diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11956 @@
+{
+  "best_metric": 0.8205392852535157,
+  "best_model_checkpoint": "videomae-base-finetuned-ucf101/checkpoint-19780",
+  "epoch": 4.2,
+  "global_step": 19780,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.527805864509606e-07,
+      "loss": 4.68,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.055611729019212e-07,
+      "loss": 4.6907,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.583417593528817e-07,
+      "loss": 4.6742,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0111223458038423e-06,
+      "loss": 4.804,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2639029322548029e-06,
+      "loss": 4.7056,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5166835187057635e-06,
+      "loss": 4.5885,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.769464105156724e-06,
+      "loss": 4.7113,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0222446916076846e-06,
+      "loss": 4.678,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2750252780586454e-06,
+      "loss": 4.7039,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5278058645096058e-06,
+      "loss": 4.7001,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7805864509605666e-06,
+      "loss": 4.7141,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.033367037411527e-06,
+      "loss": 4.6267,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2861476238624873e-06,
+      "loss": 4.6421,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.538928210313448e-06,
+      "loss": 4.7045,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.791708796764409e-06,
+      "loss": 4.6262,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.044489383215369e-06,
+      "loss": 4.6217,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2972699696663305e-06,
+      "loss": 4.589,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.550050556117291e-06,
+      "loss": 4.7515,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.802831142568251e-06,
+      "loss": 4.5265,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.0556117290192116e-06,
+      "loss": 4.6884,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.308392315470173e-06,
+      "loss": 4.6772,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.561172901921133e-06,
+      "loss": 4.5202,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8139534883720935e-06,
+      "loss": 4.5874,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.066734074823054e-06,
+      "loss": 4.4802,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.319514661274015e-06,
+      "loss": 4.5604,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.572295247724975e-06,
+      "loss": 4.6428,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.825075834175936e-06,
+      "loss": 4.6809,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.077856420626896e-06,
+      "loss": 4.5347,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3306370070778566e-06,
+      "loss": 4.6068,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.583417593528818e-06,
+      "loss": 4.6327,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.836198179979779e-06,
+      "loss": 4.5552,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.088978766430739e-06,
+      "loss": 4.6179,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.3417593528817e-06,
+      "loss": 4.461,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.594539939332661e-06,
+      "loss": 4.6736,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.84732052578362e-06,
+      "loss": 4.5406,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.100101112234582e-06,
+      "loss": 4.6944,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.352881698685541e-06,
+      "loss": 4.5823,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.605662285136502e-06,
+      "loss": 4.69,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.858442871587464e-06,
+      "loss": 4.7124,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0111223458038423e-05,
+      "loss": 4.5055,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0364004044489384e-05,
+      "loss": 4.6744,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0616784630940346e-05,
+      "loss": 4.6469,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 4.6291,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1122345803842266e-05,
+      "loss": 4.6061,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1375126390293226e-05,
+      "loss": 4.649,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1627906976744187e-05,
+      "loss": 4.5493,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1880687563195148e-05,
+      "loss": 4.6308,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2133468149646108e-05,
+      "loss": 4.6582,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2386248736097069e-05,
+      "loss": 4.5301,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.263902932254803e-05,
+      "loss": 4.4677,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.289180990899899e-05,
+      "loss": 4.5065,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.314459049544995e-05,
+      "loss": 4.5028,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.339737108190091e-05,
+      "loss": 4.547,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3650151668351872e-05,
+      "loss": 4.7245,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3902932254802831e-05,
+      "loss": 4.5828,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4155712841253792e-05,
+      "loss": 4.5556,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4408493427704754e-05,
+      "loss": 4.5894,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4661274014155713e-05,
+      "loss": 4.4456,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4914054600606674e-05,
+      "loss": 4.5844,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5166835187057636e-05,
+      "loss": 4.5346,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5419615773508595e-05,
+      "loss": 4.6335,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5672396359959558e-05,
+      "loss": 4.5046,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5925176946410518e-05,
+      "loss": 4.6511,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6177957532861477e-05,
+      "loss": 4.7109,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.643073811931244e-05,
+      "loss": 4.6781,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.66835187057634e-05,
+      "loss": 4.611,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.693629929221436e-05,
+      "loss": 4.528,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7189079878665322e-05,
+      "loss": 4.411,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.744186046511628e-05,
+      "loss": 4.5609,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.769464105156724e-05,
+      "loss": 4.5649,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.79474216380182e-05,
+      "loss": 4.4835,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8200202224469163e-05,
+      "loss": 4.401,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8452982810920123e-05,
+      "loss": 4.3958,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8705763397371082e-05,
+      "loss": 4.4687,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8958543983822045e-05,
+      "loss": 4.5011,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9211324570273005e-05,
+      "loss": 4.2695,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9464105156723964e-05,
+      "loss": 4.5461,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9716885743174927e-05,
+      "loss": 4.564,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9969666329625887e-05,
+      "loss": 4.5565,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0222446916076846e-05,
+      "loss": 4.4434,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.047522750252781e-05,
+      "loss": 4.3052,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.072800808897877e-05,
+      "loss": 4.2266,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0980788675429728e-05,
+      "loss": 4.3449,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.123356926188069e-05,
+      "loss": 4.4353,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.148634984833165e-05,
+      "loss": 4.3212,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.173913043478261e-05,
+      "loss": 4.322,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991911021233573e-05,
+      "loss": 4.5171,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2244691607684533e-05,
+      "loss": 4.3732,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2497472194135492e-05,
+      "loss": 4.4087,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.275025278058645e-05,
+      "loss": 4.2639,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3003033367037415e-05,
+      "loss": 4.3718,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3255813953488374e-05,
+      "loss": 4.5052,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3508594539939334e-05,
+      "loss": 4.3786,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3761375126390297e-05,
+      "loss": 4.1556,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4014155712841256e-05,
+      "loss": 4.1752,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4266936299292216e-05,
+      "loss": 3.7999,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.451971688574318e-05,
+      "loss": 4.0865,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.4772497472194138e-05,
+      "loss": 4.4178,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5025278058645097e-05,
+      "loss": 4.2508,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.527805864509606e-05,
+      "loss": 4.0827,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5530839231547017e-05,
+      "loss": 4.1762,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.578361981799798e-05,
+      "loss": 4.2111,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.6036400404448942e-05,
+      "loss": 4.082,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.62891809908999e-05,
+      "loss": 4.2653,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.654196157735086e-05,
+      "loss": 4.2191,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.679474216380182e-05,
+      "loss": 4.2105,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.704752275025278e-05,
+      "loss": 4.1019,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7300303336703743e-05,
+      "loss": 4.1689,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7553083923154703e-05,
+      "loss": 4.3119,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7805864509605662e-05,
+      "loss": 4.0664,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8058645096056625e-05,
+      "loss": 4.3021,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8311425682507585e-05,
+      "loss": 3.9197,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8564206268958544e-05,
+      "loss": 4.0239,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8816986855409507e-05,
+      "loss": 4.2242,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9069767441860467e-05,
+      "loss": 3.9362,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9322548028311426e-05,
+      "loss": 4.1823,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.957532861476239e-05,
+      "loss": 4.1407,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.982810920121335e-05,
+      "loss": 4.1283,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0080889787664308e-05,
+      "loss": 3.7764,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.033367037411527e-05,
+      "loss": 4.1983,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0586450960566234e-05,
+      "loss": 3.9246,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.083923154701719e-05,
+      "loss": 4.0646,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.109201213346815e-05,
+      "loss": 4.1365,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1344792719919116e-05,
+      "loss": 4.0168,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.159757330637007e-05,
+      "loss": 3.8619,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1850353892821035e-05,
+      "loss": 3.8768,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.2103134479272e-05,
+      "loss": 3.8374,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.2355915065722954e-05,
+      "loss": 3.9794,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.260869565217392e-05,
+      "loss": 3.9338,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.286147623862488e-05,
+      "loss": 3.9935,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.3114256825075836e-05,
+      "loss": 3.7239,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.33670374115268e-05,
+      "loss": 3.6177,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.361981799797776e-05,
+      "loss": 4.1565,
+      "step": 1330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.387259858442872e-05,
+      "loss": 4.0735,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.412537917087968e-05,
+      "loss": 4.1169,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4378159757330644e-05,
+      "loss": 3.8773,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.46309403437816e-05,
+      "loss": 3.8583,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.488372093023256e-05,
+      "loss": 3.7216,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.513650151668352e-05,
+      "loss": 4.0468,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.538928210313448e-05,
+      "loss": 3.4753,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5642062689585445e-05,
+      "loss": 4.0625,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.58948432760364e-05,
+      "loss": 3.8681,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6147623862487364e-05,
+      "loss": 3.8417,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.640040444893833e-05,
+      "loss": 3.735,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.665318503538928e-05,
+      "loss": 3.5258,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6905965621840246e-05,
+      "loss": 3.7124,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.715874620829121e-05,
+      "loss": 3.656,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7411526794742165e-05,
+      "loss": 3.7881,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.766430738119313e-05,
+      "loss": 3.775,
+      "step": 1490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.791708796764409e-05,
+      "loss": 3.6606,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.816986855409505e-05,
+      "loss": 3.7741,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.842264914054601e-05,
+      "loss": 3.1928,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.867542972699697e-05,
+      "loss": 3.7045,
+      "step": 1530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.892821031344793e-05,
+      "loss": 3.9253,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.918099089989889e-05,
+      "loss": 3.6154,
+      "step": 1550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9433771486349854e-05,
+      "loss": 3.6973,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.968655207280081e-05,
+      "loss": 3.6486,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9939332659251774e-05,
+      "loss": 3.414,
+      "step": 1580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.0192113245702736e-05,
+      "loss": 3.6696,
+      "step": 1590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.044489383215369e-05,
+      "loss": 3.0864,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.0697674418604655e-05,
+      "loss": 3.6524,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.095045500505562e-05,
+      "loss": 3.5955,
+      "step": 1620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.1203235591506575e-05,
+      "loss": 3.1339,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.145601617795754e-05,
+      "loss": 3.1835,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.17087967644085e-05,
+      "loss": 3.5494,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.1961577350859456e-05,
+      "loss": 3.4065,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.221435793731042e-05,
+      "loss": 3.2393,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.246713852376138e-05,
+      "loss": 3.1573,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.271991911021234e-05,
+      "loss": 3.5554,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.29726996966633e-05,
+      "loss": 3.5552,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.3225480283114264e-05,
+      "loss": 3.6993,
+      "step": 1710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.347826086956522e-05,
+      "loss": 2.9049,
+      "step": 1720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.373104145601618e-05,
+      "loss": 3.6891,
+      "step": 1730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.3983822042467146e-05,
+      "loss": 3.3102,
+      "step": 1740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.42366026289181e-05,
+      "loss": 3.51,
+      "step": 1750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4489383215369065e-05,
+      "loss": 3.7335,
+      "step": 1760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.474216380182003e-05,
+      "loss": 3.734,
+      "step": 1770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4994944388270984e-05,
+      "loss": 3.112,
+      "step": 1780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.524772497472195e-05,
+      "loss": 3.4752,
+      "step": 1790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.55005055611729e-05,
+      "loss": 3.0113,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5753286147623866e-05,
+      "loss": 3.442,
+      "step": 1810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.600606673407483e-05,
+      "loss": 3.1304,
+      "step": 1820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6258847320525785e-05,
+      "loss": 3.0062,
+      "step": 1830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.651162790697675e-05,
+      "loss": 3.2251,
+      "step": 1840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.676440849342771e-05,
+      "loss": 3.1552,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.701718907987867e-05,
+      "loss": 3.0863,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.726996966632963e-05,
+      "loss": 2.6046,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.752275025278059e-05,
+      "loss": 2.9637,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.777553083923155e-05,
+      "loss": 3.177,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.802831142568251e-05,
+      "loss": 3.3438,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8281092012133475e-05,
+      "loss": 3.0758,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.853387259858443e-05,
+      "loss": 3.1334,
+      "step": 1920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8786653185035394e-05,
+      "loss": 3.0423,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.903943377148636e-05,
+      "loss": 3.0376,
+      "step": 1940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.929221435793731e-05,
+      "loss": 2.9132,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9544994944388276e-05,
+      "loss": 2.9035,
+      "step": 1960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.979777553083924e-05,
+      "loss": 2.8816,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999438265363443e-05,
+      "loss": 3.0069,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.996629592180654e-05,
+      "loss": 3.1285,
+      "step": 1990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9938209189978655e-05,
+      "loss": 3.1156,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.991012245815077e-05,
+      "loss": 2.998,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.988203572632289e-05,
+      "loss": 3.1716,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9853948994495005e-05,
+      "loss": 2.8679,
+      "step": 2030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9825862262667115e-05,
+      "loss": 3.3838,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.979777553083924e-05,
+      "loss": 3.3697,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.976968879901135e-05,
+      "loss": 2.9499,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9741602067183466e-05,
+      "loss": 3.3205,
+      "step": 2070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.971351533535558e-05,
+      "loss": 3.1428,
+      "step": 2080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.968542860352769e-05,
+      "loss": 2.8391,
+      "step": 2090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9657341871699816e-05,
+      "loss": 2.5695,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9629255139871926e-05,
+      "loss": 2.9872,
+      "step": 2110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.960116840804404e-05,
+      "loss": 3.2846,
+      "step": 2120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.957308167621616e-05,
+      "loss": 2.9725,
+      "step": 2130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9544994944388276e-05,
+      "loss": 3.0749,
+      "step": 2140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9516908212560386e-05,
+      "loss": 3.003,
+      "step": 2150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.94888214807325e-05,
+      "loss": 2.8792,
+      "step": 2160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.946073474890462e-05,
+      "loss": 2.7323,
+      "step": 2170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9432648017076736e-05,
+      "loss": 3.1303,
+      "step": 2180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.940456128524885e-05,
+      "loss": 2.7736,
+      "step": 2190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.937647455342096e-05,
+      "loss": 2.6223,
+      "step": 2200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9348387821593086e-05,
+      "loss": 2.8449,
+      "step": 2210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9320301089765196e-05,
+      "loss": 2.8315,
+      "step": 2220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.929221435793731e-05,
+      "loss": 3.0564,
+      "step": 2230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.926412762610943e-05,
+      "loss": 2.6239,
+      "step": 2240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9236040894281547e-05,
+      "loss": 2.8278,
+      "step": 2250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9207954162453656e-05,
+      "loss": 2.8739,
+      "step": 2260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.917986743062577e-05,
+      "loss": 2.5618,
+      "step": 2270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.915178069879789e-05,
+      "loss": 2.9008,
+      "step": 2280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.912369396697001e-05,
+      "loss": 2.6284,
+      "step": 2290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9095607235142123e-05,
+      "loss": 1.9172,
+      "step": 2300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.906752050331423e-05,
+      "loss": 2.5929,
+      "step": 2310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.903943377148636e-05,
+      "loss": 3.0718,
+      "step": 2320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901134703965847e-05,
+      "loss": 2.7974,
+      "step": 2330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8983260307830584e-05,
+      "loss": 2.833,
+      "step": 2340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89551735760027e-05,
+      "loss": 2.788,
+      "step": 2350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892708684417481e-05,
+      "loss": 2.216,
+      "step": 2360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8899000112346934e-05,
+      "loss": 2.6245,
+      "step": 2370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8870913380519044e-05,
+      "loss": 2.8334,
+      "step": 2380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.884282664869116e-05,
+      "loss": 2.5791,
+      "step": 2390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.881473991686328e-05,
+      "loss": 2.4468,
+      "step": 2400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8786653185035394e-05,
+      "loss": 2.1307,
+      "step": 2410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8758566453207504e-05,
+      "loss": 2.6411,
+      "step": 2420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.873047972137962e-05,
+      "loss": 2.5514,
+      "step": 2430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.870239298955174e-05,
+      "loss": 2.5553,
+      "step": 2440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8674306257723854e-05,
+      "loss": 2.7294,
+      "step": 2450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.864621952589597e-05,
+      "loss": 2.7977,
+      "step": 2460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.861813279406808e-05,
+      "loss": 2.6219,
+      "step": 2470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8590046062240204e-05,
+      "loss": 2.3086,
+      "step": 2480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8561959330412314e-05,
+      "loss": 1.9584,
+      "step": 2490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.853387259858443e-05,
+      "loss": 2.442,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.850578586675655e-05,
+      "loss": 2.3716,
+      "step": 2510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8477699134928665e-05,
+      "loss": 2.3909,
+      "step": 2520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8449612403100775e-05,
+      "loss": 2.5932,
+      "step": 2530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.842152567127289e-05,
+      "loss": 3.0255,
+      "step": 2540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.839343893944501e-05,
+      "loss": 2.2027,
+      "step": 2550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8365352207617125e-05,
+      "loss": 2.4466,
+      "step": 2560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.833726547578924e-05,
+      "loss": 2.7153,
+      "step": 2570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.830917874396135e-05,
