rate": 1.2719665271966527e-05,
+      "loss": 0.192,
       "step": 13310
     },
     {
-      "epoch": 8.65,
-      "learning_rate": 4.50075806800953e-06,
-      "loss": 0.1044,
+      "epoch": 6.19,
+      "learning_rate": 1.2704168603750194e-05,
+      "loss": 0.1078,
       "step": 13320
     },
     {
-      "epoch": 8.66,
-      "learning_rate": 4.479098982022959e-06,
-      "loss": 0.0462,
+      "epoch": 6.2,
+      "learning_rate": 1.268867193553386e-05,
+      "loss": 0.0037,
       "step": 13330
     },
     {
-      "epoch": 8.67,
-      "learning_rate": 4.4574398960363874e-06,
-      "loss": 0.1103,
+      "epoch": 6.2,
+      "learning_rate": 1.2673175267317527e-05,
+      "loss": 0.2057,
       "step": 13340
     },
     {
-      "epoch": 8.67,
-      "learning_rate": 4.435780810049816e-06,
-      "loss": 0.1436,
+      "epoch": 6.21,
+      "learning_rate": 1.2657678599101193e-05,
+      "loss": 0.3116,
       "step": 13350
     },
     {
-      "epoch": 8.68,
-      "learning_rate": 4.414121724063245e-06,
-      "loss": 0.0242,
+      "epoch": 6.21,
+      "learning_rate": 1.264218193088486e-05,
+      "loss": 0.003,
       "step": 13360
     },
     {
-      "epoch": 8.68,
-      "learning_rate": 4.3924626380766735e-06,
-      "loss": 0.1488,
+      "epoch": 6.22,
+      "learning_rate": 1.2626685262668527e-05,
+      "loss": 0.2409,
       "step": 13370
     },
     {
-      "epoch": 8.69,
-      "learning_rate": 4.370803552090101e-06,
-      "loss": 0.2303,
+      "epoch": 6.22,
+      "learning_rate": 1.2611188594452194e-05,
+      "loss": 0.1635,
       "step": 13380
     },
     {
-      "epoch": 8.7,
-      "learning_rate": 4.349144466103531e-06,
-      "loss": 0.1304,
+      "epoch": 6.23,
+      "learning_rate": 1.259569192623586e-05,
+      "loss": 0.2466,
       "step": 13390
     },
     {
-      "epoch": 8.7,
-      "learning_rate": 4.327485380116959e-06,
-      "loss": 0.0573,
+      "epoch": 6.23,
+      "learning_rate": 1.2580195258019527e-05,
+      "loss": 0.3034,
       "step": 13400
     },
     {
-      "epoch": 8.71,
-      "learning_rate": 4.305826294130388e-06,
-      "loss": 0.0173,
+      "epoch": 6.23,
+      "learning_rate": 1.2564698589803193e-05,
+      "loss": 0.0746,
       "step": 13410
     },
     {
-      "epoch": 8.72,
-      "learning_rate": 4.284167208143816e-06,
-      "loss": 0.0028,
+      "epoch": 6.24,
+      "learning_rate": 1.254920192158686e-05,
+      "loss": 0.1879,
       "step": 13420
     },
     {
-      "epoch": 8.72,
-      "learning_rate": 4.262508122157246e-06,
-      "loss": 0.0278,
+      "epoch": 6.24,
+      "learning_rate": 1.2533705253370525e-05,
+      "loss": 0.2697,
       "step": 13430
     },
     {
-      "epoch": 8.73,
-      "learning_rate": 4.2408490361706735e-06,
-      "loss": 0.0493,
+      "epoch": 6.25,
+      "learning_rate": 1.2518208585154192e-05,
+      "loss": 0.128,
       "step": 13440
     },
     {
-      "epoch": 8.74,
-      "learning_rate": 4.219189950184103e-06,
-      "loss": 0.0705,
+      "epoch": 6.25,
+      "learning_rate": 1.2502711916937858e-05,
+      "loss": 0.2179,
       "step": 13450
     },
     {
-      "epoch": 8.74,
-      "learning_rate": 4.197530864197531e-06,
-      "loss": 0.0259,
+      "epoch": 6.26,
+      "learning_rate": 1.2487215248721525e-05,
+      "loss": 0.1675,
       "step": 13460
     },
     {
-      "epoch": 8.75,
-      "learning_rate": 4.1758717782109595e-06,
-      "loss": 0.1553,
+      "epoch": 6.26,
+      "learning_rate": 1.2471718580505191e-05,
+      "loss": 0.2036,
       "step": 13470
     },
     {
-      "epoch": 8.76,
-      "learning_rate": 4.154212692224388e-06,
-      "loss": 0.107,
+      "epoch": 6.27,
+      "learning_rate": 1.2456221912288859e-05,
+      "loss": 0.176,
       "step": 13480
     },
     {
-      "epoch": 8.76,
-      "learning_rate": 4.132553606237817e-06,
-      "loss": 0.0314,
+      "epoch": 6.27,
+      "learning_rate": 1.2440725244072524e-05,
+      "loss": 0.0898,
       "step": 13490
     },
     {
-      "epoch": 8.77,
-      "learning_rate": 4.1108945202512456e-06,
-      "loss": 0.1201,
+      "epoch": 6.28,
+      "learning_rate": 1.2425228575856192e-05,
+      "loss": 0.2254,
       "step": 13500
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 4.089235434264674e-06,
-      "loss": 0.0249,
+      "epoch": 6.28,
+      "learning_rate": 1.2409731907639858e-05,
+      "loss": 0.0578,
       "step": 13510
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 4.067576348278103e-06,
-      "loss": 0.0269,
+      "epoch": 6.29,
+      "learning_rate": 1.2394235239423525e-05,
+      "loss": 0.0738,
       "step": 13520
     },
     {
-      "epoch": 8.79,
-      "learning_rate": 4.045917262291532e-06,
-      "loss": 0.1419,
+      "epoch": 6.29,
+      "learning_rate": 1.2378738571207191e-05,
+      "loss": 0.3309,
       "step": 13530
     },
     {
-      "epoch": 8.8,
-      "learning_rate": 4.0242581763049595e-06,
-      "loss": 0.2158,
+      "epoch": 6.29,
+      "learning_rate": 1.2363241902990859e-05,
+      "loss": 0.3371,
       "step": 13540
     },
     {
-      "epoch": 8.8,
-      "learning_rate": 4.002599090318389e-06,
-      "loss": 0.1973,
+      "epoch": 6.3,
+      "learning_rate": 1.2347745234774523e-05,
+      "loss": 0.1685,
       "step": 13550
     },
     {
-      "epoch": 8.81,
-      "learning_rate": 3.980940004331817e-06,
-      "loss": 0.2355,
+      "epoch": 6.3,
+      "learning_rate": 1.233224856655819e-05,
+      "loss": 0.1555,
       "step": 13560
     },
     {
-      "epoch": 8.81,
-      "learning_rate": 3.959280918345246e-06,
-      "loss": 0.0418,
+      "epoch": 6.31,
+      "learning_rate": 1.2316751898341856e-05,
+      "loss": 0.2535,
       "step": 13570
     },
     {
-      "epoch": 8.82,
-      "learning_rate": 3.937621832358674e-06,
-      "loss": 0.1702,
+      "epoch": 6.31,
+      "learning_rate": 1.2301255230125523e-05,
+      "loss": 0.0701,
       "step": 13580
     },
     {
-      "epoch": 8.83,
-      "learning_rate": 3.915962746372104e-06,
-      "loss": 0.1813,
+      "epoch": 6.32,
+      "learning_rate": 1.228575856190919e-05,
+      "loss": 0.1508,
       "step": 13590
     },
     {
-      "epoch": 8.83,
-      "learning_rate": 3.8943036603855316e-06,
-      "loss": 0.089,
+      "epoch": 6.32,
+      "learning_rate": 1.2270261893692857e-05,
+      "loss": 0.2139,
       "step": 13600
     },
     {
-      "epoch": 8.84,
-      "learning_rate": 3.872644574398961e-06,
-      "loss": 0.0309,
+      "epoch": 6.33,
+      "learning_rate": 1.2254765225476522e-05,
+      "loss": 0.005,
       "step": 13610
     },
     {
-      "epoch": 8.85,
-      "learning_rate": 3.850985488412389e-06,
-      "loss": 0.0278,
+      "epoch": 6.33,
+      "learning_rate": 1.223926855726019e-05,
+      "loss": 0.1364,
       "step": 13620
     },
     {
-      "epoch": 8.85,
-      "learning_rate": 3.829326402425818e-06,
-      "loss": 0.0381,
+      "epoch": 6.34,
+      "learning_rate": 1.2223771889043856e-05,
+      "loss": 0.2343,
       "step": 13630
     },
     {
-      "epoch": 8.86,
-      "learning_rate": 3.8076673164392463e-06,
-      "loss": 0.014,
+      "epoch": 6.34,
+      "learning_rate": 1.2208275220827523e-05,
+      "loss": 0.2291,
       "step": 13640
     },
     {
-      "epoch": 8.87,
-      "learning_rate": 3.7860082304526746e-06,
-      "loss": 0.109,
+      "epoch": 6.35,
+      "learning_rate": 1.2192778552611189e-05,
+      "loss": 0.3999,
       "step": 13650
     },
     {
-      "epoch": 8.87,
-      "learning_rate": 3.7643491444661037e-06,
-      "loss": 0.3331,
+      "epoch": 6.35,
+      "learning_rate": 1.2177281884394856e-05,
+      "loss": 0.2729,
       "step": 13660
     },
     {
-      "epoch": 8.88,
-      "learning_rate": 3.742690058479532e-06,
-      "loss": 0.0013,
+      "epoch": 6.36,
+      "learning_rate": 1.2161785216178522e-05,
+      "loss": 0.2382,
       "step": 13670
     },
     {
-      "epoch": 8.89,
-      "learning_rate": 3.7210309724929606e-06,
-      "loss": 0.016,
+      "epoch": 6.36,
+      "learning_rate": 1.2146288547962188e-05,
+      "loss": 0.2228,
       "step": 13680
     },
     {
-      "epoch": 8.89,
-      "learning_rate": 3.6993718865063893e-06,
-      "loss": 0.1149,
+      "epoch": 6.36,
+      "learning_rate": 1.2130791879745854e-05,
+      "loss": 0.1734,
       "step": 13690
     },
     {
-      "epoch": 8.9,
-      "learning_rate": 3.677712800519818e-06,
-      "loss": 0.3126,
+      "epoch": 6.37,
+      "learning_rate": 1.2115295211529521e-05,
+      "loss": 0.1089,
       "step": 13700
     },
     {
-      "epoch": 8.91,
-      "learning_rate": 3.6560537145332467e-06,
-      "loss": 0.2456,
+      "epoch": 6.37,
+      "learning_rate": 1.2099798543313187e-05,
+      "loss": 0.0143,
       "step": 13710
     },
     {
-      "epoch": 8.91,
-      "learning_rate": 3.6343946285466754e-06,
-      "loss": 0.0794,
+      "epoch": 6.38,
+      "learning_rate": 1.2084301875096855e-05,
+      "loss": 0.0997,
       "step": 13720
     },
     {
-      "epoch": 8.92,
-      "learning_rate": 3.612735542560104e-06,
-      "loss": 0.1407,
+      "epoch": 6.38,
+      "learning_rate": 1.206880520688052e-05,
+      "loss": 0.0652,
       "step": 13730
     },
     {
-      "epoch": 8.92,
-      "learning_rate": 3.5910764565735327e-06,
-      "loss": 0.1953,
+      "epoch": 6.39,
+      "learning_rate": 1.2053308538664188e-05,
+      "loss": 0.4293,
       "step": 13740
     },
     {
-      "epoch": 8.93,
-      "learning_rate": 3.5694173705869614e-06,
-      "loss": 0.2692,
+      "epoch": 6.39,
+      "learning_rate": 1.2037811870447854e-05,
+      "loss": 0.122,
       "step": 13750
     },
     {
-      "epoch": 8.94,
-      "learning_rate": 3.5477582846003897e-06,
-      "loss": 0.235,
+      "epoch": 6.4,
+      "learning_rate": 1.2022315202231521e-05,
+      "loss": 0.1703,
       "step": 13760
     },
     {
-      "epoch": 8.94,
-      "learning_rate": 3.5260991986138184e-06,
-      "loss": 0.1149,
+      "epoch": 6.4,
+      "learning_rate": 1.2006818534015187e-05,
+      "loss": 0.0039,
       "step": 13770
     },
     {
-      "epoch": 8.95,
-      "learning_rate": 3.504440112627247e-06,
-      "loss": 0.1316,
+      "epoch": 6.41,
+      "learning_rate": 1.1991321865798854e-05,
+      "loss": 0.2788,
       "step": 13780
     },
     {
-      "epoch": 8.96,
-      "learning_rate": 3.4827810266406757e-06,
-      "loss": 0.0752,
+      "epoch": 6.41,
+      "learning_rate": 1.197582519758252e-05,
+      "loss": 0.1009,
       "step": 13790
     },
     {
-      "epoch": 8.96,
-      "learning_rate": 3.4611219406541044e-06,
-      "loss": 0.0281,
+      "epoch": 6.42,
+      "learning_rate": 1.1960328529366186e-05,
+      "loss": 0.0464,
       "step": 13800
     },
     {
-      "epoch": 8.97,
-      "learning_rate": 3.439462854667533e-06,
-      "loss": 0.161,
+      "epoch": 6.42,
+      "learning_rate": 1.1944831861149852e-05,
+      "loss": 0.141,
       "step": 13810
     },
     {
-      "epoch": 8.98,
-      "learning_rate": 3.417803768680962e-06,
-      "loss": 0.0827,
+      "epoch": 6.42,
+      "learning_rate": 1.192933519293352e-05,
+      "loss": 0.122,
       "step": 13820
     },
     {
-      "epoch": 8.98,
-      "learning_rate": 3.3961446826943905e-06,
-      "loss": 0.0172,
+      "epoch": 6.43,
+      "learning_rate": 1.1913838524717185e-05,
+      "loss": 0.1549,
       "step": 13830
     },
     {
-      "epoch": 8.99,
-      "learning_rate": 3.374485596707819e-06,
-      "loss": 0.1149,
+      "epoch": 6.43,
+      "learning_rate": 1.1899891523322486e-05,
+      "loss": 0.2195,
       "step": 13840
     },
     {
-      "epoch": 9.0,
-      "learning_rate": 3.3528265107212474e-06,
-      "loss": 0.0317,
+      "epoch": 6.44,
+      "learning_rate": 1.1884394855106152e-05,
+      "loss": 0.3122,
       "step": 13850
     },
     {
-      "epoch": 9.0,
-      "eval_accuracy": 0.984820607175713,
-      "eval_loss": 0.0851309597492218,
-      "eval_runtime": 12.8784,
-      "eval_samples_per_second": 168.809,
-      "eval_steps_per_second": 42.241,
-      "step": 13855
-    },
-    {
-      "epoch": 9.0,
-      "learning_rate": 3.331167424734676e-06,
-      "loss": 0.2191,
+      "epoch": 6.44,
+      "learning_rate": 1.1868898186889819e-05,
+      "loss": 0.2803,
       "step": 13860
     },
     {
-      "epoch": 9.01,
-      "learning_rate": 3.309508338748105e-06,
-      "loss": 0.1574,
+      "epoch": 6.45,
+      "learning_rate": 1.1853401518673485e-05,
+      "loss": 0.0183,
       "step": 13870
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 3.2878492527615335e-06,
-      "loss": 0.0723,
+      "epoch": 6.45,
+      "learning_rate": 1.1837904850457152e-05,
+      "loss": 0.1637,
       "step": 13880
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 3.266190166774962e-06,
-      "loss": 0.0881,
+      "epoch": 6.46,
+      "learning_rate": 1.1822408182240818e-05,
+      "loss": 0.1848,
       "step": 13890
     },
     {
-      "epoch": 9.03,
-      "learning_rate": 3.244531080788391e-06,
-      "loss": 0.1085,
+      "epoch": 6.46,
+      "learning_rate": 1.1806911514024486e-05,
+      "loss": 0.2463,
       "step": 13900
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 3.2228719948018195e-06,
-      "loss": 0.097,
+      "epoch": 6.47,
+      "learning_rate": 1.1791414845808151e-05,
+      "loss": 0.0659,
       "step": 13910
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 3.2012129088152482e-06,
-      "loss": 0.0165,
+      "epoch": 6.47,
+      "learning_rate": 1.1775918177591819e-05,
+      "loss": 0.0059,
       "step": 13920
     },
     {
-      "epoch": 9.05,
-      "learning_rate": 3.179553822828677e-06,
-      "loss": 0.1146,
+      "epoch": 6.48,
+      "learning_rate": 1.1760421509375485e-05,
+      "loss": 0.1391,
       "step": 13930
     },
     {
-      "epoch": 9.05,
-      "learning_rate": 3.157894736842105e-06,
-      "loss": 0.0318,
+      "epoch": 6.48,
+      "learning_rate": 1.1744924841159152e-05,
+      "loss": 0.0399,
       "step": 13940
     },
     {
-      "epoch": 9.06,
-      "learning_rate": 3.136235650855534e-06,
-      "loss": 0.0328,
+      "epoch": 6.49,
+      "learning_rate": 1.1729428172942816e-05,
+      "loss": 0.0984,
       "step": 13950
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 3.1145765648689625e-06,
-      "loss": 0.0659,
+      "epoch": 6.49,
+      "learning_rate": 1.1713931504726484e-05,
+      "loss": 0.2245,
       "step": 13960
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 3.0929174788823912e-06,
-      "loss": 0.0173,
+      "epoch": 6.49,
+      "learning_rate": 1.169843483651015e-05,
+      "loss": 0.0638,
       "step": 13970
     },
     {
-      "epoch": 9.08,
-      "learning_rate": 3.07125839289582e-06,
-      "loss": 0.0772,
+      "epoch": 6.5,
+      "learning_rate": 1.1682938168293817e-05,
+      "loss": 0.1447,
       "step": 13980
     },
     {
-      "epoch": 9.09,