+      "loss": 1.8205,
+      "step": 2580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8281092012133475e-05,
+      "loss": 2.8319,
+      "step": 2590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8253005280305585e-05,
+      "loss": 2.8584,
+      "step": 2600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.82249185484777e-05,
+      "loss": 2.8757,
+      "step": 2610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.819683181664982e-05,
+      "loss": 2.5624,
+      "step": 2620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.816874508482193e-05,
+      "loss": 2.431,
+      "step": 2630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.814065835299405e-05,
+      "loss": 2.5779,
+      "step": 2640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.811257162116616e-05,
+      "loss": 2.7651,
+      "step": 2650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.808448488933828e-05,
+      "loss": 2.7778,
+      "step": 2660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8056398157510395e-05,
+      "loss": 2.4839,
+      "step": 2670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.802831142568251e-05,
+      "loss": 2.4368,
+      "step": 2680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.800022469385462e-05,
+      "loss": 1.8938,
+      "step": 2690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.797213796202674e-05,
+      "loss": 2.0687,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7944051230198855e-05,
+      "loss": 1.9147,
+      "step": 2710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.791596449837097e-05,
+      "loss": 2.9428,
+      "step": 2720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.788787776654309e-05,
+      "loss": 2.2333,
+      "step": 2730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78597910347152e-05,
+      "loss": 2.4392,
+      "step": 2740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.783170430288732e-05,
+      "loss": 2.0556,
+      "step": 2750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.780361757105943e-05,
+      "loss": 2.1116,
+      "step": 2760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777553083923155e-05,
+      "loss": 1.8684,
+      "step": 2770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7747444107403666e-05,
+      "loss": 2.3838,
+      "step": 2780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771935737557578e-05,
+      "loss": 1.899,
+      "step": 2790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76912706437479e-05,
+      "loss": 2.3284,
+      "step": 2800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766318391192001e-05,
+      "loss": 1.9155,
+      "step": 2810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7635097180092126e-05,
+      "loss": 1.8833,
+      "step": 2820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.760701044826424e-05,
+      "loss": 2.5877,
+      "step": 2830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.757892371643636e-05,
+      "loss": 2.4796,
+      "step": 2840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.755083698460847e-05,
+      "loss": 2.2923,
+      "step": 2850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.752275025278059e-05,
+      "loss": 2.3423,
+      "step": 2860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74946635209527e-05,
+      "loss": 2.8767,
+      "step": 2870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.746657678912482e-05,
+      "loss": 2.3688,
+      "step": 2880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7438490057296936e-05,
+      "loss": 2.3477,
+      "step": 2890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7410403325469046e-05,
+      "loss": 2.3324,
+      "step": 2900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.738231659364117e-05,
+      "loss": 2.4138,
+      "step": 2910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.735422986181328e-05,
+      "loss": 2.5056,
+      "step": 2920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.73261431299854e-05,
+      "loss": 2.062,
+      "step": 2930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.729805639815751e-05,
+      "loss": 2.0543,
+      "step": 2940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.726996966632963e-05,
+      "loss": 2.1489,
+      "step": 2950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.724188293450174e-05,
+      "loss": 2.0132,
+      "step": 2960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.721379620267386e-05,
+      "loss": 1.9671,
+      "step": 2970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7185709470845974e-05,
+      "loss": 2.078,
+      "step": 2980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.715762273901809e-05,
+      "loss": 2.07,
+      "step": 2990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.712953600719021e-05,
+      "loss": 2.9755,
+      "step": 3000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.710144927536232e-05,
+      "loss": 2.3831,
+      "step": 3010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.707336254353444e-05,
+      "loss": 1.8275,
+      "step": 3020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.704527581170655e-05,
+      "loss": 2.0956,
+      "step": 3030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.701718907987867e-05,
+      "loss": 2.0516,
+      "step": 3040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.6989102348050784e-05,
+      "loss": 2.2697,
+      "step": 3050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.69610156162229e-05,
+      "loss": 2.3516,
+      "step": 3060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693292888439502e-05,
+      "loss": 2.0196,
+      "step": 3070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.690484215256713e-05,
+      "loss": 2.0119,
+      "step": 3080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6876755420739244e-05,
+      "loss": 1.7081,
+      "step": 3090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.684866868891136e-05,
+      "loss": 1.9191,
+      "step": 3100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.682058195708348e-05,
+      "loss": 1.8302,
+      "step": 3110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.679249522525559e-05,
+      "loss": 2.3978,
+      "step": 3120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.676440849342771e-05,
+      "loss": 2.0549,
+      "step": 3130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.673632176159982e-05,
+      "loss": 2.5816,
+      "step": 3140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.670823502977194e-05,
+      "loss": 1.9995,
+      "step": 3150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6680148297944054e-05,
+      "loss": 1.6153,
+      "step": 3160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6652061566116164e-05,
+      "loss": 1.4241,
+      "step": 3170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.662397483428829e-05,
+      "loss": 1.9988,
+      "step": 3180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.65958881024604e-05,
+      "loss": 1.6617,
+      "step": 3190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6567801370632515e-05,
+      "loss": 1.662,
+      "step": 3200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.653971463880463e-05,
+      "loss": 1.9961,
+      "step": 3210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.651162790697675e-05,
+      "loss": 1.9676,
+      "step": 3220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.648354117514886e-05,
+      "loss": 2.2615,
+      "step": 3230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6455454443320975e-05,
+      "loss": 2.0938,
+      "step": 3240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.642736771149309e-05,
+      "loss": 1.9016,
+      "step": 3250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.639928097966521e-05,
+      "loss": 1.9549,
+      "step": 3260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6371194247837325e-05,
+      "loss": 2.3842,
+      "step": 3270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6343107516009435e-05,
+      "loss": 1.8681,
+      "step": 3280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.631502078418156e-05,
+      "loss": 2.4576,
+      "step": 3290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.628693405235367e-05,
+      "loss": 2.1882,
+      "step": 3300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6258847320525785e-05,
+      "loss": 1.7006,
+      "step": 3310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.62307605886979e-05,
+      "loss": 1.8845,
+      "step": 3320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.620267385687002e-05,
+      "loss": 2.1922,
+      "step": 3330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6174587125042135e-05,
+      "loss": 1.5775,
+      "step": 3340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6146500393214245e-05,
+      "loss": 2.0459,
+      "step": 3350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.611841366138636e-05,
+      "loss": 1.8625,
+      "step": 3360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.609032692955848e-05,
+      "loss": 1.5601,
+      "step": 3370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6062240197730596e-05,
+      "loss": 1.881,
+      "step": 3380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6034153465902706e-05,
+      "loss": 1.6504,
+      "step": 3390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.600606673407483e-05,
+      "loss": 2.1222,
+      "step": 3400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.597798000224694e-05,
+      "loss": 1.8576,
+      "step": 3410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5949893270419056e-05,
+      "loss": 1.8391,
+      "step": 3420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.592180653859117e-05,
+      "loss": 1.6178,
+      "step": 3430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.589371980676328e-05,
+      "loss": 2.0147,
+      "step": 3440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5865633074935406e-05,
+      "loss": 1.7665,
+      "step": 3450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5837546343107516e-05,
+      "loss": 1.6191,
+      "step": 3460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.580945961127963e-05,
+      "loss": 1.9692,
+      "step": 3470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.578137287945175e-05,
+      "loss": 1.5127,
+      "step": 3480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5753286147623866e-05,
+      "loss": 1.3596,
+      "step": 3490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.572519941579598e-05,
+      "loss": 1.5121,
+      "step": 3500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.56971126839681e-05,
+      "loss": 2.1781,
+      "step": 3510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.566902595214021e-05,
+      "loss": 1.3654,
+      "step": 3520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5640939220312326e-05,
+      "loss": 2.5457,
+      "step": 3530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.561285248848444e-05,
+      "loss": 1.2025,
+      "step": 3540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.558476575665655e-05,
+      "loss": 1.5711,
+      "step": 3550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5556679024828677e-05,
+      "loss": 1.969,
+      "step": 3560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5528592293000787e-05,
+      "loss": 1.5156,
+      "step": 3570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.55005055611729e-05,
+      "loss": 1.8262,
+      "step": 3580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.547241882934502e-05,
+      "loss": 1.5133,
+      "step": 3590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.544433209751714e-05,
+      "loss": 1.3922,
+      "step": 3600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5416245365689254e-05,
+      "loss": 1.1827,
+      "step": 3610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5388158633861363e-05,
+      "loss": 1.2597,
+      "step": 3620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.536007190203348e-05,
+      "loss": 2.1833,
+      "step": 3630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.53319851702056e-05,
+      "loss": 1.4932,
+      "step": 3640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5303898438377714e-05,
+      "loss": 1.6185,
+      "step": 3650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5275811706549824e-05,
+      "loss": 1.7571,
+      "step": 3660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.524772497472195e-05,
+      "loss": 1.6672,
+      "step": 3670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.521963824289406e-05,
+      "loss": 1.7736,
+      "step": 3680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5191551511066174e-05,
+      "loss": 1.6212,
+      "step": 3690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.516346477923829e-05,
+      "loss": 1.5307,
+      "step": 3700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.51353780474104e-05,
+      "loss": 2.3102,
+      "step": 3710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5107291315582524e-05,
+      "loss": 1.4021,
+      "step": 3720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5079204583754634e-05,
+      "loss": 1.6835,
+      "step": 3730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.505111785192675e-05,
+      "loss": 1.8789,
+      "step": 3740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.502303112009887e-05,
+      "loss": 2.054,
+      "step": 3750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.4994944388270984e-05,
+      "loss": 2.0126,
+      "step": 3760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.49668576564431e-05,
+      "loss": 1.7743,
+      "step": 3770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.493877092461522e-05,
+      "loss": 1.7667,
+      "step": 3780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.491068419278733e-05,
+      "loss": 1.6471,
+      "step": 3790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.4882597460959444e-05,
+      "loss": 1.4901,
+      "step": 3800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.485451072913156e-05,
+      "loss": 1.58,
+      "step": 3810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.482642399730367e-05,
+      "loss": 1.6836,
+      "step": 3820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.4798337265475795e-05,
+      "loss": 1.6808,
+      "step": 3830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.4770250533647905e-05,
+      "loss": 1.1758,
+      "step": 3840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.474216380182003e-05,
+      "loss": 1.375,
+      "step": 3850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.471407706999214e-05,
+      "loss": 1.6212,
+      "step": 3860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.4685990338164255e-05,
+      "loss": 1.2041,
+      "step": 3870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.465790360633637e-05,
+      "loss": 2.1536,
+      "step": 3880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.462981687450848e-05,
+      "loss": 1.4693,
+      "step": 3890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.46017301426806e-05,
+      "loss": 1.3537,
+      "step": 3900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.4573643410852715e-05,
+      "loss": 1.173,
+      "step": 3910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.454555667902483e-05,
+      "loss": 1.8422,
+      "step": 3920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.451746994719694e-05,
+      "loss": 1.5115,
+      "step": 3930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.4489383215369065e-05,
+      "loss": 1.8926,
+      "step": 3940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.4461296483541175e-05,
+      "loss": 1.0704,
+      "step": 3950
+    },
+    {
+      "epoch": 0.2,
+      "eval_accuracy": 0.5346406915236743,
+      "eval_loss": 1.758298397064209,
+      "eval_runtime": 1317.0676,
+      "eval_samples_per_second": 5.885,
+      "eval_steps_per_second": 2.943,
+      "step": 3956
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.443320975171329e-05,
+      "loss": 1.536,
+      "step": 3960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.440512301988541e-05,
+      "loss": 1.69,
+      "step": 3970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.437703628805752e-05,
+      "loss": 1.0675,
+      "step": 3980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.434894955622964e-05,
+      "loss": 1.0894,
+      "step": 3990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.432086282440175e-05,
+      "loss": 1.1409,
+      "step": 4000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.429277609257387e-05,
+      "loss": 1.8051,
+      "step": 4010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4264689360745986e-05,
+      "loss": 1.7109,
+      "step": 4020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.42366026289181e-05,
+      "loss": 1.2647,
+      "step": 4030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.420851589709022e-05,
+      "loss": 1.8677,
+      "step": 4040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4180429165262336e-05,
+      "loss": 1.5998,
+      "step": 4050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4152342433434446e-05,
+      "loss": 1.171,
+      "step": 4060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.412425570160656e-05,
+      "loss": 1.5152,
+      "step": 4070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.409616896977868e-05,
+      "loss": 1.355,
+      "step": 4080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.406808223795079e-05,
+      "loss": 1.0396,
+      "step": 4090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.403999550612291e-05,
+      "loss": 1.5342,
+      "step": 4100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.401190877429502e-05,
+      "loss": 2.1099,
+      "step": 4110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3983822042467146e-05,
+      "loss": 1.4267,
+      "step": 4120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3955735310639256e-05,
+      "loss": 1.2166,
+      "step": 4130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.392764857881137e-05,
+      "loss": 1.0253,
+      "step": 4140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.389956184698349e-05,
+      "loss": 0.9691,
+      "step": 4150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.38714751151556e-05,
+      "loss": 1.2448,
+      "step": 4160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3843388383327716e-05,
+      "loss": 1.2505,
+      "step": 4170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.381530165149983e-05,
+      "loss": 0.8117,
+      "step": 4180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.378721491967195e-05,
+      "loss": 1.4709,
+      "step": 4190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3759128187844067e-05,
+      "loss": 1.4287,
+      "step": 4200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.373104145601618e-05,
+      "loss": 1.0941,
+      "step": 4210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.370295472418829e-05,
+      "loss": 1.7235,
+      "step": 4220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.367486799236041e-05,
+      "loss": 1.6953,
+      "step": 4230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.364678126053253e-05,
+      "loss": 1.054,
+      "step": 4240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.361869452870464e-05,
+      "loss": 0.9221,
+      "step": 4250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.359060779687676e-05,
+      "loss": 0.8579,
+      "step": 4260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.356252106504887e-05,
+      "loss": 1.0348,
+      "step": 4270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.353443433322099e-05,
+      "loss": 1.2875,
+      "step": 4280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3506347601393104e-05,
+      "loss": 1.0705,
+      "step": 4290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.347826086956522e-05,
+      "loss": 0.8026,
+      "step": 4300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.345017413773734e-05,
+      "loss": 1.1767,
+      "step": 4310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3422087405909454e-05,
+      "loss": 1.1856,
+      "step": 4320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3394000674081564e-05,
+      "loss": 0.9064,
+      "step": 4330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.336591394225368e-05,
+      "loss": 1.003,
+      "step": 4340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.33378272104258e-05,
+      "loss": 0.7749,
+      "step": 4350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.330974047859791e-05,
+      "loss": 0.8747,
+      "step": 4360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.328165374677003e-05,
+      "loss": 0.989,
+      "step": 4370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.325356701494214e-05,
+      "loss": 1.3786,
+      "step": 4380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3225480283114264e-05,
+      "loss": 1.4611,
+      "step": 4390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3197393551286374e-05,
+      "loss": 1.2834,
+      "step": 4400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.316930681945849e-05,
+      "loss": 1.3459,
+      "step": 4410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.314122008763061e-05,
+      "loss": 0.8945,
+      "step": 4420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.311313335580272e-05,
+      "loss": 1.2581,
+      "step": 4430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3085046623974834e-05,
+      "loss": 1.1246,
+      "step": 4440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.305695989214695e-05,
+      "loss": 1.7202,
+      "step": 4450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.302887316031907e-05,
+      "loss": 0.8289,
+      "step": 4460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.3000786428491185e-05,
+      "loss": 0.4677,
+      "step": 4470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.29726996966633e-05,
+      "loss": 1.5624,
+      "step": 4480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.294461296483541e-05,
+      "loss": 1.6121,
+      "step": 4490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.291652623300753e-05,
+      "loss": 0.998,
+      "step": 4500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.2888439501179645e-05,
+      "loss": 1.2382,
+      "step": 4510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.286035276935176e-05,
+      "loss": 1.2114,
+      "step": 4520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.283226603752388e-05,
+      "loss": 2.0071,
+      "step": 4530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.280417930569599e-05,
+      "loss": 1.507,
+      "step": 4540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.277609257386811e-05,
+      "loss": 1.2564,
+      "step": 4550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.274800584204022e-05,
+      "loss": 0.9962,
+      "step": 4560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.271991911021234e-05,
+      "loss": 1.0782,
+      "step": 4570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.2691832378384455e-05,
+      "loss": 1.5052,
+      "step": 4580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.266374564655657e-05,
+      "loss": 1.2718,
+      "step": 4590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.263565891472868e-05,
+      "loss": 1.1049,
+      "step": 4600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.26075721829008e-05,
+      "loss": 1.2068,
+      "step": 4610
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.2579485451072915e-05,
+      "loss": 1.0709,
+      "step": 4620
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.2551398719245025e-05,
+      "loss": 0.9509,
+      "step": 4630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.252331198741715e-05,
+      "loss": 1.1732,
+      "step": 4640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.249522525558926e-05,
+      "loss": 1.2884,
+      "step": 4650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.246713852376138e-05,
+      "loss": 1.0682,
+      "step": 4660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.243905179193349e-05,
+      "loss": 0.8404,
+      "step": 4670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.241096506010561e-05,
+      "loss": 1.0531,
+      "step": 4680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.2382878328277726e-05,
+      "loss": 0.8525,
+      "step": 4690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.2354791596449836e-05,
+      "loss": 0.8313,
+      "step": 4700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.232670486462195e-05,
+      "loss": 0.8385,
+      "step": 4710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.229861813279407e-05,
+      "loss": 0.8567,
+      "step": 4720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.2270531400966186e-05,
+      "loss": 1.3185,
+      "step": 4730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.22424446691383e-05,
+      "loss": 1.4366,
+      "step": 4740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.221435793731042e-05,
+      "loss": 0.952,
+      "step": 4750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.218627120548253e-05,
+      "loss": 1.3992,
+      "step": 4760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.2158184473654646e-05,
+      "loss": 0.9214,
+      "step": 4770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.213009774182676e-05,
+      "loss": 1.3596,
+      "step": 4780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.210201100999888e-05,
+      "loss": 1.0587,
+      "step": 4790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.2073924278170996e-05,
+      "loss": 1.2828,
+      "step": 4800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.2045837546343106e-05,
+      "loss": 1.0506,
+      "step": 4810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.201775081451523e-05,
+      "loss": 1.0066,
+      "step": 4820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.198966408268734e-05,
+      "loss": 1.2051,
+      "step": 4830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1961577350859456e-05,
+      "loss": 0.8884,
+      "step": 4840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.193349061903157e-05,
+      "loss": 1.1716,
+      "step": 4850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.190540388720369e-05,
+      "loss": 0.9561,
+      "step": 4860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.18773171553758e-05,
+      "loss": 1.2945,
+      "step": 4870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.184923042354792e-05,
+      "loss": 1.1603,
+      "step": 4880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.182114369172003e-05,
+      "loss": 1.1885,
+      "step": 4890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.179305695989214e-05,
+      "loss": 0.9057,
+      "step": 4900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.176497022806427e-05,
+      "loss": 1.4051,
+      "step": 4910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.173688349623638e-05,
+      "loss": 1.2735,
+      "step": 4920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.17087967644085e-05,
+      "loss": 1.1922,
+      "step": 4930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.168071003258061e-05,
+      "loss": 0.8904,
+      "step": 4940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.165262330075273e-05,
+      "loss": 0.9882,
+      "step": 4950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.1624536568924844e-05,
+      "loss": 1.0433,
+      "step": 4960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.1596449837096954e-05,
+      "loss": 1.2635,
+      "step": 4970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.156836310526907e-05,
+      "loss": 0.8175,
+      "step": 4980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.154027637344119e-05,
+      "loss": 0.9645,
+      "step": 4990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.1512189641613304e-05,
+      "loss": 1.1371,
+      "step": 5000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.148410290978542e-05,
+      "loss": 1.6277,
+      "step": 5010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.145601617795754e-05,
+      "loss": 1.2775,
+      "step": 5020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.142792944612965e-05,
+      "loss": 0.7418,
+      "step": 5030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.139984271430177e-05,
+      "loss": 0.549,
+      "step": 5040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.137175598247388e-05,
+      "loss": 1.1559,
+      "step": 5050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.1343669250646e-05,
+      "loss": 1.1057,
+      "step": 5060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.1315582518818114e-05,
+      "loss": 1.2046,
+      "step": 5070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.1287495786990224e-05,
+      "loss": 1.0863,
+      "step": 5080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.125940905516235e-05,
+      "loss": 1.0463,
+      "step": 5090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.123132232333446e-05,
+      "loss": 0.6999,
+      "step": 5100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.1203235591506575e-05,
+      "loss": 0.7451,
+      "step": 5110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.117514885967869e-05,
+      "loss": 1.0702,
+      "step": 5120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.114706212785081e-05,
+      "loss": 0.4564,
+      "step": 5130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.111897539602292e-05,
+      "loss": 0.8598,
+      "step": 5140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.1090888664195035e-05,
+      "loss": 0.8266,
+      "step": 5150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.106280193236715e-05,
+      "loss": 0.7084,
+      "step": 5160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.103471520053927e-05,
+      "loss": 1.3511,
+      "step": 5170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.1006628468711385e-05,
+      "loss": 1.1446,
+      "step": 5180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0978541736883495e-05,
+      "loss": 1.2583,
+      "step": 5190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.095045500505562e-05,
+      "loss": 1.1378,
+      "step": 5200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.092236827322773e-05,
+      "loss": 0.7655,
+      "step": 5210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0894281541399845e-05,
+      "loss": 0.6871,
+      "step": 5220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.086619480957196e-05,
+      "loss": 0.7515,
+      "step": 5230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.083810807774407e-05,
+      "loss": 1.2146,
+      "step": 5240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.081002134591619e-05,
+      "loss": 0.7645,
+      "step": 5250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0781934614088305e-05,
+      "loss": 0.8596,
+      "step": 5260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.075384788226042e-05,
+      "loss": 1.0691,
+      "step": 5270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.072576115043254e-05,
+      "loss": 0.8725,
+      "step": 5280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0697674418604655e-05,
+      "loss": 0.7043,
+      "step": 5290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0669587686776765e-05,
+      "loss": 1.3654,
+      "step": 5300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.064150095494889e-05,
+      "loss": 1.2759,
+      "step": 5310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0613414223121e-05,
+      "loss": 0.9521,
+      "step": 5320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0585327491293116e-05,
+      "loss": 1.1463,
+      "step": 5330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.055724075946523e-05,
+      "loss": 0.7876,
+      "step": 5340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.052915402763734e-05,
+      "loss": 1.6468,
+      "step": 5350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0501067295809466e-05,
+      "loss": 0.9478,
+      "step": 5360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0472980563981576e-05,
+      "loss": 0.9746,
+      "step": 5370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.044489383215369e-05,
+      "loss": 0.4251,
+      "step": 5380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.041680710032581e-05,
+      "loss": 0.6614,
+      "step": 5390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0388720368497926e-05,
+      "loss": 0.8028,
+      "step": 5400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0360633636670036e-05,
+      "loss": 0.9352,
+      "step": 5410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.033254690484215e-05,
+      "loss": 1.0759,
+      "step": 5420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.030446017301427e-05,
+      "loss": 1.0339,
+      "step": 5430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0276373441186386e-05,
+      "loss": 0.9672,
+      "step": 5440
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.02482867093585e-05,
+      "loss": 0.9615,
+      "step": 5450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.022019997753061e-05,
+      "loss": 0.8137,
+      "step": 5460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0192113245702736e-05,
+      "loss": 0.8012,
+      "step": 5470
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0164026513874846e-05,
+      "loss": 0.9234,
+      "step": 5480
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.013593978204696e-05,
+      "loss": 0.9225,
+      "step": 5490
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.010785305021908e-05,
+      "loss": 1.2532,
+      "step": 5500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.007976631839119e-05,
+      "loss": 0.568,
+      "step": 5510
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.005167958656331e-05,
+      "loss": 0.5888,
+      "step": 5520
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.002359285473542e-05,
+      "loss": 0.7198,
+      "step": 5530
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.999550612290754e-05,
+      "loss": 0.7062,
+      "step": 5540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.996741939107966e-05,
+      "loss": 1.0393,
+      "step": 5550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9939332659251774e-05,
+      "loss": 0.9848,
+      "step": 5560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9911245927423883e-05,
+      "loss": 1.2126,
+      "step": 5570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.988315919559601e-05,
+      "loss": 0.9815,
+      "step": 5580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.985507246376812e-05,
+      "loss": 0.7371,
+      "step": 5590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9826985731940234e-05,
+      "loss": 0.8056,
+      "step": 5600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.979889900011235e-05,
+      "loss": 0.815,
+      "step": 5610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.977081226828446e-05,
+      "loss": 0.7023,
+      "step": 5620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9742725536456584e-05,
+      "loss": 0.851,
+      "step": 5630
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9714638804628694e-05,
+      "loss": 1.4918,
+      "step": 5640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.968655207280081e-05,
+      "loss": 1.0232,
+      "step": 5650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.965846534097293e-05,
+      "loss": 0.9706,
+      "step": 5660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9630378609145044e-05,
+      "loss": 1.1325,
+      "step": 5670
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9602291877317154e-05,
+      "loss": 1.2365,
+      "step": 5680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.957420514548927e-05,
+      "loss": 0.8062,
+      "step": 5690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.954611841366139e-05,
+      "loss": 0.5341,
+      "step": 5700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9518031681833504e-05,
+      "loss": 1.2296,
+      "step": 5710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.948994495000562e-05,
+      "loss": 0.6538,
+      "step": 5720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.946185821817773e-05,
+      "loss": 1.1553,
+      "step": 5730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9433771486349854e-05,
+      "loss": 0.3904,
+      "step": 5740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9405684754521964e-05,
+      "loss": 1.3732,
+      "step": 5750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.937759802269408e-05,
+      "loss": 1.0621,
+      "step": 5760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.93495112908662e-05,
+      "loss": 0.5839,
+      "step": 5770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.932142455903831e-05,
+      "loss": 0.6424,
+      "step": 5780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.929333782721043e-05,
+      "loss": 0.9537,
+      "step": 5790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.926525109538254e-05,
+      "loss": 0.5404,
+      "step": 5800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.923716436355466e-05,
+      "loss": 0.8372,
+      "step": 5810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9209077631726775e-05,
+      "loss": 1.2897,
+      "step": 5820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.918099089989889e-05,
+      "loss": 0.8106,
+      "step": 5830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9152904168071e-05,
+      "loss": 0.8566,
+      "step": 5840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9124817436243125e-05,
+      "loss": 1.1213,
+      "step": 5850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9096730704415235e-05,
+      "loss": 0.9252,
+      "step": 5860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.906864397258735e-05,
+      "loss": 1.0017,
+      "step": 5870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.904055724075947e-05,
+      "loss": 0.8164,
+      "step": 5880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.901247050893158e-05,
+      "loss": 0.9821,
+      "step": 5890
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.89843837771037e-05,
+      "loss": 0.4936,
+      "step": 5900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.895629704527581e-05,
+      "loss": 1.066,
+      "step": 5910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.892821031344793e-05,
+      "loss": 0.4226,
+      "step": 5920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.8900123581620045e-05,
+      "loss": 0.7138,
+      "step": 5930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.887203684979216e-05,
+      "loss": 0.8918,
+      "step": 5940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.884395011796427e-05,
+      "loss": 0.4821,
+      "step": 5950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.881586338613639e-05,
+      "loss": 0.7073,
+      "step": 5960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.8787776654308506e-05,
+      "loss": 0.8115,
+      "step": 5970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.875968992248062e-05,
+      "loss": 1.0276,
+      "step": 5980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.873160319065274e-05,
+      "loss": 1.1758,
+      "step": 5990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.870351645882485e-05,
+      "loss": 0.4017,
+      "step": 6000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.867542972699697e-05,
+      "loss": 0.4894,
+      "step": 6010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.864734299516908e-05,
+      "loss": 0.8956,
+      "step": 6020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.86192562633412e-05,
+      "loss": 0.8063,
+      "step": 6030
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8591169531513316e-05,
+      "loss": 0.9415,
+      "step": 6040
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.856308279968543e-05,
+      "loss": 0.855,
+      "step": 6050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.853499606785755e-05,
+      "loss": 0.5531,
+      "step": 6060
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.850690933602966e-05,
+      "loss": 0.8103,
+      "step": 6070
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8478822604201776e-05,
+      "loss": 0.9308,
+      "step": 6080
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.845073587237389e-05,
+      "loss": 0.624,
+      "step": 6090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.842264914054601e-05,
+      "loss": 0.4623,
+      "step": 6100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.839456240871812e-05,
+      "loss": 0.7564,
+      "step": 6110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.836647567689024e-05,
+      "loss": 0.7589,
+      "step": 6120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.833838894506235e-05,
+      "loss": 0.8116,
+      "step": 6130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.831030221323447e-05,
+      "loss": 0.6276,
+      "step": 6140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8282215481406587e-05,
+      "loss": 0.6312,
+      "step": 6150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8254128749578697e-05,
+      "loss": 1.0153,
+      "step": 6160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.822604201775082e-05,
+      "loss": 0.7968,
+      "step": 6170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.819795528592293e-05,
+      "loss": 0.7362,
+      "step": 6180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.816986855409505e-05,
+      "loss": 1.1105,
+      "step": 6190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8141781822267163e-05,
+      "loss": 0.9504,
+      "step": 6200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.811369509043928e-05,
+      "loss": 0.4126,
+      "step": 6210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.80856083586114e-05,
+      "loss": 0.8754,
+      "step": 6220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.805752162678351e-05,
+      "loss": 0.6008,
+      "step": 6230
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8029434894955624e-05,
+      "loss": 0.628,
+      "step": 6240
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.800134816312774e-05,
+      "loss": 0.7699,
+      "step": 6250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.797326143129986e-05,
+      "loss": 0.6691,
+      "step": 6260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.794517469947197e-05,
+      "loss": 0.5924,
+      "step": 6270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.791708796764409e-05,
+      "loss": 1.1327,
+      "step": 6280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.78890012358162e-05,
+      "loss": 0.5681,
+      "step": 6290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.786091450398832e-05,
+      "loss": 0.709,
+      "step": 6300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.7832827772160434e-05,
+      "loss": 0.5845,
+      "step": 6310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.780474104033255e-05,
+      "loss": 0.8509,
+      "step": 6320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.777665430850467e-05,
+      "loss": 0.9219,
+      "step": 6330
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.774856757667678e-05,
+      "loss": 0.6495,
+      "step": 6340
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.7720480844848894e-05,
+      "loss": 1.4958,
+      "step": 6350
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.769239411302101e-05,
+      "loss": 0.5606,
+      "step": 6360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.766430738119313e-05,
+      "loss": 0.7901,
+      "step": 6370
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.763622064936524e-05,
+      "loss": 0.513,
+      "step": 6380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.760813391753736e-05,
+      "loss": 0.8909,
+      "step": 6390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.758004718570947e-05,
+      "loss": 0.4335,
+      "step": 6400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.755196045388159e-05,
+      "loss": 0.3082,
+      "step": 6410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.7523873722053705e-05,
+      "loss": 0.7107,
+      "step": 6420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7495786990225815e-05,
+      "loss": 0.3986,
+      "step": 6430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.746770025839794e-05,
+      "loss": 0.533,
+      "step": 6440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.743961352657005e-05,
+      "loss": 0.2938,
+      "step": 6450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7411526794742165e-05,
+      "loss": 0.5807,
+      "step": 6460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.738344006291428e-05,
+      "loss": 0.3448,
+      "step": 6470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.73553533310864e-05,
+      "loss": 0.2658,
+      "step": 6480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7327266599258515e-05,
+      "loss": 0.7665,
+      "step": 6490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7299179867430625e-05,
+      "loss": 0.3087,
+      "step": 6500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.727109313560274e-05,
+      "loss": 0.6744,
+      "step": 6510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.724300640377486e-05,
+      "loss": 0.5438,
+      "step": 6520
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7214919671946975e-05,
+      "loss": 1.1266,
+      "step": 6530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7186832940119085e-05,
+      "loss": 0.8935,
+      "step": 6540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.715874620829121e-05,
+      "loss": 0.2975,
+      "step": 6550
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.713065947646332e-05,
+      "loss": 0.501,
+      "step": 6560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.710257274463544e-05,
+      "loss": 0.8347,
+      "step": 6570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.707448601280755e-05,
+      "loss": 0.845,
+      "step": 6580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.704639928097967e-05,
+      "loss": 0.4272,
+      "step": 6590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7018312549151786e-05,
+      "loss": 0.4246,
+      "step": 6600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.6990225817323896e-05,
+      "loss": 1.0605,
+      "step": 6610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.696213908549601e-05,
+      "loss": 0.7649,
+      "step": 6620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.693405235366813e-05,
+      "loss": 1.2556,
+      "step": 6630
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6905965621840246e-05,
+      "loss": 0.7721,
+      "step": 6640
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6877878890012356e-05,
+      "loss": 0.2938,
+      "step": 6650
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.684979215818448e-05,
+      "loss": 0.4109,
+      "step": 6660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.682170542635659e-05,
+      "loss": 0.7686,
+      "step": 6670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6793618694528706e-05,
+      "loss": 1.0396,
+      "step": 6680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.676553196270082e-05,
+      "loss": 0.4376,
+      "step": 6690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.673744523087293e-05,
+      "loss": 0.5358,
+      "step": 6700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6709358499045056e-05,
+      "loss": 0.3283,
+      "step": 6710
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6681271767217166e-05,
+      "loss": 0.8013,
+      "step": 6720
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.665318503538928e-05,
+      "loss": 0.7522,
+      "step": 6730
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.66250983035614e-05,
+      "loss": 0.6111,
+      "step": 6740
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6597011571733516e-05,
+      "loss": 0.4451,
+      "step": 6750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.656892483990563e-05,
+      "loss": 0.5314,
+      "step": 6760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.654083810807774e-05,
+      "loss": 0.5492,
+      "step": 6770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.651275137624986e-05,
+      "loss": 0.6122,
+      "step": 6780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.6484664644421976e-05,
+      "loss": 0.7028,
+      "step": 6790
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.645657791259409e-05,
+      "loss": 0.8066,
+      "step": 6800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.64284911807662e-05,
+      "loss": 0.7244,
+      "step": 6810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.640040444893833e-05,
+      "loss": 0.3727,
+      "step": 6820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.637231771711044e-05,
+      "loss": 0.2827,
+      "step": 6830
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.634423098528256e-05,
+      "loss": 0.8827,
+      "step": 6840
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.631614425345467e-05,
+      "loss": 0.8029,
+      "step": 6850
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.628805752162679e-05,
+      "loss": 0.4557,
+      "step": 6860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6259970789798904e-05,
+      "loss": 0.8907,
+      "step": 6870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6231884057971014e-05,
+      "loss": 0.6931,
+      "step": 6880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.620379732614313e-05,
+      "loss": 0.3222,
+      "step": 6890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.617571059431525e-05,
+      "loss": 0.158,
+      "step": 6900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6147623862487364e-05,
+      "loss": 0.6244,
+      "step": 6910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6119537130659474e-05,
+      "loss": 0.2259,
+      "step": 6920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.60914503988316e-05,
+      "loss": 0.5667,
+      "step": 6930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.606336366700371e-05,
+      "loss": 0.6292,
+      "step": 6940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.6035276935175824e-05,
+      "loss": 0.1489,
+      "step": 6950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.600719020334794e-05,
+      "loss": 0.4443,
+      "step": 6960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.597910347152005e-05,
+      "loss": 0.4776,
+      "step": 6970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.5951016739692174e-05,
+      "loss": 0.4631,
+      "step": 6980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.5922930007864284e-05,
+      "loss": 0.3282,
+      "step": 6990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.58948432760364e-05,
+      "loss": 0.6206,
+      "step": 7000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.586675654420852e-05,
+      "loss": 0.1572,
+      "step": 7010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.5838669812380634e-05,
+      "loss": 0.35,
+      "step": 7020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.581058308055275e-05,
+      "loss": 0.5425,
+      "step": 7030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.578249634872486e-05,
+      "loss": 0.8094,
+      "step": 7040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.575440961689698e-05,
+      "loss": 1.2532,
+      "step": 7050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5726322885069095e-05,
+      "loss": 0.7594,
+      "step": 7060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.569823615324121e-05,
+      "loss": 0.2774,
+      "step": 7070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.567014942141332e-05,
+      "loss": 0.3976,
+      "step": 7080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5642062689585445e-05,
+      "loss": 0.3896,
+      "step": 7090
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5613975957757555e-05,
+      "loss": 0.5127,
+      "step": 7100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.558588922592968e-05,