-      "learning_rate": 3.0495993069092486e-06,
-      "loss": 0.002,
+      "epoch": 6.5,
+      "learning_rate": 1.1667441500077483e-05,
+      "loss": 0.1398,
       "step": 13990
     },
     {
-      "epoch": 9.09,
-      "learning_rate": 3.0279402209226773e-06,
-      "loss": 0.326,
+      "epoch": 6.51,
+      "learning_rate": 1.165194483186115e-05,
+      "loss": 0.0596,
       "step": 14000
     },
     {
-      "epoch": 9.1,
-      "learning_rate": 3.006281134936106e-06,
-      "loss": 0.0963,
+      "epoch": 6.51,
+      "learning_rate": 1.1636448163644816e-05,
+      "loss": 0.2698,
       "step": 14010
     },
     {
-      "epoch": 9.11,
-      "learning_rate": 2.9846220489495346e-06,
-      "loss": 0.0407,
+      "epoch": 6.52,
+      "learning_rate": 1.1620951495428484e-05,
+      "loss": 0.3052,
       "step": 14020
     },
     {
-      "epoch": 9.11,
-      "learning_rate": 2.962962962962963e-06,
-      "loss": 0.0449,
+      "epoch": 6.52,
+      "learning_rate": 1.160545482721215e-05,
+      "loss": 0.1935,
       "step": 14030
     },
     {
-      "epoch": 9.12,
-      "learning_rate": 2.9413038769763916e-06,
-      "loss": 0.196,
+      "epoch": 6.53,
+      "learning_rate": 1.1589958158995817e-05,
+      "loss": 0.2589,
       "step": 14040
     },
     {
-      "epoch": 9.13,
-      "learning_rate": 2.9196447909898203e-06,
-      "loss": 0.0058,
+      "epoch": 6.53,
+      "learning_rate": 1.1574461490779483e-05,
+      "loss": 0.3373,
       "step": 14050
     },
     {
-      "epoch": 9.13,
-      "learning_rate": 2.897985705003249e-06,
-      "loss": 0.0504,
+      "epoch": 6.54,
+      "learning_rate": 1.155896482256315e-05,
+      "loss": 0.3851,
       "step": 14060
     },
     {
-      "epoch": 9.14,
-      "learning_rate": 2.8763266190166776e-06,
-      "loss": 0.2405,
+      "epoch": 6.54,
+      "learning_rate": 1.1543468154346816e-05,
+      "loss": 0.1823,
       "step": 14070
     },
     {
-      "epoch": 9.15,
-      "learning_rate": 2.8546675330301063e-06,
-      "loss": 0.2004,
+      "epoch": 6.55,
+      "learning_rate": 1.1527971486130482e-05,
+      "loss": 0.1581,
       "step": 14080
     },
     {
-      "epoch": 9.15,
-      "learning_rate": 2.833008447043535e-06,
-      "loss": 0.1501,
+      "epoch": 6.55,
+      "learning_rate": 1.1512474817914148e-05,
+      "loss": 0.1346,
       "step": 14090
     },
     {
-      "epoch": 9.16,
-      "learning_rate": 2.8113493610569637e-06,
-      "loss": 0.0012,
+      "epoch": 6.56,
+      "learning_rate": 1.1496978149697815e-05,
+      "loss": 0.053,
       "step": 14100
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 2.789690275070392e-06,
-      "loss": 0.0938,
+      "epoch": 6.56,
+      "learning_rate": 1.148148148148148e-05,
+      "loss": 0.1271,
       "step": 14110
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 2.7680311890838206e-06,
-      "loss": 0.1683,
+      "epoch": 6.56,
+      "learning_rate": 1.1465984813265148e-05,
+      "loss": 0.0857,
       "step": 14120
     },
     {
-      "epoch": 9.18,
-      "learning_rate": 2.7463721030972493e-06,
-      "loss": 0.0288,
+      "epoch": 6.57,
+      "learning_rate": 1.1450488145048814e-05,
+      "loss": 0.1693,
       "step": 14130
     },
     {
-      "epoch": 9.18,
-      "learning_rate": 2.724713017110678e-06,
-      "loss": 0.0936,
+      "epoch": 6.57,
+      "learning_rate": 1.1434991476832482e-05,
+      "loss": 0.335,
       "step": 14140
     },
     {
-      "epoch": 9.19,
-      "learning_rate": 2.7030539311241067e-06,
-      "loss": 0.0241,
+      "epoch": 6.58,
+      "learning_rate": 1.1419494808616147e-05,
+      "loss": 0.2926,
       "step": 14150
     },
     {
-      "epoch": 9.2,
-      "learning_rate": 2.6813948451375354e-06,
-      "loss": 0.1308,
+      "epoch": 6.58,
+      "learning_rate": 1.1403998140399815e-05,
+      "loss": 0.3188,
       "step": 14160
     },
     {
-      "epoch": 9.2,
-      "learning_rate": 2.659735759150964e-06,
-      "loss": 0.0011,
+      "epoch": 6.59,
+      "learning_rate": 1.138850147218348e-05,
+      "loss": 0.3365,
       "step": 14170
     },
     {
-      "epoch": 9.21,
-      "learning_rate": 2.6380766731643928e-06,
-      "loss": 0.0892,
+      "epoch": 6.59,
+      "learning_rate": 1.1373004803967148e-05,
+      "loss": 0.2765,
       "step": 14180
     },
     {
-      "epoch": 9.22,
-      "learning_rate": 2.6164175871778214e-06,
-      "loss": 0.0683,
+      "epoch": 6.6,
+      "learning_rate": 1.1357508135750814e-05,
+      "loss": 0.2927,
       "step": 14190
     },
     {
-      "epoch": 9.22,
-      "learning_rate": 2.5947585011912497e-06,
-      "loss": 0.0821,
+      "epoch": 6.6,
+      "learning_rate": 1.134201146753448e-05,
+      "loss": 0.4226,
       "step": 14200
     },
     {
-      "epoch": 9.23,
-      "learning_rate": 2.5730994152046784e-06,
-      "loss": 0.1705,
+      "epoch": 6.61,
+      "learning_rate": 1.1326514799318145e-05,
+      "loss": 0.3253,
       "step": 14210
     },
     {
-      "epoch": 9.24,
-      "learning_rate": 2.551440329218107e-06,
-      "loss": 0.2017,
+      "epoch": 6.61,
+      "learning_rate": 1.1311018131101813e-05,
+      "loss": 0.1654,
       "step": 14220
     },
     {
-      "epoch": 9.24,
-      "learning_rate": 2.5297812432315358e-06,
-      "loss": 0.2258,
+      "epoch": 6.62,
+      "learning_rate": 1.1295521462885479e-05,
+      "loss": 0.2147,
       "step": 14230
     },
     {
-      "epoch": 9.25,
-      "learning_rate": 2.5081221572449644e-06,
-      "loss": 0.0655,
+      "epoch": 6.62,
+      "learning_rate": 1.1280024794669146e-05,
+      "loss": 0.0685,
       "step": 14240
     },
     {
-      "epoch": 9.26,
-      "learning_rate": 2.486463071258393e-06,
-      "loss": 0.0019,
+      "epoch": 6.62,
+      "learning_rate": 1.1264528126452812e-05,
+      "loss": 0.1078,
       "step": 14250
     },
     {
-      "epoch": 9.26,
-      "learning_rate": 2.464803985271822e-06,
-      "loss": 0.1095,
+      "epoch": 6.63,
+      "learning_rate": 1.124903145823648e-05,
+      "loss": 0.0877,
       "step": 14260
     },
     {
-      "epoch": 9.27,
-      "learning_rate": 2.4431448992852505e-06,
-      "loss": 0.0099,
+      "epoch": 6.63,
+      "learning_rate": 1.1233534790020145e-05,
+      "loss": 0.2331,
       "step": 14270
     },
     {
-      "epoch": 9.28,
-      "learning_rate": 2.421485813298679e-06,
-      "loss": 0.1379,
+      "epoch": 6.64,
+      "learning_rate": 1.1218038121803813e-05,
+      "loss": 0.1169,
       "step": 14280
     },
     {
-      "epoch": 9.28,
-      "learning_rate": 2.3998267273121074e-06,
-      "loss": 0.0783,
+      "epoch": 6.64,
+      "learning_rate": 1.1202541453587479e-05,
+      "loss": 0.3452,
       "step": 14290
     },
     {
-      "epoch": 9.29,
-      "learning_rate": 2.378167641325536e-06,
-      "loss": 0.0081,
+      "epoch": 6.65,
+      "learning_rate": 1.1187044785371146e-05,
+      "loss": 0.0466,
       "step": 14300
     },
     {
-      "epoch": 9.3,
-      "learning_rate": 2.356508555338965e-06,
-      "loss": 0.0195,
+      "epoch": 6.65,
+      "learning_rate": 1.1171548117154812e-05,
+      "loss": 0.0307,
       "step": 14310
     },
     {
-      "epoch": 9.3,
-      "learning_rate": 2.3348494693523935e-06,
-      "loss": 0.0019,
+      "epoch": 6.66,
+      "learning_rate": 1.115605144893848e-05,
+      "loss": 0.1214,
       "step": 14320
     },
     {
-      "epoch": 9.31,
-      "learning_rate": 2.313190383365822e-06,
-      "loss": 0.0339,
+      "epoch": 6.66,
+      "learning_rate": 1.1140554780722145e-05,
+      "loss": 0.1335,
       "step": 14330
     },
     {
-      "epoch": 9.31,
-      "learning_rate": 2.291531297379251e-06,
-      "loss": 0.0434,
+      "epoch": 6.67,
+      "learning_rate": 1.1125058112505811e-05,
+      "loss": 0.2897,
       "step": 14340
     },
     {
-      "epoch": 9.32,
-      "learning_rate": 2.2698722113926796e-06,
-      "loss": 0.1738,
+      "epoch": 6.67,
+      "learning_rate": 1.1109561444289478e-05,
+      "loss": 0.2502,
       "step": 14350
     },
     {
-      "epoch": 9.33,
-      "learning_rate": 2.2482131254061082e-06,
-      "loss": 0.1285,
+      "epoch": 6.68,
+      "learning_rate": 1.1094064776073144e-05,
+      "loss": 0.1572,
       "step": 14360
     },
     {
-      "epoch": 9.33,
-      "learning_rate": 2.226554039419537e-06,
-      "loss": 0.0907,
+      "epoch": 6.68,
+      "learning_rate": 1.1078568107856812e-05,
+      "loss": 0.216,
       "step": 14370
     },
     {
-      "epoch": 9.34,
-      "learning_rate": 2.204894953432965e-06,
-      "loss": 0.0349,
+      "epoch": 6.69,
+      "learning_rate": 1.1063071439640477e-05,
+      "loss": 0.0477,
       "step": 14380
     },
     {
-      "epoch": 9.35,
-      "learning_rate": 2.183235867446394e-06,
-      "loss": 0.0509,
+      "epoch": 6.69,
+      "learning_rate": 1.1047574771424145e-05,
+      "loss": 0.2136,
       "step": 14390
     },
     {
-      "epoch": 9.35,
-      "learning_rate": 2.1615767814598226e-06,
-      "loss": 0.0015,
+      "epoch": 6.69,
+      "learning_rate": 1.103207810320781e-05,
+      "loss": 0.0107,
       "step": 14400
     },
     {
-      "epoch": 9.36,
-      "learning_rate": 2.1399176954732512e-06,
-      "loss": 0.0035,
+      "epoch": 6.7,
+      "learning_rate": 1.1016581434991478e-05,
+      "loss": 0.1131,
       "step": 14410
     },
     {
-      "epoch": 9.37,
-      "learning_rate": 2.11825860948668e-06,
-      "loss": 0.1029,
+      "epoch": 6.7,
+      "learning_rate": 1.1001084766775144e-05,
+      "loss": 0.11,
       "step": 14420
     },
     {
-      "epoch": 9.37,
-      "learning_rate": 2.0965995235001086e-06,
-      "loss": 0.1722,
+      "epoch": 6.71,
+      "learning_rate": 1.0985588098558811e-05,
+      "loss": 0.1941,
       "step": 14430
     },
     {
-      "epoch": 9.38,
-      "learning_rate": 2.0749404375135373e-06,
-      "loss": 0.1884,
+      "epoch": 6.71,
+      "learning_rate": 1.0970091430342477e-05,
+      "loss": 0.1927,
       "step": 14440
     },
     {
-      "epoch": 9.39,
-      "learning_rate": 2.053281351526966e-06,
-      "loss": 0.153,
+      "epoch": 6.72,
+      "learning_rate": 1.0954594762126143e-05,
+      "loss": 0.1965,
       "step": 14450
     },
     {
-      "epoch": 9.39,
-      "learning_rate": 2.0316222655403942e-06,
-      "loss": 0.0016,
+      "epoch": 6.72,
+      "learning_rate": 1.0939098093909809e-05,
+      "loss": 0.0492,
       "step": 14460
     },
     {
-      "epoch": 9.4,
-      "learning_rate": 2.0099631795538225e-06,
-      "loss": 0.1055,
+      "epoch": 6.73,
+      "learning_rate": 1.0923601425693476e-05,
+      "loss": 0.1702,
       "step": 14470
     },
     {
-      "epoch": 9.41,
-      "learning_rate": 1.988304093567251e-06,
-      "loss": 0.0803,
+      "epoch": 6.73,
+      "learning_rate": 1.0908104757477142e-05,
+      "loss": 0.2058,
       "step": 14480
     },
     {
-      "epoch": 9.41,
-      "learning_rate": 1.96664500758068e-06,
-      "loss": 0.0262,
+      "epoch": 6.74,
+      "learning_rate": 1.089260808926081e-05,
+      "loss": 0.0383,
       "step": 14490
     },
     {
-      "epoch": 9.42,
-      "learning_rate": 1.9449859215941086e-06,
-      "loss": 0.0813,
+      "epoch": 6.74,
+      "learning_rate": 1.0877111421044475e-05,
+      "loss": 0.2255,
       "step": 14500
     },
     {
-      "epoch": 9.43,
-      "learning_rate": 1.9233268356075373e-06,
-      "loss": 0.1877,
+      "epoch": 6.75,
+      "learning_rate": 1.0861614752828143e-05,
+      "loss": 0.1561,
       "step": 14510
     },
     {
-      "epoch": 9.43,
-      "learning_rate": 1.9016677496209661e-06,
-      "loss": 0.0272,
+      "epoch": 6.75,
+      "learning_rate": 1.0846118084611809e-05,
+      "loss": 0.3958,
       "step": 14520
     },
     {
-      "epoch": 9.44,
-      "learning_rate": 1.8800086636343948e-06,
-      "loss": 0.1979,
+      "epoch": 6.75,
+      "learning_rate": 1.0830621416395476e-05,
+      "loss": 0.1192,
       "step": 14530
     },
     {
-      "epoch": 9.44,
-      "learning_rate": 1.8583495776478233e-06,
-      "loss": 0.1209,
+      "epoch": 6.76,
+      "learning_rate": 1.0815124748179142e-05,
+      "loss": 0.0829,
       "step": 14540
     },
     {
-      "epoch": 9.45,
-      "learning_rate": 1.836690491661252e-06,
-      "loss": 0.0729,
+      "epoch": 6.76,
+      "learning_rate": 1.079962807996281e-05,
+      "loss": 0.1938,
       "step": 14550
     },
     {
-      "epoch": 9.46,
-      "learning_rate": 1.8150314056746807e-06,
-      "loss": 0.1729,
+      "epoch": 6.77,
+      "learning_rate": 1.0784131411746475e-05,
+      "loss": 0.0591,
       "step": 14560
     },
     {
-      "epoch": 9.46,
-      "learning_rate": 1.7933723196881092e-06,
-      "loss": 0.1455,
+      "epoch": 6.77,
+      "learning_rate": 1.0768634743530143e-05,
+      "loss": 0.0797,
       "step": 14570
     },
     {
-      "epoch": 9.47,
-      "learning_rate": 1.7717132337015378e-06,
-      "loss": 0.0725,
+      "epoch": 6.78,
+      "learning_rate": 1.0753138075313807e-05,
+      "loss": 0.1026,
       "step": 14580
     },
     {
-      "epoch": 9.48,
-      "learning_rate": 1.7500541477149665e-06,
-      "loss": 0.0148,
+      "epoch": 6.78,
+      "learning_rate": 1.0737641407097474e-05,
+      "loss": 0.208,
       "step": 14590
     },
     {
-      "epoch": 9.48,
-      "learning_rate": 1.7283950617283952e-06,
-      "loss": 0.0011,
+      "epoch": 6.79,
+      "learning_rate": 1.072214473888114e-05,
+      "loss": 0.4293,
       "step": 14600
     },
     {
-      "epoch": 9.49,
-      "learning_rate": 1.7067359757418239e-06,
-      "loss": 0.002,
+      "epoch": 6.79,
+      "learning_rate": 1.0706648070664808e-05,
+      "loss": 0.2139,
       "step": 14610
     },
     {
-      "epoch": 9.5,
-      "learning_rate": 1.6850768897552524e-06,
-      "loss": 0.2557,
+      "epoch": 6.8,
+      "learning_rate": 1.0691151402448473e-05,
+      "loss": 0.3856,
       "step": 14620
     },
     {
-      "epoch": 9.5,
-      "learning_rate": 1.663417803768681e-06,
-      "loss": 0.2756,
+      "epoch": 6.8,
+      "learning_rate": 1.0675654734232141e-05,
+      "loss": 0.0709,
       "step": 14630
     },
     {
-      "epoch": 9.51,
-      "learning_rate": 1.6417587177821097e-06,
-      "loss": 0.0519,
+      "epoch": 6.81,
+      "learning_rate": 1.0660158066015807e-05,
+      "loss": 0.1399,
       "step": 14640
     },
     {
-      "epoch": 9.52,
-      "learning_rate": 1.6200996317955384e-06,
-      "loss": 0.0122,
+      "epoch": 6.81,
+      "learning_rate": 1.0644661397799474e-05,
+      "loss": 0.1463,
       "step": 14650
     },
     {
-      "epoch": 9.52,
-      "learning_rate": 1.5984405458089669e-06,
-      "loss": 0.0782,
+      "epoch": 6.82,
+      "learning_rate": 1.062916472958314e-05,
+      "loss": 0.2953,
       "step": 14660
     },
     {
-      "epoch": 9.53,
-      "learning_rate": 1.5767814598223956e-06,
-      "loss": 0.181,
+      "epoch": 6.82,
+      "learning_rate": 1.0613668061366807e-05,
+      "loss": 0.2367,
       "step": 14670
     },
     {
-      "epoch": 9.54,
-      "learning_rate": 1.5551223738358243e-06,
-      "loss": 0.0997,
+      "epoch": 6.82,
+      "learning_rate": 1.0598171393150473e-05,
+      "loss": 0.0831,
       "step": 14680
     },
     {
-      "epoch": 9.54,
-      "learning_rate": 1.533463287849253e-06,
-      "loss": 0.1639,
+      "epoch": 6.83,
+      "learning_rate": 1.058267472493414e-05,
+      "loss": 0.0993,
       "step": 14690
     },
     {
-      "epoch": 9.55,
-      "learning_rate": 1.5118042018626816e-06,
-      "loss": 0.0852,
+      "epoch": 6.83,
+      "learning_rate": 1.0567178056717805e-05,
+      "loss": 0.1645,
       "step": 14700
     },
     {
-      "epoch": 9.56,
-      "learning_rate": 1.4901451158761101e-06,
-      "loss": 0.001,
+      "epoch": 6.84,
+      "learning_rate": 1.0551681388501472e-05,
+      "loss": 0.0895,
       "step": 14710
     },
     {
-      "epoch": 9.56,
-      "learning_rate": 1.4684860298895388e-06,
-      "loss": 0.2217,
+      "epoch": 6.84,
+      "learning_rate": 1.0536184720285138e-05,
+      "loss": 0.2147,
       "step": 14720
     },
     {
-      "epoch": 9.57,
-      "learning_rate": 1.4468269439029673e-06,
-      "loss": 0.1359,
+      "epoch": 6.85,
+      "learning_rate": 1.0520688052068806e-05,
+      "loss": 0.3754,
       "step": 14730
     },
     {
-      "epoch": 9.57,
-      "learning_rate": 1.425167857916396e-06,
-      "loss": 0.1462,
+      "epoch": 6.85,
+      "learning_rate": 1.0505191383852471e-05,
+      "loss": 0.2089,
       "step": 14740
     },
     {
-      "epoch": 9.58,
-      "learning_rate": 1.4035087719298244e-06,
-      "loss": 0.002,
+      "epoch": 6.86,
+      "learning_rate": 1.0489694715636139e-05,
+      "loss": 0.1646,
       "step": 14750
     },
     {
-      "epoch": 9.59,
-      "learning_rate": 1.3818496859432531e-06,
-      "loss": 0.0147,
+      "epoch": 6.86,
+      "learning_rate": 1.0474198047419805e-05,
+      "loss": 0.1797,
       "step": 14760
     },
     {
-      "epoch": 9.59,
-      "learning_rate": 1.3601905999566818e-06,
-      "loss": 0.0607,
+      "epoch": 6.87,
+      "learning_rate": 1.0458701379203472e-05,
+      "loss": 0.2464,
       "step": 14770
     },
     {
-      "epoch": 9.6,
-      "learning_rate": 1.3385315139701105e-06,
-      "loss": 0.0903,
+      "epoch": 6.87,
+      "learning_rate": 1.0443204710987138e-05,
+      "loss": 0.3507,
       "step": 14780
     },
     {
-      "epoch": 9.61,
-      "learning_rate": 1.316872427983539e-06,
-      "loss": 0.1384,
+      "epoch": 6.88,
+      "learning_rate": 1.0427708042770805e-05,
+      "loss": 0.091,
       "step": 14790
     },
     {
-      "epoch": 9.61,
-      "learning_rate": 1.2952133419969676e-06,
-      "loss": 0.1349,
+      "epoch": 6.88,
+      "learning_rate": 1.0412211374554471e-05,
+      "loss": 0.415,
       "step": 14800
     },
     {
-      "epoch": 9.62,
-      "learning_rate": 1.2735542560103963e-06,
-      "loss": 0.145,
+      "epoch": 6.89,
+      "learning_rate": 1.0396714706338139e-05,
+      "loss": 0.2619,
       "step": 14810
     },
     {
-      "epoch": 9.63,
-      "learning_rate": 1.251895170023825e-06,
-      "loss": 0.0957,
+      "epoch": 6.89,
+      "learning_rate": 1.0381218038121804e-05,
+      "loss": 0.0688,
       "step": 14820
     },
     {
-      "epoch": 9.63,
-      "learning_rate": 1.2302360840372537e-06,
-      "loss": 0.0971,
+      "epoch": 6.89,
+      "learning_rate": 1.036572136990547e-05,
+      "loss": 0.1622,
       "step": 14830
     },
     {
-      "epoch": 9.64,
-      "learning_rate": 1.2085769980506822e-06,
-      "loss": 0.0578,
+      "epoch": 6.9,
+      "learning_rate": 1.0350224701689136e-05,
+      "loss": 0.0825,
       "step": 14840
     },
     {
-      "epoch": 9.65,
-      "learning_rate": 1.1869179120641109e-06,
-      "loss": 0.0793,
+      "epoch": 6.9,
+      "learning_rate": 1.0334728033472803e-05,
+      "loss": 0.2034,
       "step": 14850
     },
     {
-      "epoch": 9.65,
-      "learning_rate": 1.1652588260775395e-06,
-      "loss": 0.0914,
+      "epoch": 6.91,
+      "learning_rate": 1.031923136525647e-05,
+      "loss": 0.0951,
       "step": 14860
     },
     {
-      "epoch": 9.66,
-      "learning_rate": 1.1435997400909682e-06,
-      "loss": 0.0207,
+      "epoch": 6.91,
+      "learning_rate": 1.0303734697040137e-05,
+      "loss": 0.2086,
       "step": 14870
     },
     {
-      "epoch": 9.67,
-      "learning_rate": 1.1219406541043967e-06,
-      "loss": 0.0121,
+      "epoch": 6.92,
+      "learning_rate": 1.0288238028823803e-05,
+      "loss": 0.0577,
       "step": 14880
     },
     {
-      "epoch": 9.67,
-      "learning_rate": 1.1002815681178254e-06,
-      "loss": 0.0013,
+      "epoch": 6.92,
+      "learning_rate": 1.027274136060747e-05,
+      "loss": 0.2179,
       "step": 14890
     },
     {
-      "epoch": 9.68,
-      "learning_rate": 1.078622482131254e-06,
-      "loss": 0.1251,
+      "epoch": 6.93,
+      "learning_rate": 1.0257244692391136e-05,
+      "loss": 0.1198,
       "step": 14900
     },
     {
-      "epoch": 9.68,
-      "learning_rate": 1.0569633961446828e-06,
-      "loss": 0.0711,
+      "epoch": 6.93,
+      "learning_rate": 1.0241748024174803e-05,
+      "loss": 0.1049,
       "step": 14910
     },
     {
-      "epoch": 9.69,
-      "learning_rate": 1.0353043101581112e-06,
-      "loss": 0.032,
+      "epoch": 6.94,
+      "learning_rate": 1.0226251355958469e-05,
+      "loss": 0.3032,
       "step": 14920
     },
     {
-      "epoch": 9.7,
-      "learning_rate": 1.01364522417154e-06,
-      "loss": 0.1033,
+      "epoch": 6.94,
+      "learning_rate": 1.0210754687742137e-05,
+      "loss": 0.3674,
       "step": 14930
     },
     {
-      "epoch": 9.7,
-      "learning_rate": 9.919861381849686e-07,
-      "loss": 0.0706,
+      "epoch": 6.95,
+      "learning_rate": 1.0195258019525802e-05,
+      "loss": 0.1696,
       "step": 14940
     },
     {
-      "epoch": 9.71,
-      "learning_rate": 9.703270521983973e-07,
-      "loss": 0.0011,
+      "epoch": 6.95,
+      "learning_rate": 1.0179761351309468e-05,
+      "loss": 0.2785,
       "step": 14950
     },
     {
-      "epoch": 9.72,
-      "learning_rate": 9.48667966211826e-07,
-      "loss": 0.1376,
+      "epoch": 6.95,
+      "learning_rate": 1.0164264683093134e-05,
+      "loss": 0.1236,
       "step": 14960
     },
     {
-      "epoch": 9.72,
-      "learning_rate": 9.270088802252545e-07,
-      "loss": 0.0821,
+      "epoch": 6.96,
+      "learning_rate": 1.0148768014876801e-05,
+      "loss": 0.231,
       "step": 14970
     },
     {
-      "epoch": 9.73,
-      "learning_rate": 9.053497942386831e-07,
-      "loss": 0.0325,
+      "epoch": 6.96,
+      "learning_rate": 1.0133271346660467e-05,
+      "loss": 0.1885,
       "step": 14980
     },
     {
-      "epoch": 9.74,
-      "learning_rate": 8.836907082521118e-07,
-      "loss": 0.1046,
+      "epoch": 6.97,
+      "learning_rate": 1.0117774678444135e-05,
+      "loss": 0.2478,
       "step": 14990
     },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.01022780102278e-05,
+      "loss": 0.3233,
+      "step": 15000
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.0086781342011468e-05,
+      "loss": 0.1368,
+      "step": 15010
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.0071284673795134e-05,
+      "loss": 0.1488,
+      "step": 15020
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.0055788005578801e-05,
+      "loss": 0.1132,
+      "step": 15030
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.0040291337362467e-05,
+      "loss": 0.3882,
+      "step": 15040
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.0024794669146135e-05,
+      "loss": 0.1319,
+      "step": 15050
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9789265722752717,
+      "eval_loss": 0.10872756689786911,
+      "eval_runtime": 17.2429,
+      "eval_samples_per_second": 176.131,
+      "eval_steps_per_second": 44.076,
+      "step": 15057
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.00092980009298e-05,
+      "loss": 0.1708,
+      "step": 15060
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.993801332713468e-06,
+      "loss": 0.2515,
+      "step": 15070
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.978304664497132e-06,
+      "loss": 0.1733,
+      "step": 15080
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.9628079962808e-06,
+      "loss": 0.1955,
+      "step": 15090
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.947311328064465e-06,
+      "loss": 0.3123,
+      "step": 15100
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.931814659848133e-06,
+      "loss": 0.0549,
+      "step": 15110
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.916317991631798e-06,
+      "loss": 0.0458,
+      "step": 15120
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.900821323415466e-06,
+      "loss": 0.0786,
+      "step": 15130
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.885324655199132e-06,
+      "loss": 0.1529,
+      "step": 15140
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.8698279869828e-06,
+      "loss": 0.0907,
+      "step": 15150
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.854331318766465e-06,
+      "loss": 0.092,
+      "step": 15160
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.838834650550132e-06,
+      "loss": 0.3792,
+      "step": 15170
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.823337982333798e-06,
+      "loss": 0.1769,
+      "step": 15180
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.807841314117466e-06,
+      "loss": 0.1373,
+      "step": 15190
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.792344645901132e-06,
+      "loss": 0.1044,
+      "step": 15200
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.776847977684797e-06,
+      "loss": 0.0664,
+      "step": 15210
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.761351309468465e-06,
+      "loss": 0.2296,
+      "step": 15220
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.74585464125213e-06,
+      "loss": 0.3984,
+      "step": 15230
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.730357973035798e-06,
+      "loss": 0.1523,
+      "step": 15240
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.714861304819464e-06,
+      "loss": 0.0526,
+      "step": 15250
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.699364636603131e-06,
+      "loss": 0.1269,
+      "step": 15260
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.683867968386797e-06,
+      "loss": 0.289,
+      "step": 15270
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.668371300170465e-06,
+      "loss": 0.0285,
+      "step": 15280
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.65287463195413e-06,
+      "loss": 0.0599,
+      "step": 15290
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.637377963737798e-06,
+      "loss": 0.121,
+      "step": 15300
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.621881295521464e-06,
+      "loss": 0.0041,
+      "step": 15310
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.60638462730513e-06,
+      "loss": 0.2209,
+      "step": 15320
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.590887959088795e-06,
+      "loss": 0.1529,
+      "step": 15330
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.575391290872463e-06,
+      "loss": 0.1067,
+      "step": 15340
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.559894622656129e-06,
+      "loss": 0.0675,
+      "step": 15350
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.544397954439796e-06,
+      "loss": 0.0939,
+      "step": 15360
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.528901286223462e-06,
+      "loss": 0.297,
+      "step": 15370
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.51340461800713e-06,
+      "loss": 0.0538,
+      "step": 15380
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.497907949790795e-06,
+      "loss": 0.1045,
+      "step": 15390
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.482411281574463e-06,
+      "loss": 0.1477,
+      "step": 15400
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.466914613358128e-06,
+      "loss": 0.061,
+      "step": 15410
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.451417945141796e-06,
+      "loss": 0.0951,
+      "step": 15420
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.435921276925462e-06,
+      "loss": 0.1163,
+      "step": 15430
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.42042460870913e-06,
+      "loss": 0.2016,
+      "step": 15440
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.404927940492793e-06,
+      "loss": 0.0333,
+      "step": 15450
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.38943127227646e-06,
+      "loss": 0.0091,
+      "step": 15460
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.373934604060127e-06,
+      "loss": 0.3396,
+      "step": 15470
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.358437935843794e-06,
+      "loss": 0.0027,
+      "step": 15480
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.34294126762746e-06,
+      "loss": 0.1906,
+      "step": 15490
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.327444599411127e-06,
+      "loss": 0.2862,
+      "step": 15500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.311947931194793e-06,
+      "loss": 0.0614,
+      "step": 15510
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.29645126297846e-06,
+      "loss": 0.133,
+      "step": 15520
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.280954594762126e-06,
+      "loss": 0.1031,
+      "step": 15530
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.265457926545794e-06,
+      "loss": 0.1839,
+      "step": 15540
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.24996125832946e-06,
+      "loss": 0.0742,
+      "step": 15550
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.234464590113127e-06,
+      "loss": 0.2748,
+      "step": 15560
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.218967921896791e-06,
+      "loss": 0.3583,
+      "step": 15570
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.203471253680459e-06,
+      "loss": 0.0343,
+      "step": 15580
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.187974585464124e-06,
+      "loss": 0.3678,
+      "step": 15590
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.172477917247792e-06,
+      "loss": 0.0024,
+      "step": 15600
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.156981249031458e-06,
+      "loss": 0.111,
+      "step": 15610
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.141484580815125e-06,
+      "loss": 0.1994,
+      "step": 15620
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.125987912598791e-06,
+      "loss": 0.2164,
+      "step": 15630
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.110491244382459e-06,
+      "loss": 0.2001,
+      "step": 15640