+      "loss": 0.5183,
+      "step": 7110
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.555780249410179e-05,
+      "loss": 0.3577,
+      "step": 7120
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5529715762273905e-05,
+      "loss": 0.3121,
+      "step": 7130
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.550162903044602e-05,
+      "loss": 0.8092,
+      "step": 7140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.547354229861813e-05,
+      "loss": 0.7679,
+      "step": 7150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.544545556679025e-05,
+      "loss": 0.9464,
+      "step": 7160
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5417368834962365e-05,
+      "loss": 0.5635,
+      "step": 7170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.538928210313448e-05,
+      "loss": 1.0384,
+      "step": 7180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.53611953713066e-05,
+      "loss": 0.2994,
+      "step": 7190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5333108639478715e-05,
+      "loss": 0.4317,
+      "step": 7200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5305021907650825e-05,
+      "loss": 0.5196,
+      "step": 7210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.527693517582294e-05,
+      "loss": 0.0929,
+      "step": 7220
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.524884844399506e-05,
+      "loss": 0.3605,
+      "step": 7230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.522076171216717e-05,
+      "loss": 0.2597,
+      "step": 7240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.519267498033929e-05,
+      "loss": 0.3664,
+      "step": 7250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.51645882485114e-05,
+      "loss": 0.307,
+      "step": 7260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.513650151668352e-05,
+      "loss": 0.2905,
+      "step": 7270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.5108414784855636e-05,
+      "loss": 0.9549,
+      "step": 7280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.508032805302775e-05,
+      "loss": 0.5071,
+      "step": 7290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.505224132119987e-05,
+      "loss": 0.087,
+      "step": 7300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.502415458937198e-05,
+      "loss": 0.3774,
+      "step": 7310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.4996067857544096e-05,
+      "loss": 0.2979,
+      "step": 7320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.496798112571621e-05,
+      "loss": 0.854,
+      "step": 7330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.493989439388833e-05,
+      "loss": 0.6126,
+      "step": 7340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.491180766206044e-05,
+      "loss": 0.473,
+      "step": 7350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.488372093023256e-05,
+      "loss": 0.3205,
+      "step": 7360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.485563419840467e-05,
+      "loss": 0.6176,
+      "step": 7370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.4827547466576796e-05,
+      "loss": 0.5618,
+      "step": 7380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.4799460734748906e-05,
+      "loss": 0.4003,
+      "step": 7390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.477137400292102e-05,
+      "loss": 0.44,
+      "step": 7400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.474328727109314e-05,
+      "loss": 0.2038,
+      "step": 7410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.471520053926525e-05,
+      "loss": 0.3305,
+      "step": 7420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.4687113807437366e-05,
+      "loss": 0.691,
+      "step": 7430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.465902707560948e-05,
+      "loss": 0.5525,
+      "step": 7440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.46309403437816e-05,
+      "loss": 0.7706,
+      "step": 7450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.460285361195372e-05,
+      "loss": 0.5165,
+      "step": 7460
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.457476688012583e-05,
+      "loss": 0.9926,
+      "step": 7470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.454668014829794e-05,
+      "loss": 0.103,
+      "step": 7480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.451859341647006e-05,
+      "loss": 0.7124,
+      "step": 7490
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.449050668464218e-05,
+      "loss": 0.4166,
+      "step": 7500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.446241995281429e-05,
+      "loss": 0.2715,
+      "step": 7510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.443433322098641e-05,
+      "loss": 0.416,
+      "step": 7520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.440624648915852e-05,
+      "loss": 0.5691,
+      "step": 7530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.4378159757330644e-05,
+      "loss": 0.2682,
+      "step": 7540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.4350073025502754e-05,
+      "loss": 0.2504,
+      "step": 7550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.432198629367487e-05,
+      "loss": 0.5478,
+      "step": 7560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.429389956184699e-05,
+      "loss": 0.3484,
+      "step": 7570
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.4265812830019104e-05,
+      "loss": 0.8361,
+      "step": 7580
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.4237726098191214e-05,
+      "loss": 1.3379,
+      "step": 7590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.420963936636333e-05,
+      "loss": 0.6968,
+      "step": 7600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.418155263453545e-05,
+      "loss": 0.4283,
+      "step": 7610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.415346590270756e-05,
+      "loss": 0.7949,
+      "step": 7620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.412537917087968e-05,
+      "loss": 1.0044,
+      "step": 7630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.409729243905179e-05,
+      "loss": 0.6239,
+      "step": 7640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.4069205707223914e-05,
+      "loss": 0.3219,
+      "step": 7650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.4041118975396024e-05,
+      "loss": 0.3326,
+      "step": 7660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.401303224356814e-05,
+      "loss": 0.8751,
+      "step": 7670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.398494551174026e-05,
+      "loss": 0.564,
+      "step": 7680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.395685877991237e-05,
+      "loss": 0.7908,
+      "step": 7690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.3928772048084484e-05,
+      "loss": 0.7893,
+      "step": 7700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.39006853162566e-05,
+      "loss": 0.0763,
+      "step": 7710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.387259858442872e-05,
+      "loss": 0.5171,
+      "step": 7720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.3844511852600835e-05,
+      "loss": 0.4223,
+      "step": 7730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.381642512077295e-05,
+      "loss": 0.596,
+      "step": 7740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.378833838894506e-05,
+      "loss": 0.3406,
+      "step": 7750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.376025165711718e-05,
+      "loss": 0.6104,
+      "step": 7760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.3732164925289295e-05,
+      "loss": 0.6543,
+      "step": 7770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.3704078193461405e-05,
+      "loss": 0.7004,
+      "step": 7780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.367599146163353e-05,
+      "loss": 0.2828,
+      "step": 7790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.364790472980564e-05,
+      "loss": 0.2892,
+      "step": 7800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.361981799797776e-05,
+      "loss": 0.5403,
+      "step": 7810
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.359173126614987e-05,
+      "loss": 0.4357,
+      "step": 7820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.356364453432199e-05,
+      "loss": 0.1557,
+      "step": 7830
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.3535557802494105e-05,
+      "loss": 0.4338,
+      "step": 7840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.350747107066622e-05,
+      "loss": 0.2118,
+      "step": 7850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.347938433883833e-05,
+      "loss": 0.3819,
+      "step": 7860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.345129760701045e-05,
+      "loss": 0.7408,
+      "step": 7870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.3423210875182565e-05,
+      "loss": 0.222,
+      "step": 7880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.339512414335468e-05,
+      "loss": 0.7326,
+      "step": 7890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.33670374115268e-05,
+      "loss": 0.133,
+      "step": 7900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.333895067969891e-05,
+      "loss": 0.1936,
+      "step": 7910
+    },
+    {
+      "epoch": 1.2,
+      "eval_accuracy": 0.7188749838730486,
+      "eval_loss": 1.0779602527618408,
+      "eval_runtime": 1333.2284,
+      "eval_samples_per_second": 5.814,
+      "eval_steps_per_second": 2.907,
+      "step": 7912
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.331086394787103e-05,
+      "loss": 0.1527,
+      "step": 7920
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.328277721604314e-05,
+      "loss": 0.428,
+      "step": 7930
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.325469048421526e-05,
+      "loss": 0.2814,
+      "step": 7940
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3226603752387376e-05,
+      "loss": 0.3085,
+      "step": 7950
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3198517020559486e-05,
+      "loss": 0.3646,
+      "step": 7960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.31704302887316e-05,
+      "loss": 0.6297,
+      "step": 7970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.314234355690372e-05,
+      "loss": 0.0743,
+      "step": 7980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3114256825075836e-05,
+      "loss": 0.6047,
+      "step": 7990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.308617009324795e-05,
+      "loss": 0.0741,
+      "step": 8000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.305808336142007e-05,
+      "loss": 0.0561,
+      "step": 8010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.302999662959218e-05,
+      "loss": 0.4328,
+      "step": 8020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3001909897764296e-05,
+      "loss": 0.3761,
+      "step": 8030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.297382316593641e-05,
+      "loss": 0.5726,
+      "step": 8040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.294573643410852e-05,
+      "loss": 0.635,
+      "step": 8050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.2917649702280646e-05,
+      "loss": 0.5592,
+      "step": 8060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.2889562970452756e-05,
+      "loss": 0.5711,
+      "step": 8070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.286147623862488e-05,
+      "loss": 0.4493,
+      "step": 8080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.283338950679699e-05,
+      "loss": 0.1169,
+      "step": 8090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.2805302774969107e-05,
+      "loss": 0.2195,
+      "step": 8100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.277721604314122e-05,
+      "loss": 0.1202,
+      "step": 8110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.274912931131334e-05,
+      "loss": 0.0522,
+      "step": 8120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.272104257948545e-05,
+      "loss": 0.4026,
+      "step": 8130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.269295584765757e-05,
+      "loss": 0.3013,
+      "step": 8140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.2664869115829683e-05,
+      "loss": 0.139,
+      "step": 8150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.26367823840018e-05,
+      "loss": 0.059,
+      "step": 8160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.260869565217392e-05,
+      "loss": 0.3185,
+      "step": 8170
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.258060892034603e-05,
+      "loss": 0.3473,
+      "step": 8180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.255252218851815e-05,
+      "loss": 0.4267,
+      "step": 8190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.252443545669026e-05,
+      "loss": 0.5359,
+      "step": 8200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.249634872486238e-05,
+      "loss": 0.555,
+      "step": 8210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2468261993034494e-05,
+      "loss": 0.349,
+      "step": 8220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2440175261206604e-05,
+      "loss": 0.3167,
+      "step": 8230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.241208852937873e-05,
+      "loss": 0.0955,
+      "step": 8240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.238400179755084e-05,
+      "loss": 0.4953,
+      "step": 8250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2355915065722954e-05,
+      "loss": 0.1859,
+      "step": 8260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.232782833389507e-05,
+      "loss": 0.6907,
+      "step": 8270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.229974160206719e-05,
+      "loss": 0.2779,
+      "step": 8280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.22716548702393e-05,
+      "loss": 0.3102,
+      "step": 8290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2243568138411414e-05,
+      "loss": 0.513,
+      "step": 8300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.221548140658353e-05,
+      "loss": 0.2827,
+      "step": 8310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.218739467475564e-05,
+      "loss": 0.0513,
+      "step": 8320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2159307942927764e-05,
+      "loss": 0.2898,
+      "step": 8330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2131221211099874e-05,
+      "loss": 0.4058,
+      "step": 8340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2103134479272e-05,
+      "loss": 0.3355,
+      "step": 8350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.207504774744411e-05,
+      "loss": 0.0321,
+      "step": 8360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2046961015616225e-05,
+      "loss": 0.3748,
+      "step": 8370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.201887428378834e-05,
+      "loss": 0.3318,
+      "step": 8380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.199078755196046e-05,
+      "loss": 0.1727,
+      "step": 8390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.196270082013257e-05,
+      "loss": 0.1907,
+      "step": 8400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1934614088304685e-05,
+      "loss": 0.4921,
+      "step": 8410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.19065273564768e-05,
+      "loss": 0.2396,
+      "step": 8420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.187844062464892e-05,
+      "loss": 0.2757,
+      "step": 8430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1850353892821035e-05,
+      "loss": 0.9192,
+      "step": 8440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1822267160993145e-05,
+      "loss": 0.3968,
+      "step": 8450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.179418042916527e-05,
+      "loss": 0.593,
+      "step": 8460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.176609369733738e-05,
+      "loss": 0.7393,
+      "step": 8470
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1738006965509495e-05,
+      "loss": 0.5377,
+      "step": 8480
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.170992023368161e-05,
+      "loss": 0.4317,
+      "step": 8490
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.168183350185372e-05,
+      "loss": 0.3803,
+      "step": 8500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1653746770025845e-05,
+      "loss": 0.6205,
+      "step": 8510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1625660038197955e-05,
+      "loss": 0.2392,
+      "step": 8520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.159757330637007e-05,
+      "loss": 0.5311,
+      "step": 8530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.156948657454219e-05,
+      "loss": 0.5365,
+      "step": 8540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1541399842714306e-05,
+      "loss": 0.1621,
+      "step": 8550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1513313110886416e-05,
+      "loss": 0.7311,
+      "step": 8560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.148522637905853e-05,
+      "loss": 0.2579,
+      "step": 8570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.145713964723065e-05,
+      "loss": 0.1502,
+      "step": 8580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.1429052915402766e-05,
+      "loss": 0.3241,
+      "step": 8590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.140096618357488e-05,
+      "loss": 0.2354,
+      "step": 8600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.137287945174699e-05,
+      "loss": 0.3303,
+      "step": 8610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.1344792719919116e-05,
+      "loss": 0.1892,
+      "step": 8620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.1316705988091226e-05,
+      "loss": 0.2756,
+      "step": 8630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.128861925626334e-05,
+      "loss": 0.3473,
+      "step": 8640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.126053252443546e-05,
+      "loss": 0.1983,
+      "step": 8650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.1232445792607576e-05,
+      "loss": 0.2596,
+      "step": 8660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.1204359060779686e-05,
+      "loss": 0.2564,
+      "step": 8670
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.11762723289518e-05,
+      "loss": 0.2435,
+      "step": 8680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.114818559712392e-05,
+      "loss": 0.6432,
+      "step": 8690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.1120098865296036e-05,
+      "loss": 0.1312,
+      "step": 8700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.109201213346815e-05,
+      "loss": 0.3702,
+      "step": 8710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.106392540164026e-05,
+      "loss": 0.7259,
+      "step": 8720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.1035838669812387e-05,
+      "loss": 0.7203,
+      "step": 8730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.1007751937984497e-05,
+      "loss": 0.3819,
+      "step": 8740
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.097966520615661e-05,
+      "loss": 0.305,
+      "step": 8750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.095157847432873e-05,
+      "loss": 0.2647,
+      "step": 8760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.092349174250084e-05,
+      "loss": 0.2427,
+      "step": 8770
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.0895405010672963e-05,
+      "loss": 0.3887,
+      "step": 8780
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.0867318278845073e-05,
+      "loss": 0.1617,
+      "step": 8790
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.083923154701719e-05,
+      "loss": 0.3139,
+      "step": 8800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.081114481518931e-05,
+      "loss": 0.069,
+      "step": 8810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0783058083361424e-05,
+      "loss": 0.174,
+      "step": 8820
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0754971351533534e-05,
+      "loss": 0.042,
+      "step": 8830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.072688461970565e-05,
+      "loss": 0.5377,
+      "step": 8840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.069879788787777e-05,
+      "loss": 0.6893,
+      "step": 8850
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0670711156049884e-05,
+      "loss": 0.5524,
+      "step": 8860
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0642624424222e-05,
+      "loss": 0.8107,
+      "step": 8870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.061453769239411e-05,
+      "loss": 0.5393,
+      "step": 8880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0586450960566234e-05,
+      "loss": 0.3641,
+      "step": 8890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0558364228738344e-05,
+      "loss": 0.1179,
+      "step": 8900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.053027749691046e-05,
+      "loss": 0.2101,
+      "step": 8910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0502190765082577e-05,
+      "loss": 0.961,
+      "step": 8920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.047410403325469e-05,
+      "loss": 0.0858,
+      "step": 8930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.044601730142681e-05,
+      "loss": 0.3053,
+      "step": 8940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0417930569598924e-05,
+      "loss": 0.4237,
+      "step": 8950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0389843837771038e-05,
+      "loss": 0.1001,
+      "step": 8960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0361757105943154e-05,
+      "loss": 0.2652,
+      "step": 8970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.033367037411527e-05,
+      "loss": 0.0901,
+      "step": 8980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.0305583642287384e-05,
+      "loss": 0.5711,
+      "step": 8990
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.02774969104595e-05,
+      "loss": 0.338,
+      "step": 9000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0249410178631615e-05,
+      "loss": 0.3216,
+      "step": 9010
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0221323446803728e-05,
+      "loss": 0.1766,
+      "step": 9020
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0193236714975848e-05,
+      "loss": 0.4741,
+      "step": 9030
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.016514998314796e-05,
+      "loss": 0.1007,
+      "step": 9040
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.013706325132008e-05,
+      "loss": 0.3536,
+      "step": 9050
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0108976519492195e-05,
+      "loss": 0.5337,
+      "step": 9060
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0080889787664308e-05,
+      "loss": 0.1366,
+      "step": 9070
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0052803055836425e-05,
+      "loss": 0.116,
+      "step": 9080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.0024716324008538e-05,
+      "loss": 0.0603,
+      "step": 9090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.999662959218065e-05,
+      "loss": 0.2886,
+      "step": 9100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9968542860352772e-05,
+      "loss": 0.0425,
+      "step": 9110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9940456128524885e-05,
+      "loss": 0.4222,
+      "step": 9120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9912369396697005e-05,
+      "loss": 0.2005,
+      "step": 9130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.988428266486912e-05,
+      "loss": 0.1818,
+      "step": 9140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9856195933041232e-05,
+      "loss": 0.0235,
+      "step": 9150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.982810920121335e-05,
+      "loss": 0.0809,
+      "step": 9160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9800022469385462e-05,
+      "loss": 0.0143,
+      "step": 9170
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9771935737557575e-05,
+      "loss": 0.7728,
+      "step": 9180
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9743849005729696e-05,
+      "loss": 0.6792,
+      "step": 9190
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.971576227390181e-05,
+      "loss": 0.2666,
+      "step": 9200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.968767554207393e-05,
+      "loss": 0.0321,
+      "step": 9210
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9659588810246042e-05,
+      "loss": 0.7046,
+      "step": 9220
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9631502078418156e-05,
+      "loss": 0.6931,
+      "step": 9230
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9603415346590276e-05,
+      "loss": 0.523,
+      "step": 9240
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.957532861476239e-05,
+      "loss": 0.0774,
+      "step": 9250
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9547241882934503e-05,
+      "loss": 0.1332,
+      "step": 9260
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.951915515110662e-05,
+      "loss": 0.0691,
+      "step": 9270
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9491068419278733e-05,
+      "loss": 0.2895,
+      "step": 9280
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9462981687450846e-05,
+      "loss": 0.5303,
+      "step": 9290
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9434894955622966e-05,
+      "loss": 0.4615,
+      "step": 9300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.940680822379508e-05,
+      "loss": 0.2697,
+      "step": 9310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.93787214919672e-05,
+      "loss": 0.3039,
+      "step": 9320
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9350634760139313e-05,
+      "loss": 0.5948,
+      "step": 9330
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9322548028311426e-05,
+      "loss": 0.2371,
+      "step": 9340
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9294461296483543e-05,
+      "loss": 0.0077,
+      "step": 9350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9266374564655656e-05,
+      "loss": 0.1464,