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.094994576166124e-06,
+      "loss": 0.0191,
+      "step": 15650
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.079497907949792e-06,
+      "loss": 0.1286,
+      "step": 15660
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.064001239733458e-06,
+      "loss": 0.1652,
+      "step": 15670
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.048504571517125e-06,
+      "loss": 0.1685,
+      "step": 15680
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.033007903300791e-06,
+      "loss": 0.1938,
+      "step": 15690
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 9.017511235084457e-06,
+      "loss": 0.1407,
+      "step": 15700
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 9.002014566868122e-06,
+      "loss": 0.1829,
+      "step": 15710
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.98651789865179e-06,
+      "loss": 0.0607,
+      "step": 15720
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.971021230435456e-06,
+      "loss": 0.3167,
+      "step": 15730
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.955524562219123e-06,
+      "loss": 0.1105,
+      "step": 15740
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.940027894002789e-06,
+      "loss": 0.1907,
+      "step": 15750
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.924531225786456e-06,
+      "loss": 0.0641,
+      "step": 15760
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.909034557570122e-06,
+      "loss": 0.2064,
+      "step": 15770
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.89353788935379e-06,
+      "loss": 0.0045,
+      "step": 15780
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.878041221137456e-06,
+      "loss": 0.1104,
+      "step": 15790
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.862544552921123e-06,
+      "loss": 0.0997,
+      "step": 15800
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.847047884704789e-06,
+      "loss": 0.1006,
+      "step": 15810
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.831551216488455e-06,
+      "loss": 0.0891,
+      "step": 15820
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.81605454827212e-06,
+      "loss": 0.0535,
+      "step": 15830
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.800557880055788e-06,
+      "loss": 0.2747,
+      "step": 15840
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.785061211839454e-06,
+      "loss": 0.1308,
+      "step": 15850
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.769564543623121e-06,
+      "loss": 0.079,
+      "step": 15860
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.754067875406787e-06,
+      "loss": 0.0534,
+      "step": 15870
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.738571207190454e-06,
+      "loss": 0.1435,
+      "step": 15880
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.72307453897412e-06,
+      "loss": 0.0688,
+      "step": 15890
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.707577870757788e-06,
+      "loss": 0.1386,
+      "step": 15900
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.692081202541453e-06,
+      "loss": 0.2127,
+      "step": 15910
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.676584534325121e-06,
+      "loss": 0.1386,
+      "step": 15920
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.661087866108787e-06,
+      "loss": 0.0835,
+      "step": 15930
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.645591197892454e-06,
+      "loss": 0.2212,
+      "step": 15940
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.630094529676118e-06,
+      "loss": 0.2225,
+      "step": 15950
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.614597861459786e-06,
+      "loss": 0.0939,
+      "step": 15960
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.599101193243452e-06,
+      "loss": 0.0194,
+      "step": 15970
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.583604525027119e-06,
+      "loss": 0.2879,
+      "step": 15980
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.568107856810785e-06,
+      "loss": 0.0863,
+      "step": 15990
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.552611188594452e-06,
+      "loss": 0.145,
+      "step": 16000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.537114520378118e-06,
+      "loss": 0.1236,
+      "step": 16010
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.521617852161786e-06,
+      "loss": 0.2345,
+      "step": 16020
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.506121183945451e-06,
+      "loss": 0.2312,
+      "step": 16030
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.490624515729119e-06,
+      "loss": 0.0682,
+      "step": 16040
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.475127847512786e-06,
+      "loss": 0.1497,
+      "step": 16050
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.459631179296452e-06,
+      "loss": 0.0016,
+      "step": 16060
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.444134511080118e-06,
+      "loss": 0.3009,
+      "step": 16070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.428637842863784e-06,
+      "loss": 0.2538,
+      "step": 16080
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.413141174647451e-06,
+      "loss": 0.0944,
+      "step": 16090
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.397644506431117e-06,
+      "loss": 0.1941,
+      "step": 16100
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.382147838214785e-06,
+      "loss": 0.007,
+      "step": 16110
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.36665116999845e-06,
+      "loss": 0.3934,
+      "step": 16120
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.351154501782118e-06,
+      "loss": 0.0915,
+      "step": 16130
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.335657833565784e-06,
+      "loss": 0.1373,
+      "step": 16140
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.320161165349451e-06,
+      "loss": 0.0626,
+      "step": 16150
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.304664497133117e-06,
+      "loss": 0.0624,
+      "step": 16160
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.289167828916784e-06,
+      "loss": 0.0458,
+      "step": 16170
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.27367116070045e-06,
+      "loss": 0.2246,
+      "step": 16180
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.258174492484118e-06,
+      "loss": 0.1318,
+      "step": 16190
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.242677824267782e-06,
+      "loss": 0.0221,
+      "step": 16200
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.22718115605145e-06,
+      "loss": 0.2664,
+      "step": 16210
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.211684487835115e-06,
+      "loss": 0.0027,
+      "step": 16220
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 8.196187819618782e-06,
+      "loss": 0.1597,
+      "step": 16230
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 8.180691151402448e-06,
+      "loss": 0.2088,
+      "step": 16240
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 8.165194483186116e-06,
+      "loss": 0.2231,
+      "step": 16250
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 8.149697814969782e-06,
+      "loss": 0.2694,
+      "step": 16260
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 8.134201146753449e-06,
+      "loss": 0.0034,
+      "step": 16270
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 8.118704478537115e-06,
+      "loss": 0.0549,
+      "step": 16280
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 8.103207810320782e-06,
+      "loss": 0.1103,
+      "step": 16290
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 8.087711142104448e-06,
+      "loss": 0.0032,
+      "step": 16300
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 8.072214473888116e-06,
+      "loss": 0.0827,
+      "step": 16310
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 8.05671780567178e-06,
+      "loss": 0.1303,
+      "step": 16320
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 8.041221137455447e-06,
+      "loss": 0.129,
+      "step": 16330
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 8.025724469239113e-06,
+      "loss": 0.1739,
+      "step": 16340
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 8.01022780102278e-06,
+      "loss": 0.0765,
+      "step": 16350
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.994731132806446e-06,
+      "loss": 0.1083,
+      "step": 16360
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.979234464590114e-06,
+      "loss": 0.0901,
+      "step": 16370
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.96373779637378e-06,
+      "loss": 0.0506,
+      "step": 16380
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.948241128157447e-06,
+      "loss": 0.0382,
+      "step": 16390
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.932744459941113e-06,
+      "loss": 0.0841,
+      "step": 16400
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.91724779172478e-06,
+      "loss": 0.2492,
+      "step": 16410
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.901751123508446e-06,
+      "loss": 0.0846,
+      "step": 16420
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.886254455292114e-06,
+      "loss": 0.0028,
+      "step": 16430
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.87075778707578e-06,
+      "loss": 0.1131,
+      "step": 16440
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.855261118859445e-06,
+      "loss": 0.1876,
+      "step": 16450
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.839764450643111e-06,
+      "loss": 0.088,
+      "step": 16460
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.824267782426778e-06,
+      "loss": 0.2712,
+      "step": 16470
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.808771114210444e-06,
+      "loss": 0.1309,
+      "step": 16480
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.793274445994112e-06,
+      "loss": 0.1477,
+      "step": 16490
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.777777777777777e-06,
+      "loss": 0.1527,
+      "step": 16500
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.762281109561445e-06,
+      "loss": 0.0907,
+      "step": 16510
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.74678444134511e-06,
+      "loss": 0.0049,
+      "step": 16520
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.731287773128778e-06,
+      "loss": 0.2083,
+      "step": 16530
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.715791104912444e-06,
+      "loss": 0.1156,
+      "step": 16540
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.700294436696112e-06,
+      "loss": 0.2825,
+      "step": 16550
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.684797768479777e-06,
+      "loss": 0.1159,
+      "step": 16560
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.669301100263443e-06,
+      "loss": 0.0711,
+      "step": 16570
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.653804432047109e-06,
+      "loss": 0.2192,
+      "step": 16580
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.638307763830776e-06,
+      "loss": 0.1037,
+      "step": 16590
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.622811095614442e-06,
+      "loss": 0.1424,
+      "step": 16600
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.60731442739811e-06,
+      "loss": 0.216,
+      "step": 16610
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.5918177591817754e-06,
+      "loss": 0.0876,
+      "step": 16620
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.576321090965443e-06,
+      "loss": 0.1751,
+      "step": 16630
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.560824422749109e-06,
+      "loss": 0.1445,
+      "step": 16640
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.545327754532776e-06,
+      "loss": 0.1459,
+      "step": 16650
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.529831086316442e-06,
+      "loss": 0.222,
+      "step": 16660
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.514334418100109e-06,
+      "loss": 0.0253,
+      "step": 16670
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.498837749883775e-06,
+      "loss": 0.2423,
+      "step": 16680
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.483341081667442e-06,
+      "loss": 0.2054,
+      "step": 16690
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.4678444134511085e-06,
+      "loss": 0.1258,
+      "step": 16700
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.452347745234775e-06,
+      "loss": 0.0495,
+      "step": 16710
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.436851077018442e-06,
+      "loss": 0.0716,
+      "step": 16720
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.421354408802108e-06,
+      "loss": 0.0371,
+      "step": 16730
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.405857740585774e-06,
+      "loss": 0.1346,
+      "step": 16740
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.390361072369441e-06,
+      "loss": 0.1465,
+      "step": 16750
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.3748644041531075e-06,
+      "loss": 0.025,
+      "step": 16760
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.359367735936774e-06,
+      "loss": 0.2293,
+      "step": 16770
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.343871067720441e-06,
+      "loss": 0.1768,
+      "step": 16780
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.3283743995041066e-06,
+      "loss": 0.091,
+      "step": 16790
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.312877731287773e-06,
+      "loss": 0.1603,
+      "step": 16800
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.29738106307144e-06,
+      "loss": 0.0076,
+      "step": 16810
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.2818843948551065e-06,
+      "loss": 0.2233,
+      "step": 16820
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.266387726638773e-06,
+      "loss": 0.0052,
+      "step": 16830
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.25089105842244e-06,
+      "loss": 0.0976,
+      "step": 16840
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.2353943902061055e-06,
+      "loss": 0.1401,
+      "step": 16850
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.219897721989772e-06,
+      "loss": 0.1559,
+      "step": 16860
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.204401053773439e-06,
+      "loss": 0.2169,
+      "step": 16870
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.1889043855571055e-06,
+      "loss": 0.0884,
+      "step": 16880
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.173407717340772e-06,
+      "loss": 0.0743,
+      "step": 16890
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 7.157911049124439e-06,
+      "loss": 0.3376,
+      "step": 16900
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 7.142414380908105e-06,
+      "loss": 0.0255,
+      "step": 16910
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 7.126917712691771e-06,
+      "loss": 0.0286,
+      "step": 16920
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 7.111421044475438e-06,
+      "loss": 0.0028,
+      "step": 16930
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 7.0959243762591044e-06,
+      "loss": 0.2599,
+      "step": 16940
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 7.080427708042771e-06,
+      "loss": 0.0389,
+      "step": 16950
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 7.064931039826438e-06,
+      "loss": 0.1285,
+      "step": 16960
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 7.049434371610104e-06,
+      "loss": 0.1612,
+      "step": 16970
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 7.03393770339377e-06,
+      "loss": 0.0285,
+      "step": 16980
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 7.018441035177437e-06,
+      "loss": 0.0944,
+      "step": 16990
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 7.002944366961103e-06,
+      "loss": 0.1581,
+      "step": 17000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.98744769874477e-06,
+      "loss": 0.202,
+      "step": 17010
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.971951030528437e-06,
+      "loss": 0.1744,
+      "step": 17020
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.9580040291337366e-06,
+      "loss": 0.3009,
+      "step": 17030
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.942507360917403e-06,
+      "loss": 0.0864,
+      "step": 17040
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.92701069270107e-06,
+      "loss": 0.212,
+      "step": 17050
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.9115140244847365e-06,
+      "loss": 0.2234,
+      "step": 17060
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.896017356268402e-06,
+      "loss": 0.0449,
+      "step": 17070
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.880520688052069e-06,
+      "loss": 0.0892,
+      "step": 17080
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.8650240198357356e-06,
+      "loss": 0.2014,
+      "step": 17090
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.849527351619402e-06,
+      "loss": 0.174,
+      "step": 17100
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.834030683403069e-06,
+      "loss": 0.0537,
+      "step": 17110
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.8185340151867355e-06,
+      "loss": 0.2684,
+      "step": 17120
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.803037346970401e-06,
+      "loss": 0.0025,
+      "step": 17130
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.787540678754068e-06,
+      "loss": 0.007,
+      "step": 17140
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.7720440105377345e-06,
+      "loss": 0.1587,
+      "step": 17150
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.756547342321401e-06,
+      "loss": 0.2571,
+      "step": 17160
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.741050674105068e-06,
+      "loss": 0.0534,
+      "step": 17170
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.7255540058887344e-06,
+      "loss": 0.1896,
+      "step": 17180
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.7100573376724e-06,
+      "loss": 0.1856,
+      "step": 17190
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.694560669456067e-06,
+      "loss": 0.0646,
+      "step": 17200
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.9739874876522885,
+      "eval_loss": 0.13712750375270844,
+      "eval_runtime": 17.5393,
+      "eval_samples_per_second": 173.154,
+      "eval_steps_per_second": 43.331,
+      "step": 17208
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.6790640012397335e-06,
+      "loss": 0.2262,
+      "step": 17210
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.6635673330234e-06,
+      "loss": 0.1837,
+      "step": 17220
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.648070664807067e-06,
+      "loss": 0.1265,
+      "step": 17230
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.632573996590733e-06,
+      "loss": 0.1258,
+      "step": 17240
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.617077328374399e-06,
+      "loss": 0.0651,
+      "step": 17250
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.601580660158066e-06,
+      "loss": 0.3102,
+      "step": 17260
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.5860839919417325e-06,
+      "loss": 0.1612,
+      "step": 17270
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.570587323725399e-06,
+      "loss": 0.1483,
+      "step": 17280
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.555090655509066e-06,
+      "loss": 0.0733,
+      "step": 17290
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.539593987292732e-06,
+      "loss": 0.0171,
+      "step": 17300
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.524097319076399e-06,
+      "loss": 0.2636,
+      "step": 17310
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.508600650860065e-06,
+      "loss": 0.2274,
+      "step": 17320
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.4931039826437315e-06,
+      "loss": 0.0884,
+      "step": 17330
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.477607314427398e-06,
+      "loss": 0.2319,
+      "step": 17340
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.462110646211065e-06,
+      "loss": 0.0403,
+      "step": 17350
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.446613977994731e-06,
+      "loss": 0.0503,
+      "step": 17360
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.431117309778398e-06,
+      "loss": 0.0743,
+      "step": 17370
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.415620641562064e-06,
+      "loss": 0.0994,
+      "step": 17380
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.4001239733457304e-06,
+      "loss": 0.0959,
+      "step": 17390
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.384627305129397e-06,
+      "loss": 0.1529,
+      "step": 17400
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.369130636913064e-06,
+      "loss": 0.0031,
+      "step": 17410
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.35363396869673e-06,
+      "loss": 0.1075,
+      "step": 17420
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.338137300480397e-06,
+      "loss": 0.0573,
+      "step": 17430
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.322640632264063e-06,
+      "loss": 0.2425,
+      "step": 17440
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.307143964047729e-06,
+      "loss": 0.1474,
+      "step": 17450
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.291647295831396e-06,
+      "loss": 0.2,
+      "step": 17460
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.276150627615063e-06,
+      "loss": 0.1569,
+      "step": 17470
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.260653959398729e-06,
+      "loss": 0.1137,
+      "step": 17480
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.245157291182396e-06,
+      "loss": 0.0912,
+      "step": 17490
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.229660622966062e-06,
+      "loss": 0.1301,
+      "step": 17500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.214163954749728e-06,
+      "loss": 0.0062,
+      "step": 17510
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.198667286533395e-06,
+      "loss": 0.1619,
+      "step": 17520
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.183170618317062e-06,
+      "loss": 0.1377,
+      "step": 17530
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.167673950100728e-06,
+      "loss": 0.0386,
+      "step": 17540
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.152177281884395e-06,
+      "loss": 0.0695,
+      "step": 17550
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.136680613668062e-06,
+      "loss": 0.0603,
+      "step": 17560
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.121183945451727e-06,
+      "loss": 0.2312,
+      "step": 17570
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 6.105687277235394e-06,
+      "loss": 0.018,
+      "step": 17580
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.090190609019061e-06,
+      "loss": 0.1271,
+      "step": 17590
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 6.074693940802727e-06,
+      "loss": 0.1665,
+      "step": 17600
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.059197272586394e-06,
+      "loss": 0.1065,
+      "step": 17610
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 6.043700604370061e-06,
+      "loss": 0.5536,
+      "step": 17620
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.028203936153727e-06,
+      "loss": 0.1297,
+      "step": 17630
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 6.012707267937394e-06,
+      "loss": 0.097,
+      "step": 17640
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.9972105997210605e-06,
+      "loss": 0.0464,
+      "step": 17650
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.981713931504727e-06,
+      "loss": 0.0168,
+      "step": 17660
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.966217263288394e-06,
+      "loss": 0.0492,
+      "step": 17670
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.95072059507206e-06,
+      "loss": 0.1815,
+      "step": 17680
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.935223926855726e-06,
+      "loss": 0.2915,
+      "step": 17690
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.919727258639393e-06,
+      "loss": 0.0628,
+      "step": 17700
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.9042305904230594e-06,
+      "loss": 0.0442,
+      "step": 17710
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.888733922206726e-06,
+      "loss": 0.0826,
+      "step": 17720
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.873237253990393e-06,
+      "loss": 0.1191,
+      "step": 17730
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.857740585774059e-06,
+      "loss": 0.1648,
+      "step": 17740
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.842243917557725e-06,
+      "loss": 0.0023,
+      "step": 17750
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.826747249341392e-06,
+      "loss": 0.1562,
+      "step": 17760
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.811250581125058e-06,
+      "loss": 0.0411,
+      "step": 17770
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.795753912908725e-06,
+      "loss": 0.0362,
+      "step": 17780
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.780257244692392e-06,
+      "loss": 0.1763,
+      "step": 17790
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.764760576476058e-06,
+      "loss": 0.0708,
+      "step": 17800
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.749263908259724e-06,
+      "loss": 0.3768,
+      "step": 17810
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.733767240043391e-06,
+      "loss": 0.277,
+      "step": 17820
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.718270571827057e-06,
+      "loss": 0.1965,
+      "step": 17830
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.702773903610724e-06,
+      "loss": 0.1794,
+      "step": 17840
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.687277235394391e-06,
+      "loss": 0.0979,
+      "step": 17850
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.671780567178057e-06,
+      "loss": 0.072,
+      "step": 17860
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.656283898961724e-06,
+      "loss": 0.1873,
+      "step": 17870
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.64078723074539e-06,
+      "loss": 0.0778,
+      "step": 17880
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.625290562529056e-06,
+      "loss": 0.173,
+      "step": 17890
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.609793894312723e-06,
+      "loss": 0.1449,
+      "step": 17900
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.59429722609639e-06,
+      "loss": 0.0297,
+      "step": 17910
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.578800557880056e-06,
+      "loss": 0.0418,
+      "step": 17920
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.563303889663723e-06,
+      "loss": 0.0307,
+      "step": 17930
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.547807221447389e-06,
+      "loss": 0.1582,
+      "step": 17940
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.532310553231055e-06,
+      "loss": 0.1351,
+      "step": 17950
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.516813885014722e-06,
+      "loss": 0.0991,
+      "step": 17960
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.501317216798389e-06,
+      "loss": 0.0395,
+      "step": 17970
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.485820548582055e-06,
+      "loss": 0.3475,
+      "step": 17980
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.470323880365722e-06,
+      "loss": 0.1425,
+      "step": 17990
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.454827212149388e-06,
+      "loss": 0.1808,
+      "step": 18000
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.439330543933054e-06,
+      "loss": 0.034,
+      "step": 18010
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.423833875716721e-06,
+      "loss": 0.1168,
+      "step": 18020
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.408337207500388e-06,
+      "loss": 0.1169,
+      "step": 18030
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.392840539284054e-06,
+      "loss": 0.05,
+      "step": 18040
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.377343871067721e-06,
+      "loss": 0.1126,
+      "step": 18050
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.361847202851387e-06,
+      "loss": 0.0158,
+      "step": 18060
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.346350534635053e-06,
+      "loss": 0.015,
+      "step": 18070
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.33085386641872e-06,
+      "loss": 0.0189,
+      "step": 18080
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.315357198202387e-06,
+      "loss": 0.1123,
+      "step": 18090
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.299860529986053e-06,
+      "loss": 0.1492,
+      "step": 18100
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.28436386176972e-06,
+      "loss": 0.0464,
+      "step": 18110
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.2688671935533865e-06,
+      "loss": 0.2331,
+      "step": 18120
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.253370525337052e-06,
+      "loss": 0.1383,
+      "step": 18130
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.237873857120719e-06,
+      "loss": 0.1527,
+      "step": 18140
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.2223771889043856e-06,
+      "loss": 0.1064,
+      "step": 18150
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.206880520688052e-06,
+      "loss": 0.1365,
+      "step": 18160