+      "step": 9360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.923828783282777e-05,
+      "loss": 0.0127,
+      "step": 9370
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.921020110099989e-05,
+      "loss": 0.2673,
+      "step": 9380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9182114369172003e-05,
+      "loss": 0.0185,
+      "step": 9390
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9154027637344123e-05,
+      "loss": 0.5188,
+      "step": 9400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9125940905516237e-05,
+      "loss": 0.0235,
+      "step": 9410
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.909785417368835e-05,
+      "loss": 0.0399,
+      "step": 9420
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9069767441860467e-05,
+      "loss": 0.1907,
+      "step": 9430
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.904168071003258e-05,
+      "loss": 0.2062,
+      "step": 9440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9013593978204693e-05,
+      "loss": 0.0529,
+      "step": 9450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8985507246376814e-05,
+      "loss": 0.2058,
+      "step": 9460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8957420514548927e-05,
+      "loss": 0.1838,
+      "step": 9470
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8929333782721047e-05,
+      "loss": 0.2133,
+      "step": 9480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.890124705089316e-05,
+      "loss": 0.1268,
+      "step": 9490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8873160319065274e-05,
+      "loss": 0.3107,
+      "step": 9500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8845073587237394e-05,
+      "loss": 0.0589,
+      "step": 9510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8816986855409507e-05,
+      "loss": 0.0605,
+      "step": 9520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.878890012358162e-05,
+      "loss": 0.0839,
+      "step": 9530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8760813391753737e-05,
+      "loss": 0.4657,
+      "step": 9540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.873272665992585e-05,
+      "loss": 0.8741,
+      "step": 9550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.870463992809797e-05,
+      "loss": 0.0462,
+      "step": 9560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8676553196270084e-05,
+      "loss": 0.234,
+      "step": 9570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8648466464442197e-05,
+      "loss": 0.0592,
+      "step": 9580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.8620379732614318e-05,
+      "loss": 0.5951,
+      "step": 9590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.859229300078643e-05,
+      "loss": 0.2689,
+      "step": 9600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8564206268958544e-05,
+      "loss": 0.2143,
+      "step": 9610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.853611953713066e-05,
+      "loss": 0.4141,
+      "step": 9620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8508032805302774e-05,
+      "loss": 0.0451,
+      "step": 9630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8479946073474888e-05,
+      "loss": 0.0275,
+      "step": 9640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8451859341647008e-05,
+      "loss": 0.2775,
+      "step": 9650
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.842377260981912e-05,
+      "loss": 0.3787,
+      "step": 9660
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.839568587799124e-05,
+      "loss": 0.4379,
+      "step": 9670
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8367599146163355e-05,
+      "loss": 0.0244,
+      "step": 9680
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8339512414335468e-05,
+      "loss": 0.1208,
+      "step": 9690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8311425682507585e-05,
+      "loss": 0.0565,
+      "step": 9700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8283338950679698e-05,
+      "loss": 0.2893,
+      "step": 9710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.825525221885181e-05,
+      "loss": 0.2822,
+      "step": 9720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.822716548702393e-05,
+      "loss": 0.7578,
+      "step": 9730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8199078755196045e-05,
+      "loss": 0.317,
+      "step": 9740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8170992023368165e-05,
+      "loss": 0.5164,
+      "step": 9750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.814290529154028e-05,
+      "loss": 0.3778,
+      "step": 9760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8114818559712392e-05,
+      "loss": 0.5201,
+      "step": 9770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8086731827884512e-05,
+      "loss": 0.4249,
+      "step": 9780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.8058645096056625e-05,
+      "loss": 0.0947,
+      "step": 9790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.803055836422874e-05,
+      "loss": 0.1716,
+      "step": 9800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.8002471632400855e-05,
+      "loss": 0.2041,
+      "step": 9810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.797438490057297e-05,
+      "loss": 0.0367,
+      "step": 9820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.794629816874509e-05,
+      "loss": 0.2542,
+      "step": 9830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7918211436917202e-05,
+      "loss": 0.516,
+      "step": 9840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7890124705089316e-05,
+      "loss": 0.4607,
+      "step": 9850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7862037973261436e-05,
+      "loss": 0.143,
+      "step": 9860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.783395124143355e-05,
+      "loss": 0.1799,
+      "step": 9870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7805864509605662e-05,
+      "loss": 0.0652,
+      "step": 9880
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.3438,
+      "step": 9890
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7749691045949892e-05,
+      "loss": 0.3123,
+      "step": 9900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7721604314122013e-05,
+      "loss": 0.3548,
+      "step": 9910
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7693517582294126e-05,
+      "loss": 0.3546,
+      "step": 9920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.766543085046624e-05,
+      "loss": 0.0425,
+      "step": 9930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.763734411863836e-05,
+      "loss": 0.0113,
+      "step": 9940
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7609257386810473e-05,
+      "loss": 0.0175,
+      "step": 9950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7581170654982586e-05,
+      "loss": 0.2991,
+      "step": 9960
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7553083923154703e-05,
+      "loss": 0.2057,
+      "step": 9970
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.7524997191326816e-05,
+      "loss": 0.0535,
+      "step": 9980
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7496910459498933e-05,
+      "loss": 0.2534,
+      "step": 9990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.746882372767105e-05,
+      "loss": 0.2388,
+      "step": 10000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7440736995843163e-05,
+      "loss": 0.4119,
+      "step": 10010
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7412650264015283e-05,
+      "loss": 0.0783,
+      "step": 10020
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7384563532187397e-05,
+      "loss": 0.3031,
+      "step": 10030
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.735647680035951e-05,
+      "loss": 0.4968,
+      "step": 10040
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.732839006853163e-05,
+      "loss": 0.2388,
+      "step": 10050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7300303336703743e-05,
+      "loss": 0.1887,
+      "step": 10060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7272216604875857e-05,
+      "loss": 0.2498,
+      "step": 10070
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7244129873047973e-05,
+      "loss": 0.277,
+      "step": 10080
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7216043141220087e-05,
+      "loss": 0.7838,
+      "step": 10090
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7187956409392207e-05,
+      "loss": 0.6741,
+      "step": 10100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.715986967756432e-05,
+      "loss": 0.6697,
+      "step": 10110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7131782945736434e-05,
+      "loss": 0.0807,
+      "step": 10120
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7103696213908554e-05,
+      "loss": 0.1239,
+      "step": 10130
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7075609482080667e-05,
+      "loss": 0.3427,
+      "step": 10140
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.704752275025278e-05,
+      "loss": 0.3546,
+      "step": 10150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.7019436018424897e-05,
+      "loss": 0.5538,
+      "step": 10160
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.699134928659701e-05,
+      "loss": 0.346,
+      "step": 10170
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.696326255476913e-05,
+      "loss": 0.668,
+      "step": 10180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6935175822941244e-05,
+      "loss": 0.3388,
+      "step": 10190
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6907089091113357e-05,
+      "loss": 0.2091,
+      "step": 10200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6879002359285477e-05,
+      "loss": 0.0324,
+      "step": 10210
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.685091562745759e-05,
+      "loss": 0.0084,
+      "step": 10220
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6822828895629704e-05,
+      "loss": 0.1367,
+      "step": 10230
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.679474216380182e-05,
+      "loss": 0.1677,
+      "step": 10240
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6766655431973938e-05,
+      "loss": 0.0417,
+      "step": 10250
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6738568700146054e-05,
+      "loss": 0.4514,
+      "step": 10260
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6710481968318168e-05,
+      "loss": 0.0162,
+      "step": 10270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.668239523649028e-05,
+      "loss": 0.2189,
+      "step": 10280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.66543085046624e-05,
+      "loss": 0.5889,
+      "step": 10290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6626221772834515e-05,
+      "loss": 0.4427,
+      "step": 10300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6598135041006628e-05,
+      "loss": 0.0301,
+      "step": 10310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6570048309178748e-05,
+      "loss": 0.2991,
+      "step": 10320
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.654196157735086e-05,
+      "loss": 0.2227,
+      "step": 10330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6513874845522975e-05,
+      "loss": 0.5861,
+      "step": 10340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.648578811369509e-05,
+      "loss": 0.3113,
+      "step": 10350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6457701381867205e-05,
+      "loss": 0.4378,
+      "step": 10360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6429614650039325e-05,
+      "loss": 0.0686,
+      "step": 10370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6401527918211438e-05,
+      "loss": 0.4675,
+      "step": 10380
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.637344118638355e-05,
+      "loss": 0.3048,
+      "step": 10390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6345354454555672e-05,
+      "loss": 0.0628,
+      "step": 10400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6317267722727785e-05,
+      "loss": 0.0292,
+      "step": 10410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.62891809908999e-05,
+      "loss": 0.2756,
+      "step": 10420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6261094259072015e-05,
+      "loss": 0.2843,
+      "step": 10430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.623300752724413e-05,
+      "loss": 0.2695,
+      "step": 10440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.620492079541625e-05,
+      "loss": 0.3109,
+      "step": 10450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6176834063588362e-05,
+      "loss": 0.7247,
+      "step": 10460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6148747331760475e-05,
+      "loss": 0.2102,
+      "step": 10470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6120660599932596e-05,
+      "loss": 0.2235,
+      "step": 10480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.609257386810471e-05,
+      "loss": 0.0282,
+      "step": 10490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6064487136276822e-05,
+      "loss": 0.329,
+      "step": 10500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6036400404448942e-05,
+      "loss": 0.0474,
+      "step": 10510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6008313672621056e-05,
+      "loss": 0.2207,
+      "step": 10520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5980226940793172e-05,
+      "loss": 0.0838,
+      "step": 10530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5952140208965286e-05,
+      "loss": 0.3554,
+      "step": 10540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.59240534771374e-05,
+      "loss": 0.0688,
+      "step": 10550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.589596674530952e-05,
+      "loss": 0.0981,
+      "step": 10560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5867880013481633e-05,
+      "loss": 0.0557,
+      "step": 10570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5839793281653746e-05,
+      "loss": 0.4021,
+      "step": 10580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5811706549825866e-05,
+      "loss": 0.3132,
+      "step": 10590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.578361981799798e-05,
+      "loss": 0.2272,
+      "step": 10600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5755533086170096e-05,
+      "loss": 0.4241,
+      "step": 10610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.572744635434221e-05,
+      "loss": 0.5019,
+      "step": 10620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5699359622514323e-05,
+      "loss": 0.1336,
+      "step": 10630
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5671272890686443e-05,
+      "loss": 0.4276,
+      "step": 10640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5643186158858556e-05,
+      "loss": 0.1032,
+      "step": 10650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.561509942703067e-05,
+      "loss": 0.2629,
+      "step": 10660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.558701269520279e-05,
+      "loss": 0.2075,
+      "step": 10670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5558925963374903e-05,
+      "loss": 0.0489,
+      "step": 10680
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5530839231547017e-05,
+      "loss": 0.0248,
+      "step": 10690
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5502752499719133e-05,
+      "loss": 0.0313,
+      "step": 10700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5474665767891247e-05,
+      "loss": 0.1751,
+      "step": 10710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5446579036063367e-05,
+      "loss": 0.2057,
+      "step": 10720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.541849230423548e-05,
+      "loss": 0.1532,
+      "step": 10730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5390405572407593e-05,
+      "loss": 0.2646,
+      "step": 10740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5362318840579714e-05,
+      "loss": 0.0534,
+      "step": 10750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5334232108751827e-05,
+      "loss": 0.0374,
+      "step": 10760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.530614537692394e-05,
+      "loss": 0.0214,
+      "step": 10770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.527805864509606e-05,
+      "loss": 0.1003,
+      "step": 10780
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5249971913268174e-05,
+      "loss": 0.0367,
+      "step": 10790
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.522188518144029e-05,
+      "loss": 0.2377,
+      "step": 10800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5193798449612404e-05,
+      "loss": 0.3069,
+      "step": 10810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5165711717784517e-05,
+      "loss": 0.3599,
+      "step": 10820
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5137624985956637e-05,
+      "loss": 0.0535,
+      "step": 10830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.510953825412875e-05,
+      "loss": 0.3965,
+      "step": 10840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5081451522300864e-05,
+      "loss": 0.0132,
+      "step": 10850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5053364790472984e-05,
+      "loss": 0.4889,
+      "step": 10860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5025278058645097e-05,
+      "loss": 0.1869,
+      "step": 10870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4997191326817214e-05,
+      "loss": 0.5947,
+      "step": 10880
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4969104594989328e-05,
+      "loss": 0.0595,
+      "step": 10890
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4941017863161444e-05,
+      "loss": 0.2607,
+      "step": 10900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4912931131333558e-05,
+      "loss": 0.6526,
+      "step": 10910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4884844399505674e-05,
+      "loss": 0.0168,
+      "step": 10920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.485675766767779e-05,
+      "loss": 0.0104,
+      "step": 10930
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4828670935849908e-05,
+      "loss": 0.209,
+      "step": 10940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.480058420402202e-05,
+      "loss": 0.239,
+      "step": 10950
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.4772497472194138e-05,
+      "loss": 0.2979,
+      "step": 10960
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.474441074036625e-05,
+      "loss": 0.4175,
+      "step": 10970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4716324008538368e-05,
+      "loss": 0.0292,
+      "step": 10980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.468823727671048e-05,
+      "loss": 0.5338,
+      "step": 10990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4660150544882598e-05,
+      "loss": 0.179,
+      "step": 11000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4632063813054715e-05,
+      "loss": 0.1873,
+      "step": 11010
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4603977081226828e-05,
+      "loss": 0.0336,
+      "step": 11020
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4575890349398945e-05,
+      "loss": 0.1203,
+      "step": 11030
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4547803617571062e-05,
+      "loss": 0.4073,
+      "step": 11040
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.451971688574318e-05,
+      "loss": 0.1588,
+      "step": 11050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4491630153915292e-05,
+      "loss": 0.0213,
+      "step": 11060
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4463543422087405e-05,
+      "loss": 0.0534,
+      "step": 11070
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4435456690259522e-05,
+      "loss": 0.4908,
+      "step": 11080
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.440736995843164e-05,
+      "loss": 1.0586,
+      "step": 11090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4379283226603752e-05,
+      "loss": 0.4335,
+      "step": 11100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.435119649477587e-05,
+      "loss": 0.5386,
+      "step": 11110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4323109762947985e-05,
+      "loss": 0.0656,
+      "step": 11120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4295023031120102e-05,
+      "loss": 0.0275,
+      "step": 11130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4266936299292216e-05,
+      "loss": 1.0354,
+      "step": 11140
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4238849567464332e-05,
+      "loss": 0.1257,
+      "step": 11150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4210762835636446e-05,
+      "loss": 0.0161,
+      "step": 11160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.4182676103808562e-05,
+      "loss": 0.1694,
+      "step": 11170
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.4154589371980676e-05,
+      "loss": 0.0753,
+      "step": 11180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.4126502640152792e-05,
+      "loss": 0.1351,
+      "step": 11190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.409841590832491e-05,
+      "loss": 0.0465,
+      "step": 11200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.4070329176497026e-05,
+      "loss": 0.1268,
+      "step": 11210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.404224244466914e-05,
+      "loss": 0.4352,
+      "step": 11220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.4014155712841256e-05,
+      "loss": 0.5901,
+      "step": 11230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.398606898101337e-05,
+      "loss": 0.1323,
+      "step": 11240
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3957982249185486e-05,
+      "loss": 0.6654,
+      "step": 11250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.39298955173576e-05,
+      "loss": 0.7204,
+      "step": 11260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3901808785529716e-05,
+      "loss": 0.2244,
+      "step": 11270
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3873722053701833e-05,
+      "loss": 0.0052,
+      "step": 11280
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.384563532187395e-05,
+      "loss": 0.0031,
+      "step": 11290
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3817548590046063e-05,
+      "loss": 0.0163,
+      "step": 11300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.378946185821818e-05,
+      "loss": 0.4305,
+      "step": 11310
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3761375126390297e-05,
+      "loss": 0.0032,
+      "step": 11320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.373328839456241e-05,
+      "loss": 0.3445,
+      "step": 11330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3705201662734523e-05,
+      "loss": 0.2879,
+      "step": 11340
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.367711493090664e-05,
+      "loss": 0.5253,
+      "step": 11350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.3649028199078757e-05,
+      "loss": 0.4869,
+      "step": 11360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.362094146725087e-05,
+      "loss": 0.4635,
+      "step": 11370
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3592854735422987e-05,
+      "loss": 0.2209,
+      "step": 11380
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3564768003595104e-05,
+      "loss": 0.2095,
+      "step": 11390
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.353668127176722e-05,
+      "loss": 0.3037,
+      "step": 11400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3508594539939334e-05,
+      "loss": 0.018,
+      "step": 11410
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.348050780811145e-05,
+      "loss": 0.1462,
+      "step": 11420
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3452421076283564e-05,
+      "loss": 0.1968,
+      "step": 11430
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.342433434445568e-05,
+      "loss": 0.0504,
+      "step": 11440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3396247612627794e-05,
+      "loss": 0.0378,
+      "step": 11450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.336816088079991e-05,
+      "loss": 0.0175,
+      "step": 11460
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3340074148972027e-05,
+      "loss": 0.4443,
+      "step": 11470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3311987417144144e-05,
+      "loss": 0.2672,
+      "step": 11480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3283900685316257e-05,
+      "loss": 0.5056,
+      "step": 11490
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3255813953488374e-05,
+      "loss": 0.1364,
+      "step": 11500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3227727221660487e-05,
+      "loss": 0.5535,
+      "step": 11510
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3199640489832604e-05,
+      "loss": 0.3297,
+      "step": 11520
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3171553758004718e-05,
+      "loss": 0.0175,
+      "step": 11530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3143467026176834e-05,
+      "loss": 0.0302,
+      "step": 11540
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.311538029434895e-05,
+      "loss": 0.5749,
+      "step": 11550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.3087293562521068e-05,
+      "loss": 0.3222,
+      "step": 11560
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.305920683069318e-05,
+      "loss": 0.4816,
+      "step": 11570
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3031120098865298e-05,
+      "loss": 0.4296,
+      "step": 11580
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.3003033367037415e-05,
+      "loss": 0.1802,
+      "step": 11590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2974946635209528e-05,
+      "loss": 0.4238,
+      "step": 11600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.294685990338164e-05,
+      "loss": 0.0186,
+      "step": 11610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2918773171553758e-05,
+      "loss": 0.1639,
+      "step": 11620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2890686439725875e-05,
+      "loss": 0.5457,
+      "step": 11630
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.286259970789799e-05,
+      "loss": 0.1327,
+      "step": 11640
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2834512976070105e-05,
+      "loss": 0.6064,
+      "step": 11650
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.280642624424222e-05,
+      "loss": 0.2175,
+      "step": 11660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2778339512414338e-05,
+      "loss": 0.1389,
+      "step": 11670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.275025278058645e-05,
+      "loss": 0.5683,
+      "step": 11680
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.272216604875857e-05,
+      "loss": 0.1721,
+      "step": 11690
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2694079316930682e-05,
+      "loss": 0.3705,
+      "step": 11700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.26659925851028e-05,
+      "loss": 0.025,
+      "step": 11710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2637905853274912e-05,
+      "loss": 0.2671,
+      "step": 11720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.260981912144703e-05,
+      "loss": 0.1183,
+      "step": 11730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2581732389619145e-05,
+      "loss": 0.0099,
+      "step": 11740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2553645657791262e-05,
+      "loss": 0.0063,
+      "step": 11750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2525558925963375e-05,
+      "loss": 0.1743,
+      "step": 11760
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2497472194135492e-05,
+      "loss": 0.0061,
+      "step": 11770
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.246938546230761e-05,
+      "loss": 0.0259,
+      "step": 11780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2441298730479722e-05,
+      "loss": 0.1699,
+      "step": 11790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2413211998651836e-05,
+      "loss": 0.0214,
+      "step": 11800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2385125266823952e-05,
+      "loss": 0.2427,
+      "step": 11810
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.235703853499607e-05,
+      "loss": 0.0661,
+      "step": 11820
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2328951803168186e-05,
+      "loss": 0.3282,
+      "step": 11830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.23008650713403e-05,
+      "loss": 0.1208,
+      "step": 11840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2272778339512416e-05,
+      "loss": 0.1659,
+      "step": 11850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2244691607684533e-05,
+      "loss": 0.1014,
+      "step": 11860
+    },
+    {
+      "epoch": 2.2,
+      "eval_accuracy": 0.7415817313894981,
+      "eval_loss": 1.183919072151184,
+      "eval_runtime": 1229.7188,
+      "eval_samples_per_second": 6.303,
+      "eval_steps_per_second": 3.152,
+      "step": 11868
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2216604875856646e-05,
+      "loss": 0.1397,
+      "step": 11870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.218851814402876e-05,
+      "loss": 0.3042,
+      "step": 11880
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2160431412200876e-05,
+      "loss": 0.0711,
+      "step": 11890
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2132344680372993e-05,
+      "loss": 0.2664,
+      "step": 11900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.210425794854511e-05,
+      "loss": 0.0215,
+      "step": 11910
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2076171216717223e-05,
+      "loss": 0.1595,
+      "step": 11920
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.204808448488934e-05,
+      "loss": 0.0028,
+      "step": 11930
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2019997753061456e-05,
+      "loss": 0.0895,
+      "step": 11940
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1991911021233573e-05,
+      "loss": 0.0299,
+      "step": 11950
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1963824289405686e-05,
+      "loss": 0.0054,
+      "step": 11960
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.19357375575778e-05,
+      "loss": 0.0594,
+      "step": 11970
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1907650825749917e-05,
+      "loss": 0.0339,
+      "step": 11980
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1879564093922033e-05,
+      "loss": 0.2771,
+      "step": 11990
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1851477362094147e-05,
+      "loss": 0.0821,
+      "step": 12000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1823390630266263e-05,
+      "loss": 0.3412,
+      "step": 12010
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.179530389843838e-05,
+      "loss": 0.0235,
+      "step": 12020
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1767217166610493e-05,
+      "loss": 0.2042,
+      "step": 12030
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.173913043478261e-05,
+      "loss": 0.105,
+      "step": 12040
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1711043702954727e-05,
+      "loss": 0.0753,
+      "step": 12050
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.168295697112684e-05,
+      "loss": 0.0044,
+      "step": 12060
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1654870239298954e-05,
+      "loss": 0.1457,
+      "step": 12070
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.162678350747107e-05,
+      "loss": 0.0067,
+      "step": 12080
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1598696775643187e-05,
+      "loss": 0.0757,
+      "step": 12090
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1570610043815304e-05,
+      "loss": 0.0042,
+      "step": 12100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1542523311987417e-05,
+      "loss": 0.2074,
+      "step": 12110
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1514436580159534e-05,
+      "loss": 0.3848,
+      "step": 12120
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.148634984833165e-05,
+      "loss": 0.3169,
+      "step": 12130
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1458263116503764e-05,
+      "loss": 0.047,
+      "step": 12140
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.143017638467588e-05,
+      "loss": 0.1108,
+      "step": 12150
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.1402089652847994e-05,
+      "loss": 0.1339,
+      "step": 12160
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.137400292102011e-05,
+      "loss": 0.2602,
+      "step": 12170
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1345916189192228e-05,
+      "loss": 0.0239,
+      "step": 12180
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.131782945736434e-05,
+      "loss": 0.0014,
+      "step": 12190
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1289742725536458e-05,
+      "loss": 0.2475,
+      "step": 12200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1261655993708574e-05,
+      "loss": 0.031,
+      "step": 12210
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.123356926188069e-05,
+      "loss": 0.3204,
+      "step": 12220
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1205482530052804e-05,
+      "loss": 0.014,
+      "step": 12230
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1177395798224918e-05,
+      "loss": 0.0065,
+      "step": 12240
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1149309066397035e-05,
+      "loss": 0.163,
+      "step": 12250
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.112122233456915e-05,
+      "loss": 0.421,
+      "step": 12260
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1093135602741265e-05,
+      "loss": 0.1281,
+      "step": 12270
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.106504887091338e-05,
+      "loss": 0.0061,
+      "step": 12280
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1036962139085498e-05,
+      "loss": 0.0084,
+      "step": 12290
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.1008875407257615e-05,
+      "loss": 0.1375,
+      "step": 12300
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0980788675429728e-05,
+      "loss": 0.2595,
+      "step": 12310
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0952701943601845e-05,
+      "loss": 0.0139,
+      "step": 12320
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.092461521177396e-05,
+      "loss": 0.6045,
+      "step": 12330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.089652847994607e-05,
+      "loss": 0.2658,
+      "step": 12340
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.086844174811819e-05,
+      "loss": 0.0061,
+      "step": 12350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.0840355016290305e-05,
+      "loss": 0.0937,
+      "step": 12360
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0812268284462422e-05,
+      "loss": 0.203,
+      "step": 12370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0784181552634535e-05,
+      "loss": 0.0081,
+      "step": 12380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0756094820806652e-05,
+      "loss": 0.0133,
+      "step": 12390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.072800808897877e-05,
+      "loss": 0.266,
+      "step": 12400
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0699921357150885e-05,
+      "loss": 0.1842,
+      "step": 12410
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0671834625323e-05,
+      "loss": 0.0131,
+      "step": 12420
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0643747893495112e-05,
+      "loss": 0.0497,
+      "step": 12430
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.061566116166723e-05,
+      "loss": 0.32,
+      "step": 12440
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0587574429839346e-05,
+      "loss": 0.1859,
+      "step": 12450
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.055948769801146e-05,
+      "loss": 0.0556,
+      "step": 12460
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0531400966183576e-05,
+      "loss": 0.0054,
+      "step": 12470
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0503314234355692e-05,
+      "loss": 0.156,
+      "step": 12480
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.047522750252781e-05,
+      "loss": 0.2781,
+      "step": 12490
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0447140770699923e-05,
+      "loss": 0.0699,
+      "step": 12500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0419054038872036e-05,
+      "loss": 0.3054,
+      "step": 12510
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0390967307044153e-05,
+      "loss": 0.0195,
+      "step": 12520
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.036288057521627e-05,
+      "loss": 0.0376,
+      "step": 12530
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0334793843388383e-05,
+      "loss": 0.2314,
+      "step": 12540
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.03067071115605e-05,
+      "loss": 0.0158,
+      "step": 12550
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.0278620379732616e-05,
+      "loss": 0.0147,
+      "step": 12560
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0250533647904733e-05,
+      "loss": 0.1045,
+      "step": 12570
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0222446916076846e-05,
+      "loss": 0.0036,
+      "step": 12580
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0194360184248963e-05,
+      "loss": 0.1238,
+      "step": 12590
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0166273452421076e-05,
+      "loss": 0.2414,
+      "step": 12600
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0138186720593193e-05,
+      "loss": 0.0428,
+      "step": 12610
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0110099988765306e-05,
+      "loss": 0.2941,
+      "step": 12620
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0082013256937423e-05,
+      "loss": 0.0073,
+      "step": 12630
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.005392652510954e-05,
+      "loss": 0.4548,
+      "step": 12640
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.0025839793281657e-05,
+      "loss": 0.0064,
+      "step": 12650
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.999775306145377e-05,
+      "loss": 0.0257,
+      "step": 12660
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9969666329625887e-05,
+      "loss": 0.0279,
+      "step": 12670
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9941579597798004e-05,
+      "loss": 0.0073,
+      "step": 12680
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9913492865970117e-05,
+      "loss": 0.0072,
+      "step": 12690
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.988540613414223e-05,
+      "loss": 0.0039,
+      "step": 12700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9857319402314347e-05,
+      "loss": 0.0015,
+      "step": 12710
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9829232670486464e-05,
+      "loss": 0.3217,
+      "step": 12720
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9801145938658577e-05,
+      "loss": 0.155,
+      "step": 12730
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9773059206830694e-05,
+      "loss": 0.1053,
+      "step": 12740
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.974497247500281e-05,
+      "loss": 0.0028,
+      "step": 12750
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9716885743174927e-05,
+      "loss": 0.3649,
+      "step": 12760
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.968879901134704e-05,
+      "loss": 0.0039,
+      "step": 12770
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9660712279519154e-05,
+      "loss": 0.0079,
+      "step": 12780
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.963262554769127e-05,
+      "loss": 0.3682,
+      "step": 12790
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9604538815863387e-05,
+      "loss": 0.0483,
+      "step": 12800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.95764520840355e-05,
+      "loss": 0.0643,
+      "step": 12810
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9548365352207618e-05,
+      "loss": 0.0355,
+      "step": 12820
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9520278620379734e-05,
+      "loss": 0.0011,
+      "step": 12830
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.949219188855185e-05,
+      "loss": 0.0486,
+      "step": 12840
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9464105156723964e-05,
+      "loss": 0.0879,
+      "step": 12850
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.943601842489608e-05,
+      "loss": 0.3343,
+      "step": 12860
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9407931693068194e-05,
+      "loss": 0.0527,
+      "step": 12870
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.937984496124031e-05,
+      "loss": 0.1269,
+      "step": 12880
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9351758229412425e-05,
+      "loss": 0.1594,
+      "step": 12890
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.932367149758454e-05,
+      "loss": 0.0467,
+      "step": 12900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9295584765756658e-05,
+      "loss": 0.0049,
+      "step": 12910
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9267498033928775e-05,
+      "loss": 0.0033,
+      "step": 12920
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9239411302100888e-05,
+      "loss": 0.4019,
+      "step": 12930
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9211324570273005e-05,
+      "loss": 0.3228,
+      "step": 12940
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.918323783844512e-05,
+      "loss": 0.0929,
+      "step": 12950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9155151106617235e-05,
+      "loss": 0.0014,
+      "step": 12960
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9127064374789348e-05,
+      "loss": 0.0129,
+      "step": 12970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9098977642961465e-05,
+      "loss": 0.1457,
+      "step": 12980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9070890911133582e-05,
+      "loss": 0.0032,
+      "step": 12990
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.90428041793057e-05,
+      "loss": 0.0178,
+      "step": 13000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9014717447477812e-05,
+      "loss": 0.0035,
+      "step": 13010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.898663071564993e-05,
+      "loss": 0.1054,
+      "step": 13020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8958543983822045e-05,
+      "loss": 0.0014,
+      "step": 13030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.893045725199416e-05,
+      "loss": 0.0011,
+      "step": 13040
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8902370520166275e-05,
+      "loss": 0.1039,
+      "step": 13050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.887428378833839e-05,
+      "loss": 0.0136,
+      "step": 13060
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8846197056510505e-05,
+      "loss": 0.3307,
+      "step": 13070
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.881811032468262e-05,
+      "loss": 0.1581,
+      "step": 13080
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8790023592854736e-05,
+      "loss": 0.0079,
+      "step": 13090
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8761936861026852e-05,
+      "loss": 0.0711,
+      "step": 13100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.873385012919897e-05,
+      "loss": 0.1181,
+      "step": 13110
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8705763397371082e-05,
+      "loss": 0.2159,
+      "step": 13120
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.86776766655432e-05,
+      "loss": 0.0072,
+      "step": 13130
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8649589933715312e-05,
+      "loss": 0.0463,
+      "step": 13140
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.862150320188743e-05,
+      "loss": 0.4762,
+      "step": 13150
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8593416470059543e-05,
+      "loss": 0.0021,
+      "step": 13160
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.856532973823166e-05,
+      "loss": 0.0601,
+      "step": 13170
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8537243006403776e-05,
+      "loss": 0.0576,
+      "step": 13180
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8509156274575893e-05,
+      "loss": 0.3033,
+      "step": 13190
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8481069542748006e-05,
+      "loss": 0.0026,
+      "step": 13200
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8452982810920123e-05,
+      "loss": 0.0012,
+      "step": 13210
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.842489607909224e-05,
+      "loss": 0.0394,
+      "step": 13220
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8396809347264353e-05,
+      "loss": 0.1193,
+      "step": 13230
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8368722615436466e-05,
+      "loss": 0.0015,
+      "step": 13240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8340635883608583e-05,
+      "loss": 0.1269,
+      "step": 13250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.83125491517807e-05,
+      "loss": 0.2704,
+      "step": 13260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8284462419952817e-05,
+      "loss": 0.1427,
+      "step": 13270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.825637568812493e-05,
+      "loss": 0.0409,
+      "step": 13280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8228288956297047e-05,
+      "loss": 0.0131,
+      "step": 13290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8200202224469163e-05,
+      "loss": 0.201,
+      "step": 13300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.817211549264128e-05,
+      "loss": 0.0071,
+      "step": 13310
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8144028760813393e-05,
+      "loss": 0.1367,
+      "step": 13320
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8115942028985507e-05,
+      "loss": 0.401,
+      "step": 13330
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8087855297157624e-05,
+      "loss": 0.2915,
+      "step": 13340
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8059768565329737e-05,
+      "loss": 0.0051,
+      "step": 13350
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8031681833501854e-05,
+      "loss": 0.0012,
+      "step": 13360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.800359510167397e-05,
+      "loss": 0.2171,
+      "step": 13370
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7975508369846087e-05,
+      "loss": 0.0013,
+      "step": 13380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.79474216380182e-05,
+      "loss": 0.0073,
+      "step": 13390
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7919334906190317e-05,
+      "loss": 0.117,
+      "step": 13400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.789124817436243e-05,
+      "loss": 0.006,
+      "step": 13410
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7863161442534547e-05,
+      "loss": 0.0066,
+      "step": 13420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.783507471070666e-05,
+      "loss": 0.4244,
+      "step": 13430
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7806987978878777e-05,
+      "loss": 0.3481,
+      "step": 13440
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7778901247050894e-05,
+      "loss": 0.0475,
+      "step": 13450
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.775081451522301e-05,
+      "loss": 0.2287,
+      "step": 13460
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7722727783395124e-05,
+      "loss": 0.1568,
+      "step": 13470
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.769464105156724e-05,
+      "loss": 0.2766,
+      "step": 13480
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7666554319739358e-05,
+      "loss": 0.1093,
+      "step": 13490
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.763846758791147e-05,
+      "loss": 0.3349,
+      "step": 13500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7610380856083584e-05,
+      "loss": 0.095,
+      "step": 13510
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.75822941242557e-05,
+      "loss": 0.2097,
+      "step": 13520
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7554207392427818e-05,
+      "loss": 0.1635,
+      "step": 13530
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7526120660599935e-05,
+      "loss": 0.0015,
+      "step": 13540
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7498033928772048e-05,
+      "loss": 0.004,
+      "step": 13550
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7469947196944165e-05,
+      "loss": 0.0133,
+      "step": 13560
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.744186046511628e-05,
+      "loss": 0.0239,
+      "step": 13570
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7413773733288398e-05,
+      "loss": 0.0026,
+      "step": 13580
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.738568700146051e-05,
+      "loss": 0.005,
+      "step": 13590
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7357600269632625e-05,
+      "loss": 0.0552,
+      "step": 13600
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.732951353780474e-05,
+      "loss": 0.1046,
+      "step": 13610
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.730142680597686e-05,
+      "loss": 0.1957,
+      "step": 13620
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.727334007414897e-05,
+      "loss": 0.0104,
+      "step": 13630
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.724525334232109e-05,
+      "loss": 0.2135,
+      "step": 13640
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7217166610493205e-05,
+      "loss": 0.0881,
+      "step": 13650
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7189079878665322e-05,
+      "loss": 0.0182,
+      "step": 13660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7160993146837435e-05,
+      "loss": 0.0086,
+      "step": 13670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7132906415009552e-05,
+      "loss": 0.0056,
+      "step": 13680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7104819683181665e-05,
+      "loss": 0.0958,
+      "step": 13690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.707673295135378e-05,
+      "loss": 0.1603,
+      "step": 13700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7048646219525895e-05,
+      "loss": 0.0141,
+      "step": 13710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7020559487698012e-05,
+      "loss": 0.0015,
+      "step": 13720
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.699247275587013e-05,
+      "loss": 0.0879,
+      "step": 13730
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6964386024042242e-05,
+      "loss": 0.0021,
+      "step": 13740
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.693629929221436e-05,
+      "loss": 0.7012,
+      "step": 13750
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6908212560386476e-05,
+      "loss": 0.018,
+      "step": 13760
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.688012582855859e-05,
+      "loss": 0.0032,
+      "step": 13770
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6852039096730702e-05,
+      "loss": 0.2655,
+      "step": 13780
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.682395236490282e-05,
+      "loss": 0.4603,
+      "step": 13790
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6795865633074936e-05,
+      "loss": 0.2564,
+      "step": 13800
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6767778901247053e-05,
+      "loss": 0.2806,
+      "step": 13810
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6739692169419166e-05,
+      "loss": 0.0075,
+      "step": 13820
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6711605437591283e-05,
+      "loss": 0.0365,
+      "step": 13830
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.66835187057634e-05,
+      "loss": 0.1768,