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.191383852471719e-06,
+      "loss": 0.1485,
+      "step": 18170
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.1758871842553855e-06,
+      "loss": 0.0667,
+      "step": 18180
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.160390516039051e-06,
+      "loss": 0.2232,
+      "step": 18190
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.144893847822718e-06,
+      "loss": 0.1358,
+      "step": 18200
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.1293971796063845e-06,
+      "loss": 0.2614,
+      "step": 18210
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.113900511390051e-06,
+      "loss": 0.1439,
+      "step": 18220
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.098403843173718e-06,
+      "loss": 0.1751,
+      "step": 18230
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.0829071749573844e-06,
+      "loss": 0.0027,
+      "step": 18240
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.06741050674105e-06,
+      "loss": 0.2695,
+      "step": 18250
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.051913838524717e-06,
+      "loss": 0.187,
+      "step": 18260
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 5.0364171703083835e-06,
+      "loss": 0.3292,
+      "step": 18270
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.02092050209205e-06,
+      "loss": 0.0624,
+      "step": 18280
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.005423833875717e-06,
+      "loss": 0.0631,
+      "step": 18290
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.9899271656593834e-06,
+      "loss": 0.0393,
+      "step": 18300
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.974430497443049e-06,
+      "loss": 0.1088,
+      "step": 18310
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.958933829226716e-06,
+      "loss": 0.0956,
+      "step": 18320
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.9434371610103825e-06,
+      "loss": 0.0216,
+      "step": 18330
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.927940492794049e-06,
+      "loss": 0.1236,
+      "step": 18340
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.912443824577716e-06,
+      "loss": 0.3147,
+      "step": 18350
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.896947156361382e-06,
+      "loss": 0.3294,
+      "step": 18360
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.881450488145049e-06,
+      "loss": 0.0474,
+      "step": 18370
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.865953819928715e-06,
+      "loss": 0.1255,
+      "step": 18380
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.8504571517123815e-06,
+      "loss": 0.0888,
+      "step": 18390
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.834960483496048e-06,
+      "loss": 0.1456,
+      "step": 18400
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.819463815279715e-06,
+      "loss": 0.1489,
+      "step": 18410
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.803967147063381e-06,
+      "loss": 0.0018,
+      "step": 18420
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.788470478847048e-06,
+      "loss": 0.1444,
+      "step": 18430
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.772973810630714e-06,
+      "loss": 0.1743,
+      "step": 18440
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.7574771424143804e-06,
+      "loss": 0.0658,
+      "step": 18450
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.741980474198047e-06,
+      "loss": 0.1066,
+      "step": 18460
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.726483805981714e-06,
+      "loss": 0.1067,
+      "step": 18470
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.71098713776538e-06,
+      "loss": 0.0035,
+      "step": 18480
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.695490469549048e-06,
+      "loss": 0.1949,
+      "step": 18490
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.679993801332714e-06,
+      "loss": 0.0048,
+      "step": 18500
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.66449713311638e-06,
+      "loss": 0.2269,
+      "step": 18510
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.649000464900047e-06,
+      "loss": 0.0265,
+      "step": 18520
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.6335037966837135e-06,
+      "loss": 0.0228,
+      "step": 18530
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.61800712846738e-06,
+      "loss": 0.243,
+      "step": 18540
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.602510460251047e-06,
+      "loss": 0.2173,
+      "step": 18550
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.587013792034713e-06,
+      "loss": 0.0026,
+      "step": 18560
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.571517123818379e-06,
+      "loss": 0.0094,
+      "step": 18570
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.556020455602046e-06,
+      "loss": 0.1997,
+      "step": 18580
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.5405237873857125e-06,
+      "loss": 0.2019,
+      "step": 18590
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.525027119169379e-06,
+      "loss": 0.2062,
+      "step": 18600
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.509530450953046e-06,
+      "loss": 0.1384,
+      "step": 18610
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.494033782736712e-06,
+      "loss": 0.0141,
+      "step": 18620
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.478537114520378e-06,
+      "loss": 0.0996,
+      "step": 18630
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.463040446304045e-06,
+      "loss": 0.4844,
+      "step": 18640
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.4475437780877115e-06,
+      "loss": 0.1961,
+      "step": 18650
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.432047109871378e-06,
+      "loss": 0.1568,
+      "step": 18660
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.416550441655045e-06,
+      "loss": 0.1181,
+      "step": 18670
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.401053773438711e-06,
+      "loss": 0.176,
+      "step": 18680
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.385557105222377e-06,
+      "loss": 0.2728,
+      "step": 18690
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.370060437006044e-06,
+      "loss": 0.2356,
+      "step": 18700
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.3545637687897105e-06,
+      "loss": 0.1333,
+      "step": 18710
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.339067100573377e-06,
+      "loss": 0.1476,
+      "step": 18720
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.323570432357044e-06,
+      "loss": 0.1592,
+      "step": 18730
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.30807376414071e-06,
+      "loss": 0.1635,
+      "step": 18740
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.292577095924376e-06,
+      "loss": 0.0086,
+      "step": 18750
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.277080427708043e-06,
+      "loss": 0.2478,
+      "step": 18760
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.2615837594917095e-06,
+      "loss": 0.1388,
+      "step": 18770
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.246087091275376e-06,
+      "loss": 0.0523,
+      "step": 18780
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.230590423059043e-06,
+      "loss": 0.1501,
+      "step": 18790
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.215093754842709e-06,
+      "loss": 0.0029,
+      "step": 18800
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.199597086626375e-06,
+      "loss": 0.063,
+      "step": 18810
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.184100418410042e-06,
+      "loss": 0.0687,
+      "step": 18820
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.1686037501937084e-06,
+      "loss": 0.1215,
+      "step": 18830
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.153107081977375e-06,
+      "loss": 0.1735,
+      "step": 18840
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.137610413761042e-06,
+      "loss": 0.0016,
+      "step": 18850
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.122113745544708e-06,
+      "loss": 0.0648,
+      "step": 18860
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.106617077328374e-06,
+      "loss": 0.2182,
+      "step": 18870
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.091120409112041e-06,
+      "loss": 0.1601,
+      "step": 18880
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 4.075623740895707e-06,
+      "loss": 0.11,
+      "step": 18890
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.060127072679374e-06,
+      "loss": 0.1182,
+      "step": 18900
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 4.044630404463041e-06,
+      "loss": 0.2342,
+      "step": 18910
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.029133736246707e-06,
+      "loss": 0.0474,
+      "step": 18920
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 4.013637068030374e-06,
+      "loss": 0.1305,
+      "step": 18930
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.99814039981404e-06,
+      "loss": 0.141,
+      "step": 18940
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.982643731597706e-06,
+      "loss": 0.2511,
+      "step": 18950
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.967147063381373e-06,
+      "loss": 0.1161,
+      "step": 18960
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.95165039516504e-06,
+      "loss": 0.1686,
+      "step": 18970
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.936153726948706e-06,
+      "loss": 0.2254,
+      "step": 18980
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.920657058732373e-06,
+      "loss": 0.079,
+      "step": 18990
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.905160390516039e-06,
+      "loss": 0.0219,
+      "step": 19000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.889663722299705e-06,
+      "loss": 0.1602,
+      "step": 19010
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.874167054083372e-06,
+      "loss": 0.0658,
+      "step": 19020
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.858670385867039e-06,
+      "loss": 0.0332,
+      "step": 19030
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.843173717650705e-06,
+      "loss": 0.3521,
+      "step": 19040
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.827677049434372e-06,
+      "loss": 0.0052,
+      "step": 19050
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.812180381218038e-06,
+      "loss": 0.2244,
+      "step": 19060
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.7966837130017048e-06,
+      "loss": 0.2198,
+      "step": 19070
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.781187044785371e-06,
+      "loss": 0.1291,
+      "step": 19080
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.7656903765690376e-06,
+      "loss": 0.0885,
+      "step": 19090
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.7501937083527043e-06,
+      "loss": 0.1136,
+      "step": 19100
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.7346970401363705e-06,
+      "loss": 0.0143,
+      "step": 19110
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.719200371920037e-06,
+      "loss": 0.1678,
+      "step": 19120
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.101,
+      "step": 19130
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.6882070354873704e-06,
+      "loss": 0.0724,
+      "step": 19140
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.672710367271037e-06,
+      "loss": 0.172,
+      "step": 19150
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.6572136990547037e-06,
+      "loss": 0.058,
+      "step": 19160
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.64171703083837e-06,
+      "loss": 0.0545,
+      "step": 19170
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.6262203626220365e-06,
+      "loss": 0.0608,
+      "step": 19180
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.610723694405703e-06,
+      "loss": 0.0018,
+      "step": 19190
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5952270261893694e-06,
+      "loss": 0.1482,
+      "step": 19200
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.579730357973036e-06,
+      "loss": 0.1264,
+      "step": 19210
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5642336897567026e-06,
+      "loss": 0.058,
+      "step": 19220
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.548737021540369e-06,
+      "loss": 0.184,
+      "step": 19230
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.5332403533240355e-06,
+      "loss": 0.1888,
+      "step": 19240
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.517743685107702e-06,
+      "loss": 0.0683,
+      "step": 19250
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.5022470168913683e-06,
+      "loss": 0.0876,
+      "step": 19260
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.486750348675035e-06,
+      "loss": 0.0173,
+      "step": 19270
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.4712536804587016e-06,
+      "loss": 0.3668,
+      "step": 19280
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.4557570122423682e-06,
+      "loss": 0.1494,
+      "step": 19290
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.4402603440260345e-06,
+      "loss": 0.0465,
+      "step": 19300
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.424763675809701e-06,
+      "loss": 0.2211,
+      "step": 19310
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.4092670075933677e-06,
+      "loss": 0.1247,
+      "step": 19320
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.393770339377034e-06,
+      "loss": 0.0963,
+      "step": 19330
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.3782736711607006e-06,
+      "loss": 0.0996,
+      "step": 19340
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.3627770029443672e-06,
+      "loss": 0.1035,
+      "step": 19350
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9776094830424761,
+      "eval_loss": 0.1219358742237091,
+      "eval_runtime": 17.1629,
+      "eval_samples_per_second": 176.951,
+      "eval_steps_per_second": 44.281,
+      "step": 19359
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.3472803347280334e-06,
+      "loss": 0.1543,
+      "step": 19360
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.3317836665117e-06,
+      "loss": 0.0617,
+      "step": 19370
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.3162869982953667e-06,
+      "loss": 0.1007,
+      "step": 19380
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.300790330079033e-06,
+      "loss": 0.158,
+      "step": 19390
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2852936618626996e-06,
+      "loss": 0.0319,
+      "step": 19400
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.269796993646366e-06,
+      "loss": 0.2304,
+      "step": 19410