+      "step": 13840
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6655431973935516e-05,
+      "loss": 0.0103,
+      "step": 13850
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.662734524210763e-05,
+      "loss": 0.013,
+      "step": 13860
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6599258510279743e-05,
+      "loss": 0.6491,
+      "step": 13870
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.657117177845186e-05,
+      "loss": 0.0407,
+      "step": 13880
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6543085046623976e-05,
+      "loss": 0.26,
+      "step": 13890
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.651499831479609e-05,
+      "loss": 0.1246,
+      "step": 13900
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6486911582968206e-05,
+      "loss": 0.2129,
+      "step": 13910
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6458824851140323e-05,
+      "loss": 0.1321,
+      "step": 13920
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.643073811931244e-05,
+      "loss": 0.1057,
+      "step": 13930
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6402651387484553e-05,
+      "loss": 0.3116,
+      "step": 13940
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.637456465565667e-05,
+      "loss": 0.016,
+      "step": 13950
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6346477923828783e-05,
+      "loss": 0.2803,
+      "step": 13960
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.63183911920009e-05,
+      "loss": 0.2986,
+      "step": 13970
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6290304460173013e-05,
+      "loss": 0.0977,
+      "step": 13980
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.626221772834513e-05,
+      "loss": 0.0037,
+      "step": 13990
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6234130996517247e-05,
+      "loss": 0.0074,
+      "step": 14000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6206044264689364e-05,
+      "loss": 0.0032,
+      "step": 14010
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6177957532861477e-05,
+      "loss": 0.6385,
+      "step": 14020
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6149870801033594e-05,
+      "loss": 0.2933,
+      "step": 14030
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6121784069205707e-05,
+      "loss": 0.0313,
+      "step": 14040
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.609369733737782e-05,
+      "loss": 0.0014,
+      "step": 14050
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6065610605549937e-05,
+      "loss": 0.004,
+      "step": 14060
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6037523873722054e-05,
+      "loss": 0.1043,
+      "step": 14070
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.600943714189417e-05,
+      "loss": 0.0861,
+      "step": 14080
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5981350410066284e-05,
+      "loss": 0.0244,
+      "step": 14090
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.59532636782384e-05,
+      "loss": 0.0008,
+      "step": 14100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5925176946410518e-05,
+      "loss": 0.3805,
+      "step": 14110
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5897090214582634e-05,
+      "loss": 0.0017,
+      "step": 14120
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5869003482754748e-05,
+      "loss": 0.004,
+      "step": 14130
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.584091675092686e-05,
+      "loss": 0.3629,
+      "step": 14140
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5812830019098978e-05,
+      "loss": 0.0017,
+      "step": 14150
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5784743287271094e-05,
+      "loss": 0.0015,
+      "step": 14160
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5756656555443208e-05,
+      "loss": 0.0074,
+      "step": 14170
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5728569823615325e-05,
+      "loss": 0.6303,
+      "step": 14180
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.570048309178744e-05,
+      "loss": 0.0148,
+      "step": 14190
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5672396359959558e-05,
+      "loss": 0.3778,
+      "step": 14200
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.564430962813167e-05,
+      "loss": 0.0082,
+      "step": 14210
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5616222896303788e-05,
+      "loss": 0.0073,
+      "step": 14220
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.55881361644759e-05,
+      "loss": 0.001,
+      "step": 14230
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5560049432648018e-05,
+      "loss": 0.1529,
+      "step": 14240
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.553196270082013e-05,
+      "loss": 0.0027,
+      "step": 14250
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5503875968992248e-05,
+      "loss": 0.2603,
+      "step": 14260
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5475789237164365e-05,
+      "loss": 0.0012,
+      "step": 14270
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5447702505336482e-05,
+      "loss": 0.3793,
+      "step": 14280
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5419615773508595e-05,
+      "loss": 0.005,
+      "step": 14290
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5391529041680712e-05,
+      "loss": 0.0053,
+      "step": 14300
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5363442309852825e-05,
+      "loss": 0.0011,
+      "step": 14310
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5335355578024942e-05,
+      "loss": 0.0081,
+      "step": 14320
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5307268846197055e-05,
+      "loss": 0.168,
+      "step": 14330
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5279182114369172e-05,
+      "loss": 0.4157,
+      "step": 14340
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5251095382541289e-05,
+      "loss": 0.2973,
+      "step": 14350
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5223008650713405e-05,
+      "loss": 0.2508,
+      "step": 14360
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5194921918885519e-05,
+      "loss": 0.0022,
+      "step": 14370
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5166835187057636e-05,
+      "loss": 0.0021,
+      "step": 14380
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.513874845522975e-05,
+      "loss": 0.0123,
+      "step": 14390
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5110661723401864e-05,
+      "loss": 0.1859,
+      "step": 14400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.508257499157398e-05,
+      "loss": 0.0105,
+      "step": 14410
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5054488259746097e-05,
+      "loss": 0.0059,
+      "step": 14420
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5026401527918212e-05,
+      "loss": 0.0043,
+      "step": 14430
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4998314796090326e-05,
+      "loss": 0.012,
+      "step": 14440
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4970228064262443e-05,
+      "loss": 0.0016,
+      "step": 14450
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.494214133243456e-05,
+      "loss": 0.0244,
+      "step": 14460
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4914054600606674e-05,
+      "loss": 0.1848,
+      "step": 14470
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4885967868778788e-05,
+      "loss": 0.0312,
+      "step": 14480
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4857881136950904e-05,
+      "loss": 0.0019,
+      "step": 14490
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4829794405123021e-05,
+      "loss": 0.153,
+      "step": 14500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4801707673295138e-05,
+      "loss": 0.0008,
+      "step": 14510
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4773620941467251e-05,
+      "loss": 0.1258,
+      "step": 14520
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4745534209639366e-05,
+      "loss": 0.4804,
+      "step": 14530
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4717447477811483e-05,
+      "loss": 0.04,
+      "step": 14540
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.46893607459836e-05,
+      "loss": 0.349,
+      "step": 14550
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4661274014155713e-05,
+      "loss": 0.0051,
+      "step": 14560
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4633187282327828e-05,
+      "loss": 0.3428,
+      "step": 14570
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4605100550499945e-05,
+      "loss": 0.0983,
+      "step": 14580
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4577013818672062e-05,
+      "loss": 0.1762,
+      "step": 14590
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4548927086844175e-05,
+      "loss": 0.0405,
+      "step": 14600
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.452084035501629e-05,
+      "loss": 0.1872,
+      "step": 14610
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4492753623188407e-05,
+      "loss": 0.0035,
+      "step": 14620
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4464666891360524e-05,
+      "loss": 0.191,
+      "step": 14630
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4436580159532637e-05,
+      "loss": 0.0788,
+      "step": 14640
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4408493427704754e-05,
+      "loss": 0.0031,
+      "step": 14650
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4380406695876869e-05,
+      "loss": 0.0688,
+      "step": 14660
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4352319964048985e-05,
+      "loss": 0.1718,
+      "step": 14670
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4324233232221099e-05,
+      "loss": 0.0008,
+      "step": 14680
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4296146500393215e-05,
+      "loss": 0.5227,
+      "step": 14690
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.426805976856533e-05,
+      "loss": 0.015,
+      "step": 14700
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4239973036737444e-05,
+      "loss": 0.0136,
+      "step": 14710
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.421188630490956e-05,
+      "loss": 0.0152,
+      "step": 14720
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4183799573081677e-05,
+      "loss": 0.0041,
+      "step": 14730
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4155712841253792e-05,
+      "loss": 0.1738,
+      "step": 14740
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4127626109425906e-05,
+      "loss": 0.0151,
+      "step": 14750
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4099539377598022e-05,
+      "loss": 0.0137,
+      "step": 14760
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.407145264577014e-05,
+      "loss": 0.1463,
+      "step": 14770
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4043365913942256e-05,
+      "loss": 0.1483,
+      "step": 14780
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.401527918211437e-05,
+      "loss": 0.0325,
+      "step": 14790
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3987192450286484e-05,
+      "loss": 0.2209,
+      "step": 14800
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3959105718458601e-05,
+      "loss": 0.0011,
+      "step": 14810
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3931018986630718e-05,
+      "loss": 0.2596,
+      "step": 14820
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3902932254802831e-05,
+      "loss": 0.001,
+      "step": 14830
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3874845522974946e-05,
+      "loss": 0.0035,
+      "step": 14840
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3846758791147063e-05,
+      "loss": 0.4387,
+      "step": 14850
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.381867205931918e-05,
+      "loss": 0.0023,
+      "step": 14860
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3790585327491293e-05,
+      "loss": 0.4257,
+      "step": 14870
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3762498595663408e-05,
+      "loss": 0.0009,
+      "step": 14880
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3734411863835525e-05,
+      "loss": 0.0043,
+      "step": 14890
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3706325132007642e-05,
+      "loss": 0.0027,
+      "step": 14900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3678238400179755e-05,
+      "loss": 0.006,
+      "step": 14910
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3650151668351872e-05,
+      "loss": 0.0117,
+      "step": 14920
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.3622064936523987e-05,
+      "loss": 0.0971,
+      "step": 14930
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3593978204696103e-05,
+      "loss": 0.0184,
+      "step": 14940
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3565891472868217e-05,
+      "loss": 0.3738,
+      "step": 14950
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3537804741040334e-05,
+      "loss": 0.001,
+      "step": 14960
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3509718009212449e-05,
+      "loss": 0.1329,
+      "step": 14970
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3481631277384565e-05,
+      "loss": 0.0171,
+      "step": 14980
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3453544545556679e-05,
+      "loss": 0.0146,
+      "step": 14990
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3425457813728795e-05,
+      "loss": 0.0016,
+      "step": 15000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.339737108190091e-05,
+      "loss": 0.001,
+      "step": 15010
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3369284350073027e-05,
+      "loss": 0.012,
+      "step": 15020
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.334119761824514e-05,
+      "loss": 0.0008,
+      "step": 15030
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3313110886417257e-05,
+      "loss": 0.5451,
+      "step": 15040
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3285024154589374e-05,
+      "loss": 0.182,
+      "step": 15050
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3256937422761487e-05,
+      "loss": 0.1511,
+      "step": 15060
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3228850690933602e-05,
+      "loss": 0.0255,
+      "step": 15070
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3200763959105719e-05,
+      "loss": 0.0061,
+      "step": 15080
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3172677227277836e-05,
+      "loss": 0.0016,
+      "step": 15090
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.314459049544995e-05,
+      "loss": 0.4358,
+      "step": 15100
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3116503763622064e-05,
+      "loss": 0.1072,
+      "step": 15110
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3088417031794181e-05,
+      "loss": 0.4036,
+      "step": 15120
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.3060330299966298e-05,
+      "loss": 0.2798,
+      "step": 15130
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3032243568138411e-05,
+      "loss": 0.0009,
+      "step": 15140
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.3004156836310528e-05,
+      "loss": 0.0022,
+      "step": 15150
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2976070104482643e-05,
+      "loss": 0.002,
+      "step": 15160
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.294798337265476e-05,
+      "loss": 0.0302,
+      "step": 15170
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2919896640826873e-05,
+      "loss": 0.0033,
+      "step": 15180
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.289180990899899e-05,
+      "loss": 0.1281,
+      "step": 15190
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2863723177171105e-05,
+      "loss": 0.0297,
+      "step": 15200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2835636445343222e-05,
+      "loss": 0.6917,
+      "step": 15210
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2807549713515335e-05,
+      "loss": 0.1993,
+      "step": 15220
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2779462981687452e-05,
+      "loss": 0.0008,
+      "step": 15230
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2751376249859567e-05,
+      "loss": 0.1286,
+      "step": 15240
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2723289518031683e-05,
+      "loss": 0.0056,
+      "step": 15250
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2695202786203797e-05,
+      "loss": 0.0248,
+      "step": 15260
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2667116054375913e-05,
+      "loss": 0.6,
+      "step": 15270
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.263902932254803e-05,
+      "loss": 0.4821,
+      "step": 15280
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2610942590720145e-05,
+      "loss": 0.406,
+      "step": 15290
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2582855858892259e-05,
+      "loss": 0.0015,
+      "step": 15300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2554769127064375e-05,
+      "loss": 0.0014,
+      "step": 15310
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2526682395236492e-05,
+      "loss": 0.0651,
+      "step": 15320
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2498595663408607e-05,
+      "loss": 0.1126,
+      "step": 15330
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2470508931580722e-05,
+      "loss": 0.0278,
+      "step": 15340
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2442422199752837e-05,
+      "loss": 0.3241,
+      "step": 15350
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2414335467924954e-05,
+      "loss": 0.1976,
+      "step": 15360
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2386248736097069e-05,
+      "loss": 0.0006,
+      "step": 15370
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2358162004269184e-05,
+      "loss": 0.3559,
+      "step": 15380
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2330075272441299e-05,
+      "loss": 0.3139,
+      "step": 15390
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2301988540613414e-05,
+      "loss": 0.001,
+      "step": 15400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2273901808785531e-05,
+      "loss": 0.1977,
+      "step": 15410
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2245815076957646e-05,
+      "loss": 0.2917,
+      "step": 15420
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2217728345129761e-05,
+      "loss": 0.0029,
+      "step": 15430
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2189641613301876e-05,
+      "loss": 0.0241,
+      "step": 15440
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2161554881473993e-05,
+      "loss": 0.3519,
+      "step": 15450
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2133468149646108e-05,
+      "loss": 0.4929,
+      "step": 15460
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2105381417818223e-05,
+      "loss": 0.0078,
+      "step": 15470
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2077294685990338e-05,
+      "loss": 0.0742,
+      "step": 15480
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2049207954162455e-05,
+      "loss": 0.001,
+      "step": 15490
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.202112122233457e-05,
+      "loss": 0.3457,
+      "step": 15500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1993034490506685e-05,
+      "loss": 0.2625,
+      "step": 15510
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.19649477586788e-05,
+      "loss": 0.287,
+      "step": 15520
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1936861026850916e-05,
+      "loss": 0.0357,
+      "step": 15530
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1908774295023032e-05,
+      "loss": 0.2885,
+      "step": 15540
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1880687563195148e-05,
+      "loss": 0.0014,
+      "step": 15550
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1852600831367262e-05,
+      "loss": 0.0241,
+      "step": 15560
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1824514099539378e-05,
+      "loss": 0.001,
+      "step": 15570
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1796427367711493e-05,
+      "loss": 0.0172,
+      "step": 15580
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.176834063588361e-05,
+      "loss": 0.1738,
+      "step": 15590
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1740253904055725e-05,
+      "loss": 0.1105,
+      "step": 15600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.171216717222784e-05,
+      "loss": 0.4393,
+      "step": 15610
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1684080440399955e-05,
+      "loss": 0.0259,
+      "step": 15620
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1655993708572072e-05,
+      "loss": 0.0014,
+      "step": 15630
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1627906976744187e-05,
+      "loss": 0.0292,
+      "step": 15640
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1599820244916302e-05,
+      "loss": 0.0014,
+      "step": 15650
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1571733513088417e-05,
+      "loss": 0.0018,
+      "step": 15660
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1543646781260534e-05,
+      "loss": 0.1849,
+      "step": 15670
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1515560049432649e-05,
+      "loss": 0.3119,
+      "step": 15680
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1487473317604764e-05,
+      "loss": 0.0077,
+      "step": 15690
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1459386585776879e-05,
+      "loss": 0.0005,
+      "step": 15700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1431299853948996e-05,
+      "loss": 0.0091,
+      "step": 15710
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.140321312212111e-05,
+      "loss": 0.0021,
+      "step": 15720
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1375126390293226e-05,
+      "loss": 0.4307,
+      "step": 15730
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1347039658465341e-05,
+      "loss": 0.0017,
+      "step": 15740
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1318952926637456e-05,
+      "loss": 0.0029,
+      "step": 15750
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1290866194809573e-05,
+      "loss": 0.0761,
+      "step": 15760
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1262779462981688e-05,
+      "loss": 0.1169,
+      "step": 15770
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1234692731153804e-05,
+      "loss": 0.1949,
+      "step": 15780
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1206605999325918e-05,
+      "loss": 0.0029,
+      "step": 15790
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1178519267498035e-05,
+      "loss": 0.1022,
+      "step": 15800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.115043253567015e-05,
+      "loss": 0.0483,
+      "step": 15810
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1122345803842266e-05,
+      "loss": 0.0049,
+      "step": 15820
+    },
+    {
+      "epoch": 3.2,
+      "eval_accuracy": 0.7900916010837311,
+      "eval_loss": 1.0053784847259521,
+      "eval_runtime": 1303.5818,
+      "eval_samples_per_second": 5.946,
+      "eval_steps_per_second": 2.973,
+      "step": 15824
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.109425907201438e-05,
+      "loss": 0.0009,
+      "step": 15830
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.1066172340186496e-05,
+      "loss": 0.1814,
+      "step": 15840
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.1038085608358611e-05,
+      "loss": 0.1293,
+      "step": 15850
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.1009998876530728e-05,
+      "loss": 0.0741,
+      "step": 15860
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0981912144702843e-05,
+      "loss": 0.474,
+      "step": 15870
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0953825412874958e-05,
+      "loss": 0.003,
+      "step": 15880
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0925738681047073e-05,
+      "loss": 0.0309,
+      "step": 15890
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.089765194921919e-05,
+      "loss": 0.001,
+      "step": 15900
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0869565217391305e-05,
+      "loss": 0.1312,
+      "step": 15910
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.084147848556342e-05,
+      "loss": 0.0063,
+      "step": 15920
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0813391753735535e-05,
+      "loss": 0.2984,
+      "step": 15930
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0785305021907652e-05,
+      "loss": 0.1325,
+      "step": 15940
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0757218290079767e-05,
+      "loss": 0.0331,
+      "step": 15950
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0729131558251882e-05,
+      "loss": 0.0021,
+      "step": 15960
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0701044826423997e-05,
+      "loss": 0.0432,
+      "step": 15970
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0672958094596114e-05,
+      "loss": 0.0189,
+      "step": 15980
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0644871362768229e-05,
+      "loss": 0.0334,
+      "step": 15990
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0616784630940346e-05,
+      "loss": 0.0008,
+      "step": 16000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0588697899112459e-05,
+      "loss": 0.0165,
+      "step": 16010
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0560611167284576e-05,
+      "loss": 0.0016,
+      "step": 16020
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.053252443545669e-05,
+      "loss": 0.1505,
+      "step": 16030
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0504437703628807e-05,
+      "loss": 0.1337,
+      "step": 16040
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0476350971800922e-05,
+      "loss": 0.0007,
+      "step": 16050
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0448264239973036e-05,
+      "loss": 0.0008,