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.2543003254300324e-06,
+      "loss": 0.0612,
+      "step": 19420
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.238803657213699e-06,
+      "loss": 0.1506,
+      "step": 19430
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.2233069889973657e-06,
+      "loss": 0.0054,
+      "step": 19440
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.207810320781032e-06,
+      "loss": 0.15,
+      "step": 19450
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.1923136525646985e-06,
+      "loss": 0.2068,
+      "step": 19460
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.176816984348365e-06,
+      "loss": 0.0853,
+      "step": 19470
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.1613203161320314e-06,
+      "loss": 0.0149,
+      "step": 19480
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.145823647915698e-06,
+      "loss": 0.0633,
+      "step": 19490
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.1303269796993647e-06,
+      "loss": 0.1486,
+      "step": 19500
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.114830311483031e-06,
+      "loss": 0.1867,
+      "step": 19510
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0993336432666975e-06,
+      "loss": 0.0844,
+      "step": 19520
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.083836975050364e-06,
+      "loss": 0.0564,
+      "step": 19530
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.068340306834031e-06,
+      "loss": 0.076,
+      "step": 19540
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.052843638617697e-06,
+      "loss": 0.1528,
+      "step": 19550
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.0373469704013636e-06,
+      "loss": 0.2519,
+      "step": 19560
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.0218503021850307e-06,
+      "loss": 0.1174,
+      "step": 19570
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 3.006353633968697e-06,
+      "loss": 0.1181,
+      "step": 19580
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9908569657523636e-06,
+      "loss": 0.121,
+      "step": 19590
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.97536029753603e-06,
+      "loss": 0.0385,
+      "step": 19600
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.9598636293196964e-06,
+      "loss": 0.0413,
+      "step": 19610
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.944366961103363e-06,
+      "loss": 0.1413,
+      "step": 19620
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.9288702928870297e-06,
+      "loss": 0.0248,
+      "step": 19630
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.913373624670696e-06,
+      "loss": 0.1415,
+      "step": 19640
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8978769564543625e-06,
+      "loss": 0.139,
+      "step": 19650
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.882380288238029e-06,
+      "loss": 0.0755,
+      "step": 19660
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8668836200216954e-06,
+      "loss": 0.1845,
+      "step": 19670
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.851386951805362e-06,
+      "loss": 0.0354,
+      "step": 19680
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.8358902835890287e-06,
+      "loss": 0.1874,
+      "step": 19690
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.820393615372695e-06,
+      "loss": 0.0131,
+      "step": 19700
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.8048969471563615e-06,
+      "loss": 0.147,
+      "step": 19710
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7909499457616614e-06,
+      "loss": 0.3334,
+      "step": 19720
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7754532775453276e-06,
+      "loss": 0.2987,
+      "step": 19730
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.7599566093289942e-06,
+      "loss": 0.0704,
+      "step": 19740
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.744459941112661e-06,
+      "loss": 0.1546,
+      "step": 19750
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.728963272896327e-06,
+      "loss": 0.0019,
+      "step": 19760
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.7134666046799937e-06,
+      "loss": 0.0304,
+      "step": 19770
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6979699364636604e-06,
+      "loss": 0.0064,
+      "step": 19780
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6824732682473266e-06,
+      "loss": 0.1358,
+      "step": 19790
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.6669766000309932e-06,
+      "loss": 0.0273,
+      "step": 19800
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.65147993181466e-06,
+      "loss": 0.0752,
+      "step": 19810
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.635983263598326e-06,
+      "loss": 0.2115,
+      "step": 19820
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.620486595381993e-06,
+      "loss": 0.0997,
+      "step": 19830
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.6049899271656598e-06,
+      "loss": 0.0718,
+      "step": 19840
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.589493258949326e-06,
+      "loss": 0.1535,
+      "step": 19850
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5739965907329926e-06,
+      "loss": 0.0045,
+      "step": 19860
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.5584999225166593e-06,
+      "loss": 0.0463,
+      "step": 19870
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.5430032543003255e-06,
+      "loss": 0.1217,
+      "step": 19880
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.527506586083992e-06,
+      "loss": 0.0023,
+      "step": 19890
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.5120099178676588e-06,
+      "loss": 0.1413,
+      "step": 19900
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.496513249651325e-06,
+      "loss": 0.1249,
+      "step": 19910
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4810165814349916e-06,
+      "loss": 0.2606,
+      "step": 19920
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.4655199132186582e-06,
+      "loss": 0.0981,
+      "step": 19930
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.4500232450023245e-06,
+      "loss": 0.0172,
+      "step": 19940
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.434526576785991e-06,
+      "loss": 0.2191,
+      "step": 19950
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.4190299085696577e-06,
+      "loss": 0.126,
+      "step": 19960
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.4035332403533244e-06,
+      "loss": 0.0535,
+      "step": 19970
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3880365721369906e-06,
+      "loss": 0.1782,
+      "step": 19980
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3725399039206572e-06,
+      "loss": 0.0335,
+      "step": 19990
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.357043235704324e-06,
+      "loss": 0.061,
+      "step": 20000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.34154656748799e-06,
+      "loss": 0.1047,
+      "step": 20010
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.3260498992716567e-06,
+      "loss": 0.0417,
+      "step": 20020
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.3105532310553233e-06,
+      "loss": 0.1548,
+      "step": 20030
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2950565628389896e-06,
+      "loss": 0.1097,
+      "step": 20040
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.279559894622656e-06,
+      "loss": 0.1449,
+      "step": 20050
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.264063226406323e-06,
+      "loss": 0.1538,
+      "step": 20060
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.248566558189989e-06,
+      "loss": 0.0429,
+      "step": 20070
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.2330698899736557e-06,
+      "loss": 0.2382,
+      "step": 20080
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.2175732217573223e-06,
+      "loss": 0.2756,
+      "step": 20090
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.2020765535409885e-06,
+      "loss": 0.0469,
+      "step": 20100
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.186579885324655e-06,
+      "loss": 0.4543,
+      "step": 20110
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.171083217108322e-06,
+      "loss": 0.1226,
+      "step": 20120
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.155586548891988e-06,
+      "loss": 0.1591,
+      "step": 20130
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1400898806756547e-06,
+      "loss": 0.043,
+      "step": 20140
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.1245932124593213e-06,
+      "loss": 0.0016,
+      "step": 20150
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.1090965442429875e-06,
+      "loss": 0.0589,
+      "step": 20160
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.093599876026654e-06,
+      "loss": 0.1102,
+      "step": 20170
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0781032078103208e-06,
+      "loss": 0.035,
+      "step": 20180
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.062606539593987e-06,
+      "loss": 0.0646,
+      "step": 20190
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0471098713776536e-06,
+      "loss": 0.1542,
+      "step": 20200
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.0316132031613203e-06,
+      "loss": 0.0998,
+      "step": 20210
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.016116534944987e-06,
+      "loss": 0.0874,
+      "step": 20220
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.000619866728653e-06,
+      "loss": 0.0861,
+      "step": 20230
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9851231985123198e-06,
+      "loss": 0.0186,
+      "step": 20240
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9696265302959864e-06,
+      "loss": 0.1726,
+      "step": 20250
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9541298620796526e-06,
+      "loss": 0.0626,
+      "step": 20260
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9386331938633197e-06,
+      "loss": 0.0663,
+      "step": 20270
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.9231365256469863e-06,
+      "loss": 0.0071,
+      "step": 20280
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.9076398574306525e-06,
+      "loss": 0.3218,
+      "step": 20290
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8921431892143192e-06,
+      "loss": 0.1188,
+      "step": 20300
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8766465209979856e-06,
+      "loss": 0.0067,
+      "step": 20310
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.861149852781652e-06,
+      "loss": 0.2181,
+      "step": 20320
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8456531845653184e-06,
+      "loss": 0.2211,
+      "step": 20330
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8301565163489849e-06,
+      "loss": 0.1074,
+      "step": 20340
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8146598481326515e-06,
+      "loss": 0.1332,
+      "step": 20350
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.799163179916318e-06,
+      "loss": 0.0103,
+      "step": 20360
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7836665116999844e-06,
+      "loss": 0.1835,
+      "step": 20370
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7681698434836512e-06,
+      "loss": 0.2221,
+      "step": 20380
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7526731752673176e-06,
+      "loss": 0.0057,
+      "step": 20390
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.737176507050984e-06,
+      "loss": 0.4088,
+      "step": 20400
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.7216798388346507e-06,
+      "loss": 0.1164,
+      "step": 20410
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.7061831706183171e-06,
+      "loss": 0.1692,
+      "step": 20420
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6906865024019838e-06,
+      "loss": 0.2444,
+      "step": 20430
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6751898341856502e-06,
+      "loss": 0.1398,
+      "step": 20440
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6596931659693166e-06,
+      "loss": 0.0526,
+      "step": 20450
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6441964977529832e-06,
+      "loss": 0.0673,
+      "step": 20460
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.6286998295366497e-06,
+      "loss": 0.1691,
+      "step": 20470
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.613203161320316e-06,
+      "loss": 0.1642,
+      "step": 20480
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5977064931039827e-06,
+      "loss": 0.1555,
+      "step": 20490
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5822098248876492e-06,
+      "loss": 0.0718,
+      "step": 20500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5667131566713156e-06,
+      "loss": 0.0409,
+      "step": 20510
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5512164884549822e-06,
+      "loss": 0.2544,
+      "step": 20520
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5357198202386486e-06,
+      "loss": 0.122,
+      "step": 20530
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.5202231520223153e-06,
+      "loss": 0.0019,
+      "step": 20540
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.5047264838059817e-06,
+      "loss": 0.0084,
+      "step": 20550
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4892298155896481e-06,
+      "loss": 0.053,
+      "step": 20560
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4737331473733148e-06,
+      "loss": 0.0945,
+      "step": 20570
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4582364791569812e-06,
+      "loss": 0.1029,
+      "step": 20580
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4427398109406478e-06,
+      "loss": 0.0045,
+      "step": 20590
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.4272431427243145e-06,
+      "loss": 0.0014,
+      "step": 20600
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.411746474507981e-06,
+      "loss": 0.1044,
+      "step": 20610
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3962498062916473e-06,
+      "loss": 0.1153,
+      "step": 20620
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.380753138075314e-06,
+      "loss": 0.0612,
+      "step": 20630
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3652564698589804e-06,
+      "loss": 0.0897,
+      "step": 20640
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3497598016426468e-06,
+      "loss": 0.0051,
+      "step": 20650
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3342631334263134e-06,
+      "loss": 0.1668,
+      "step": 20660
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3187664652099799e-06,
+      "loss": 0.0149,
+      "step": 20670
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.3032697969936465e-06,
+      "loss": 0.0824,
+      "step": 20680
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.287773128777313e-06,
+      "loss": 0.052,
+      "step": 20690
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2722764605609794e-06,
+      "loss": 0.2652,
+      "step": 20700
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.256779792344646e-06,
+      "loss": 0.1066,
+      "step": 20710
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2412831241283124e-06,
+      "loss": 0.0279,
+      "step": 20720
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.2257864559119789e-06,
+      "loss": 0.1973,
+      "step": 20730
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.2102897876956455e-06,
+      "loss": 0.0739,
+      "step": 20740
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.194793119479312e-06,
+      "loss": 0.2884,
+      "step": 20750
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1792964512629783e-06,
+      "loss": 0.0648,
+      "step": 20760
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.163799783046645e-06,
+      "loss": 0.1665,
+      "step": 20770
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1483031148303114e-06,
+      "loss": 0.092,
+      "step": 20780
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1328064466139778e-06,
+      "loss": 0.0071,
+      "step": 20790
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1173097783976445e-06,
+      "loss": 0.081,
+      "step": 20800
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.101813110181311e-06,
+      "loss": 0.1156,
+      "step": 20810
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0863164419649777e-06,
+      "loss": 0.2028,
+      "step": 20820
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0708197737486442e-06,
+      "loss": 0.1469,
+      "step": 20830
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0553231055323106e-06,
+      "loss": 0.3479,
+      "step": 20840
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0398264373159772e-06,
+      "loss": 0.0947,
+      "step": 20850
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.0243297690996437e-06,
+      "loss": 0.0593,
+      "step": 20860
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.00883310088331e-06,
+      "loss": 0.2242,
+      "step": 20870
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.933364326669767e-07,
+      "loss": 0.1784,
+      "step": 20880
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.778397644506431e-07,
+      "loss": 0.0704,
+      "step": 20890
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.623430962343096e-07,
+      "loss": 0.074,
+      "step": 20900
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.468464280179762e-07,
+      "loss": 0.2409,
+      "step": 20910
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.313497598016426e-07,
+      "loss": 0.0582,
+      "step": 20920
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.158530915853092e-07,
+      "loss": 0.0093,
+      "step": 20930
+    },
     {
       "epoch": 9.74,
-      "learning_rate": 8.620316222655404e-07,
-      "loss": 0.001,
-      "step": 15000
+      "learning_rate": 9.003564233689757e-07,
+      "loss": 0.0868,
+      "step": 20940
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.848597551526421e-07,
+      "loss": 0.0017,
+      "step": 20950
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.693630869363087e-07,
+      "loss": 0.0244,
+      "step": 20960
     },
     {
       "epoch": 9.75,
-      "learning_rate": 8.403725362789691e-07,
-      "loss": 0.0316,
-      "step": 15010
+      "learning_rate": 8.538664187199753e-07,
+      "loss": 0.0585,
+      "step": 20970
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.383697505036418e-07,
+      "loss": 0.2519,
+      "step": 20980
     },
     {
       "epoch": 9.76,
-      "learning_rate": 8.187134502923977e-07,
-      "loss": 0.0583,
-      "step": 15020
+      "learning_rate": 8.228730822873082e-07,
+      "loss": 0.2187,
+      "step": 20990
     },
     {
       "epoch": 9.76,
-      "learning_rate": 7.970543643058263e-07,
-      "loss": 0.3065,
-      "step": 15030
+      "learning_rate": 8.073764140709748e-07,
+      "loss": 0.1091,
+      "step": 21000
     },
     {
       "epoch": 9.77,
-      "learning_rate": 7.75395278319255e-07,
-      "loss": 0.0385,
-      "step": 15040
+      "learning_rate": 7.918797458546413e-07,
+      "loss": 0.0514,
+      "step": 21010
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.763830776383077e-07,
+      "loss": 0.0412,
+      "step": 21020
     },
     {
       "epoch": 9.78,
-      "learning_rate": 7.537361923326836e-07,
-      "loss": 0.0798,
-      "step": 15050
+      "learning_rate": 7.608864094219743e-07,
+      "loss": 0.125,
+      "step": 21030
     },
     {
       "epoch": 9.78,
-      "learning_rate": 7.320771063461123e-07,
-      "loss": 0.0639,
-      "step": 15060
+      "learning_rate": 7.453897412056408e-07,
+      "loss": 0.1665,
+      "step": 21040
     },
     {
       "epoch": 9.79,
-      "learning_rate": 7.104180203595409e-07,
-      "loss": 0.1085,
-      "step": 15070
+      "learning_rate": 7.298930729893073e-07,
+      "loss": 0.3963,
+      "step": 21050
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.143964047729738e-07,
+      "loss": 0.1001,
+      "step": 21060
     },
     {
       "epoch": 9.8,
-      "learning_rate": 6.887589343729695e-07,
-      "loss": 0.0518,
-      "step": 15080
+      "learning_rate": 6.988997365566403e-07,
+      "loss": 0.0236,
+      "step": 21070
     },
     {
       "epoch": 9.8,
-      "learning_rate": 6.670998483863981e-07,
-      "loss": 0.16,
-      "step": 15090
+      "learning_rate": 6.834030683403069e-07,
+      "loss": 0.1739,
+      "step": 21080
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.679064001239734e-07,
+      "loss": 0.1423,
+      "step": 21090
     },
     {
       "epoch": 9.81,
-      "learning_rate": 6.454407623998268e-07,
-      "loss": 0.0204,
-      "step": 15100
+      "learning_rate": 6.524097319076399e-07,
+      "loss": 0.061,
+      "step": 21100
     },
     {
       "epoch": 9.81,
-      "learning_rate": 6.237816764132554e-07,
-      "loss": 0.1528,
-      "step": 15110
+      "learning_rate": 6.369130636913064e-07,
+      "loss": 0.0032,
+      "step": 21110
     },
     {
       "epoch": 9.82,
-      "learning_rate": 6.02122590426684e-07,
-      "loss": 0.1124,
-      "step": 15120
+      "learning_rate": 6.214163954749729e-07,
+      "loss": 0.0032,
+      "step": 21120
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.059197272586394e-07,
+      "loss": 0.0029,
+      "step": 21130
     },
     {
       "epoch": 9.83,
-      "learning_rate": 5.804635044401126e-07,
-      "loss": 0.169,
-      "step": 15130
+      "learning_rate": 5.904230590423059e-07,
+      "loss": 0.0771,
+      "step": 21140
     },
     {
       "epoch": 9.83,
-      "learning_rate": 5.588044184535412e-07,
-      "loss": 0.0796,
-      "step": 15140
+      "learning_rate": 5.749263908259724e-07,
+      "loss": 0.0379,
+      "step": 21150
     },
     {
       "epoch": 9.84,
-      "learning_rate": 5.371453324669699e-07,
-      "loss": 0.1531,
-      "step": 15150
+      "learning_rate": 5.59429722609639e-07,
+      "loss": 0.2727,
+      "step": 21160
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.439330543933054e-07,
+      "loss": 0.0425,
+      "step": 21170
     },
     {
       "epoch": 9.85,
-      "learning_rate": 5.154862464803985e-07,
-      "loss": 0.1019,
-      "step": 15160
+      "learning_rate": 5.284363861769719e-07,
+      "loss": 0.0654,
+      "step": 21180
     },
     {
       "epoch": 9.85,
-      "learning_rate": 4.938271604938272e-07,
-      "loss": 0.0018,
-      "step": 15170
+      "learning_rate": 5.129397179606386e-07,
+      "loss": 0.0291,
+      "step": 21190
     },
     {
       "epoch": 9.86,
-      "learning_rate": 4.7216807450725577e-07,
-      "loss": 0.021,
-      "step": 15180
+      "learning_rate": 4.97443049744305e-07,
+      "loss": 0.0981,
+      "step": 21200
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.819463815279715e-07,
+      "loss": 0.3052,
+      "step": 21210
     },
     {
       "epoch": 9.87,
-      "learning_rate": 4.505089885206844e-07,
-      "loss": 0.0517,
-      "step": 15190
+      "learning_rate": 4.6644971331163805e-07,
+      "loss": 0.087,
+      "step": 21220
     },
     {
       "epoch": 9.87,
-      "learning_rate": 4.2884990253411303e-07,
-      "loss": 0.0759,
-      "step": 15200
+      "learning_rate": 4.5095304509530453e-07,
+      "loss": 0.1416,
+      "step": 21230
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.35456376878971e-07,
+      "loss": 0.0807,
+      "step": 21240
     },
     {
       "epoch": 9.88,
-      "learning_rate": 4.071908165475417e-07,
-      "loss": 0.2548,
-      "step": 15210
+      "learning_rate": 4.1995970866263754e-07,
+      "loss": 0.0561,
+      "step": 21250
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.04463040446304e-07,
+      "loss": 0.215,
+      "step": 21260
     },
     {
       "epoch": 9.89,
-      "learning_rate": 3.8553173056097035e-07,
-      "loss": 0.1891,
-      "step": 15220
+      "learning_rate": 3.889663722299706e-07,
+      "loss": 0.2723,
+      "step": 21270
     },
     {
       "epoch": 9.89,
-      "learning_rate": 3.63872644574399e-07,
-      "loss": 0.1289,
-      "step": 15230
+      "learning_rate": 3.734697040136371e-07,
+      "loss": 0.0546,
+      "step": 21280
     },
     {
       "epoch": 9.9,
-      "learning_rate": 3.422135585878276e-07,
-      "loss": 0.0315,
-      "step": 15240
+      "learning_rate": 3.579730357973036e-07,
+      "loss": 0.108,
+      "step": 21290
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.424763675809701e-07,
+      "loss": 0.0655,
+      "step": 21300
     },
     {
       "epoch": 9.91,
-      "learning_rate": 3.2055447260125624e-07,
-      "loss": 0.0379,
-      "step": 15250
+      "learning_rate": 3.269796993646366e-07,
+      "loss": 0.4156,
+      "step": 21310
     },
     {
       "epoch": 9.91,
-      "learning_rate": 2.988953866146849e-07,
-      "loss": 0.1247,
-      "step": 15260
+      "learning_rate": 3.114830311483031e-07,
+      "loss": 0.1386,
+      "step": 21320
     },
     {
       "epoch": 9.92,
-      "learning_rate": 2.772363006281135e-07,
-      "loss": 0.1522,
-      "step": 15270
+      "learning_rate": 2.9598636293196963e-07,
+      "loss": 0.4026,
+      "step": 21330
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.8048969471563616e-07,
+      "loss": 0.167,
+      "step": 21340
     },
     {
       "epoch": 9.93,
-      "learning_rate": 2.5557721464154214e-07,
-      "loss": 0.1842,
-      "step": 15280
+      "learning_rate": 2.6499302649930264e-07,
+      "loss": 0.1878,
+      "step": 21350
     },
     {
       "epoch": 9.93,
-      "learning_rate": 2.3391812865497075e-07,
-      "loss": 0.1641,
-      "step": 15290
+      "learning_rate": 2.4949635828296917e-07,
+      "loss": 0.1537,
+      "step": 21360
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.3399969006663568e-07,
+      "loss": 0.0609,
+      "step": 21370
     },
     {
       "epoch": 9.94,
-      "learning_rate": 2.122590426683994e-07,
-      "loss": 0.0641,
-      "step": 15300
+      "learning_rate": 2.1850302185030218e-07,
+      "loss": 0.3141,
+      "step": 21380
     },
     {
       "epoch": 9.94,
-      "learning_rate": 1.9059995668182804e-07,
-      "loss": 0.0768,
-      "step": 15310
+      "learning_rate": 2.030063536339687e-07,
+      "loss": 0.2971,
+      "step": 21390
     },
     {
       "epoch": 9.95,
-      "learning_rate": 1.6894087069525667e-07,
-      "loss": 0.0525,
-      "step": 15320
+      "learning_rate": 1.8750968541763522e-07,
+      "loss": 0.1231,
+      "step": 21400
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.7201301720130172e-07,
+      "loss": 0.0516,
+      "step": 21410
     },
     {
       "epoch": 9.96,
-      "learning_rate": 1.472817847086853e-07,
-      "loss": 0.0982,
-      "step": 15330
+      "learning_rate": 1.5651634898496825e-07,
+      "loss": 0.2032,
+      "step": 21420
     },
     {
       "epoch": 9.96,
-      "learning_rate": 1.256226987221139e-07,
-      "loss": 0.197,
-      "step": 15340
+      "learning_rate": 1.4101968076863473e-07,
+      "loss": 0.0427,
+      "step": 21430
     },
     {
       "epoch": 9.97,
-      "learning_rate": 1.0396361273554257e-07,
-      "loss": 0.1669,
-      "step": 15350
+      "learning_rate": 1.2552301255230124e-07,
+      "loss": 0.1642,
+      "step": 21440
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1002634433596777e-07,
+      "loss": 0.1457,
+      "step": 21450
     },
     {
       "epoch": 9.98,
-      "learning_rate": 8.230452674897118e-08,
-      "loss": 0.0749,
-      "step": 15360
+      "learning_rate": 9.452967611963427e-08,
+      "loss": 0.1659,
+      "step": 21460
     },
     {
       "epoch": 9.98,
-      "learning_rate": 6.064544076239983e-08,
-      "loss": 0.1194,
-      "step": 15370
+      "learning_rate": 7.903300790330079e-08,
+      "loss": 0.1698,
+      "step": 21470
     },
     {
       "epoch": 9.99,
-      "learning_rate": 3.898635477582846e-08,
-      "loss": 0.1512,
-      "step": 15380
+      "learning_rate": 6.353633968696731e-08,
+      "loss": 0.1093,
+      "step": 21480
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.8039671470633816e-08,
+      "loss": 0.2687,
+      "step": 21490
     },
     {
       "epoch": 10.0,
-      "learning_rate": 1.7327268789257094e-08,
-      "loss": 0.068,
-      "step": 15390
+      "learning_rate": 3.254300325430033e-08,
+      "loss": 0.0359,
+      "step": 21500
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.985280588776449,
-      "eval_loss": 0.0865689218044281,
-      "eval_runtime": 13.3643,
-      "eval_samples_per_second": 162.673,
-      "eval_steps_per_second": 40.706,
-      "step": 15390
+      "learning_rate": 1.704633503796684e-08,
+      "loss": 0.1148,
+      "step": 21510
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.9782680276588739,
+      "eval_loss": 0.11879671365022659,
+      "eval_runtime": 17.2485,
+      "eval_samples_per_second": 176.073,
+      "eval_steps_per_second": 44.062,
+      "step": 21510
     },
     {
       "epoch": 10.0,
-      "step": 15390,
-      "total_flos": 1.1936517061793697e+18,
-      "train_loss": 0.29998462175571583,
-      "train_runtime": 2944.6409,
-      "train_samples_per_second": 41.825,
-      "train_steps_per_second": 5.226
+      "step": 21510,
+      "total_flos": 1.7013193016084636e+18,
+      "train_loss": 0.3477993663641144,
+      "train_runtime": 4068.2187,
+      "train_samples_per_second": 42.294,
+      "train_steps_per_second": 5.287
     }
   ],
   "logging_steps": 10,
-  "max_steps": 15390,
+  "max_steps": 21510,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 10,
   "save_steps": 500,
-  "total_flos": 1.1936517061793697e+18,
+  "total_flos": 1.7013193016084636e+18,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null