+      "step": 16060
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0420177508145153e-05,
+      "loss": 0.0018,
+      "step": 16070
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0392090776317268e-05,
+      "loss": 0.1396,
+      "step": 16080
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0364004044489384e-05,
+      "loss": 0.0032,
+      "step": 16090
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.03359173126615e-05,
+      "loss": 0.0668,
+      "step": 16100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0307830580833614e-05,
+      "loss": 0.3123,
+      "step": 16110
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.027974384900573e-05,
+      "loss": 0.0723,
+      "step": 16120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0251657117177846e-05,
+      "loss": 0.0021,
+      "step": 16130
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0223570385349961e-05,
+      "loss": 0.0048,
+      "step": 16140
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0195483653522076e-05,
+      "loss": 0.0035,
+      "step": 16150
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0167396921694191e-05,
+      "loss": 0.0022,
+      "step": 16160
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0139310189866308e-05,
+      "loss": 0.1684,
+      "step": 16170
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0111223458038423e-05,
+      "loss": 0.0008,
+      "step": 16180
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0083136726210538e-05,
+      "loss": 0.0106,
+      "step": 16190
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.0055049994382653e-05,
+      "loss": 0.0551,
+      "step": 16200
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.002696326255477e-05,
+      "loss": 0.0056,
+      "step": 16210
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.998876530726885e-06,
+      "loss": 0.0013,
+      "step": 16220
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.970789798899002e-06,
+      "loss": 0.0004,
+      "step": 16230
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.942703067071115e-06,
+      "loss": 0.2048,
+      "step": 16240
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.914616335243232e-06,
+      "loss": 0.086,
+      "step": 16250
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.886529603415347e-06,
+      "loss": 0.1927,
+      "step": 16260
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.858442871587464e-06,
+      "loss": 0.0007,
+      "step": 16270
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.830356139759577e-06,
+      "loss": 0.004,
+      "step": 16280
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.802269407931694e-06,
+      "loss": 0.0018,
+      "step": 16290
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.774182676103809e-06,
+      "loss": 0.008,
+      "step": 16300
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.746095944275925e-06,
+      "loss": 0.0093,
+      "step": 16310
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.71800921244804e-06,
+      "loss": 0.0008,
+      "step": 16320
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.689922480620156e-06,
+      "loss": 0.0014,
+      "step": 16330
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.66183574879227e-06,
+      "loss": 0.0034,
+      "step": 16340
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.633749016964387e-06,
+      "loss": 0.0005,
+      "step": 16350
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.605662285136502e-06,
+      "loss": 0.0006,
+      "step": 16360
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.577575553308617e-06,
+      "loss": 0.0011,
+      "step": 16370
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.549488821480732e-06,
+      "loss": 0.0044,
+      "step": 16380
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.52140208965285e-06,
+      "loss": 0.0191,
+      "step": 16390
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.493315357824964e-06,
+      "loss": 0.0306,
+      "step": 16400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.46522862599708e-06,
+      "loss": 0.0078,
+      "step": 16410
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.437141894169194e-06,
+      "loss": 0.0045,
+      "step": 16420
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.40905516234131e-06,
+      "loss": 0.0005,
+      "step": 16430
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.380968430513426e-06,
+      "loss": 0.0022,
+      "step": 16440
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.352881698685541e-06,
+      "loss": 0.0021,
+      "step": 16450
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.324794966857656e-06,
+      "loss": 0.0005,
+      "step": 16460
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.296708235029771e-06,
+      "loss": 0.0034,
+      "step": 16470
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.268621503201888e-06,
+      "loss": 0.0071,
+      "step": 16480
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.240534771374003e-06,
+      "loss": 0.0057,
+      "step": 16490
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.21244803954612e-06,
+      "loss": 0.0012,
+      "step": 16500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.184361307718233e-06,
+      "loss": 0.003,
+      "step": 16510
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.15627457589035e-06,
+      "loss": 0.3917,
+      "step": 16520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.128187844062465e-06,
+      "loss": 0.0004,
+      "step": 16530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.100101112234582e-06,
+      "loss": 0.0434,
+      "step": 16540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.072014380406697e-06,
+      "loss": 0.0025,
+      "step": 16550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.043927648578812e-06,
+      "loss": 0.0085,
+      "step": 16560
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.015840916750927e-06,
+      "loss": 0.4074,
+      "step": 16570
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.987754184923044e-06,
+      "loss": 0.0008,
+      "step": 16580
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.959667453095159e-06,
+      "loss": 0.0058,
+      "step": 16590
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.931580721267274e-06,
+      "loss": 0.0005,
+      "step": 16600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.903493989439389e-06,
+      "loss": 0.0007,
+      "step": 16610
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.875407257611505e-06,
+      "loss": 0.0028,
+      "step": 16620
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.84732052578362e-06,
+      "loss": 0.0013,
+      "step": 16630
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.819233793955736e-06,
+      "loss": 0.0004,
+      "step": 16640
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.79114706212785e-06,
+      "loss": 0.0123,
+      "step": 16650
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.763060330299967e-06,
+      "loss": 0.1072,
+      "step": 16660
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.734973598472082e-06,
+      "loss": 0.0007,
+      "step": 16670
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.706886866644199e-06,
+      "loss": 0.0052,
+      "step": 16680
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.678800134816312e-06,
+      "loss": 0.0005,
+      "step": 16690
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.65071340298843e-06,
+      "loss": 0.0005,
+      "step": 16700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 8.622626671160544e-06,
+      "loss": 0.0008,
+      "step": 16710
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.594539939332661e-06,
+      "loss": 0.0041,
+      "step": 16720
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.566453207504776e-06,
+      "loss": 0.004,
+      "step": 16730
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.53836647567689e-06,
+      "loss": 0.0932,
+      "step": 16740
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.510279743849006e-06,
+      "loss": 0.0006,
+      "step": 16750
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.482193012021121e-06,
+      "loss": 0.0016,
+      "step": 16760
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.454106280193238e-06,
+      "loss": 0.336,
+      "step": 16770
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.426019548365351e-06,
+      "loss": 0.0027,
+      "step": 16780
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.397932816537468e-06,
+      "loss": 0.0007,
+      "step": 16790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.369846084709583e-06,
+      "loss": 0.0004,
+      "step": 16800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.3417593528817e-06,
+      "loss": 0.001,
+      "step": 16810
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.313672621053815e-06,
+      "loss": 0.001,
+      "step": 16820
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.28558588922593e-06,
+      "loss": 0.0006,
+      "step": 16830
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.257499157398045e-06,
+      "loss": 0.0999,
+      "step": 16840
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.229412425570162e-06,
+      "loss": 0.0018,
+      "step": 16850
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.201325693742277e-06,
+      "loss": 0.0021,
+      "step": 16860
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.173238961914392e-06,
+      "loss": 0.0008,
+      "step": 16870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.145152230086507e-06,
+      "loss": 0.001,
+      "step": 16880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.117065498258623e-06,
+      "loss": 0.0004,
+      "step": 16890
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.088978766430739e-06,
+      "loss": 0.0003,
+      "step": 16900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.060892034602854e-06,
+      "loss": 0.0004,
+      "step": 16910
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.032805302774969e-06,
+      "loss": 0.001,
+      "step": 16920
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.004718570947085e-06,
+      "loss": 0.0016,
+      "step": 16930
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.9766318391192e-06,
+      "loss": 0.0016,
+      "step": 16940
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.948545107291317e-06,
+      "loss": 0.2856,
+      "step": 16950
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.92045837546343e-06,
+      "loss": 0.0011,
+      "step": 16960
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.892371643635547e-06,
+      "loss": 0.2128,
+      "step": 16970
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.864284911807662e-06,
+      "loss": 0.0008,
+      "step": 16980
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.836198179979779e-06,
+      "loss": 0.0289,
+      "step": 16990
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.808111448151894e-06,
+      "loss": 0.0004,
+      "step": 17000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.780024716324009e-06,
+      "loss": 0.0019,
+      "step": 17010
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.751937984496124e-06,
+      "loss": 0.0109,
+      "step": 17020
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.723851252668241e-06,
+      "loss": 0.0034,
+      "step": 17030
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.695764520840356e-06,
+      "loss": 0.0009,
+      "step": 17040
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.667677789012471e-06,
+      "loss": 0.0039,
+      "step": 17050
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.639591057184586e-06,
+      "loss": 0.0084,
+      "step": 17060
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.611504325356703e-06,
+      "loss": 0.0008,
+      "step": 17070
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.583417593528818e-06,
+      "loss": 0.0004,
+      "step": 17080
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.555330861700932e-06,
+      "loss": 0.0661,
+      "step": 17090
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.527244129873049e-06,
+      "loss": 0.0058,
+      "step": 17100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.499157398045163e-06,
+      "loss": 0.0004,
+      "step": 17110
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.47107066621728e-06,
+      "loss": 0.0134,
+      "step": 17120
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.442983934389394e-06,
+      "loss": 0.0008,
+      "step": 17130
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.414897202561511e-06,
+      "loss": 0.0004,
+      "step": 17140
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.386810470733626e-06,
+      "loss": 0.0682,
+      "step": 17150
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.3587237389057415e-06,
+      "loss": 0.0005,
+      "step": 17160
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.3306370070778566e-06,
+      "loss": 0.1008,
+      "step": 17170
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.3025502752499725e-06,
+      "loss": 0.0006,
+      "step": 17180
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.2744635434220875e-06,
+      "loss": 0.0024,
+      "step": 17190
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.246376811594203e-06,
+      "loss": 0.0969,
+      "step": 17200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.2182900797663184e-06,
+      "loss": 0.184,
+      "step": 17210
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.190203347938434e-06,
+      "loss": 0.0009,
+      "step": 17220
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.162116616110549e-06,
+      "loss": 0.0009,
+      "step": 17230
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.134029884282665e-06,
+      "loss": 0.0069,
+      "step": 17240
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.10594315245478e-06,
+      "loss": 0.0183,
+      "step": 17250
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.077856420626896e-06,
+      "loss": 0.0005,
+      "step": 17260
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.049769688799011e-06,
+      "loss": 0.2438,
+      "step": 17270
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.021682956971128e-06,
+      "loss": 0.0008,
+      "step": 17280
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.993596225143242e-06,
+      "loss": 0.0014,
+      "step": 17290
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.965509493315359e-06,
+      "loss": 0.0272,
+      "step": 17300
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.937422761487473e-06,
+      "loss": 0.0012,
+      "step": 17310
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.90933602965959e-06,
+      "loss": 0.0054,
+      "step": 17320
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.881249297831704e-06,
+      "loss": 0.0005,
+      "step": 17330
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.853162566003821e-06,
+      "loss": 0.0145,
+      "step": 17340
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.825075834175936e-06,
+      "loss": 0.0022,
+      "step": 17350
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.796989102348052e-06,
+      "loss": 0.0199,
+      "step": 17360
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.768902370520167e-06,
+      "loss": 0.0132,
+      "step": 17370
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.740815638692283e-06,
+      "loss": 0.002,
+      "step": 17380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.712728906864398e-06,
+      "loss": 0.0008,
+      "step": 17390
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.684642175036514e-06,
+      "loss": 0.0037,
+      "step": 17400
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.656555443208629e-06,
+      "loss": 0.0008,
+      "step": 17410
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.628468711380744e-06,
+      "loss": 0.0021,
+      "step": 17420
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.6003819795528596e-06,
+      "loss": 0.021,
+      "step": 17430
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.572295247724975e-06,
+      "loss": 0.0003,
+      "step": 17440
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.5442085158970905e-06,
+      "loss": 0.1209,
+      "step": 17450
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.5161217840692056e-06,
+      "loss": 0.002,
+      "step": 17460
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.4880350522413214e-06,
+      "loss": 0.0004,
+      "step": 17470
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.4599483204134365e-06,
+      "loss": 0.0032,
+      "step": 17480
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.431861588585552e-06,
+      "loss": 0.0012,
+      "step": 17490
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.4037748567576674e-06,
+      "loss": 0.2031,
+      "step": 17500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.375688124929783e-06,
+      "loss": 0.0006,
+      "step": 17510
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.347601393101898e-06,
+      "loss": 0.001,
+      "step": 17520
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.319514661274015e-06,
+      "loss": 0.0005,
+      "step": 17530
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.291427929446129e-06,
+      "loss": 0.2285,
+      "step": 17540
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.263341197618246e-06,
+      "loss": 0.0334,
+      "step": 17550
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.235254465790361e-06,
+      "loss": 0.0004,
+      "step": 17560
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.207167733962477e-06,
+      "loss": 0.0092,
+      "step": 17570
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.179081002134592e-06,
+      "loss": 0.0017,
+      "step": 17580
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.150994270306707e-06,
+      "loss": 0.0003,
+      "step": 17590
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.122907538478823e-06,
+      "loss": 0.0024,
+      "step": 17600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.094820806650938e-06,
+      "loss": 0.0009,
+      "step": 17610
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.066734074823054e-06,
+      "loss": 0.0016,
+      "step": 17620
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.038647342995169e-06,
+      "loss": 0.0092,
+      "step": 17630
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.010560611167285e-06,
+      "loss": 0.003,
+      "step": 17640
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.9824738793394e-06,
+      "loss": 0.0004,
+      "step": 17650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.954387147511516e-06,
+      "loss": 0.0005,
+      "step": 17660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.926300415683631e-06,
+      "loss": 0.0372,
+      "step": 17670
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.898213683855747e-06,
+      "loss": 0.0678,
+      "step": 17680
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.870126952027863e-06,
+      "loss": 0.0176,
+      "step": 17690
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.842040220199978e-06,
+      "loss": 0.001,
+      "step": 17700
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.8139534883720935e-06,
+      "loss": 0.0003,
+      "step": 17710
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.7858667565442086e-06,
+      "loss": 0.0192,
+      "step": 17720
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.7577800247163245e-06,
+      "loss": 0.001,
+      "step": 17730
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.7296932928884395e-06,
+      "loss": 0.128,
+      "step": 17740
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.701606561060555e-06,
+      "loss": 0.0006,
+      "step": 17750
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.6735198292326704e-06,
+      "loss": 0.0092,
+      "step": 17760
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.645433097404786e-06,
+      "loss": 0.001,
+      "step": 17770
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.617346365576902e-06,
+      "loss": 0.2708,
+      "step": 17780
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.589259633749017e-06,
+      "loss": 0.0007,
+      "step": 17790
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.561172901921133e-06,
+      "loss": 0.0006,
+      "step": 17800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.533086170093248e-06,
+      "loss": 0.0003,
+      "step": 17810
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.504999438265364e-06,
+      "loss": 0.001,
+      "step": 17820
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.476912706437479e-06,
+      "loss": 0.0006,
+      "step": 17830
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.448825974609595e-06,
+      "loss": 0.0008,
+      "step": 17840
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.42073924278171e-06,
+      "loss": 0.0712,
+      "step": 17850
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.392652510953826e-06,
+      "loss": 0.0003,
+      "step": 17860
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.364565779125941e-06,
+      "loss": 0.0104,
+      "step": 17870
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.336479047298057e-06,
+      "loss": 0.0005,
+      "step": 17880
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.308392315470173e-06,
+      "loss": 0.3098,
+      "step": 17890
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 5.280305583642288e-06,
+      "loss": 0.0008,
+      "step": 17900
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.252218851814404e-06,
+      "loss": 0.0023,
+      "step": 17910
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.224132119986518e-06,
+      "loss": 0.0016,
+      "step": 17920
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.196045388158634e-06,
+      "loss": 0.0028,
+      "step": 17930
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.16795865633075e-06,
+      "loss": 0.0004,
+      "step": 17940
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.139871924502865e-06,
+      "loss": 0.0398,
+      "step": 17950
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.111785192674981e-06,
+      "loss": 0.0399,
+      "step": 17960
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.083698460847096e-06,
+      "loss": 0.2243,
+      "step": 17970
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.0556117290192116e-06,
+      "loss": 0.0009,
+      "step": 17980
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.027524997191327e-06,
+      "loss": 0.0189,
+      "step": 17990
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.9994382653634425e-06,
+      "loss": 0.0009,
+      "step": 18000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.9713515335355576e-06,
+      "loss": 0.0004,
+      "step": 18010
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.9432648017076734e-06,
+      "loss": 0.0004,
+      "step": 18020
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.9151780698797885e-06,
+      "loss": 0.5231,
+      "step": 18030
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.887091338051904e-06,
+      "loss": 0.0005,
+      "step": 18040
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.85900460622402e-06,
+      "loss": 0.0007,
+      "step": 18050
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.830917874396135e-06,
+      "loss": 0.0112,
+      "step": 18060
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.802831142568251e-06,
+      "loss": 0.0074,
+      "step": 18070
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.774744410740366e-06,
+      "loss": 0.0904,
+      "step": 18080
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.746657678912482e-06,
+      "loss": 0.1521,
+      "step": 18090
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.718570947084597e-06,
+      "loss": 0.0004,
+      "step": 18100
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.690484215256713e-06,
+      "loss": 0.0091,
+      "step": 18110
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.662397483428828e-06,
+      "loss": 0.0012,
+      "step": 18120
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.634310751600944e-06,
+      "loss": 0.001,
+      "step": 18130
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.60622401977306e-06,
+      "loss": 0.0005,
+      "step": 18140
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.578137287945175e-06,
+      "loss": 0.0005,
+      "step": 18150
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.550050556117291e-06,
+      "loss": 0.0005,
+      "step": 18160
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.521963824289406e-06,
+      "loss": 0.0005,
+      "step": 18170
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.493877092461522e-06,
+      "loss": 0.0003,
+      "step": 18180
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.465790360633637e-06,
+      "loss": 0.0023,
+      "step": 18190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.437703628805753e-06,
+      "loss": 0.0015,
+      "step": 18200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.409616896977868e-06,
+      "loss": 0.0711,
+      "step": 18210
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.381530165149984e-06,
+      "loss": 0.0014,
+      "step": 18220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.3534434333220995e-06,
+      "loss": 0.2476,
+      "step": 18230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.325356701494215e-06,
+      "loss": 0.0335,
+      "step": 18240
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.2972699696663305e-06,
+      "loss": 0.0006,
+      "step": 18250
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.269183237838445e-06,
+      "loss": 0.0123,
+      "step": 18260
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.2410965060105606e-06,
+      "loss": 0.128,
+      "step": 18270
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.213009774182676e-06,
+      "loss": 0.0005,
+      "step": 18280
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.1849230423547915e-06,
+      "loss": 0.0565,
+      "step": 18290
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.156836310526907e-06,
+      "loss": 0.0006,
+      "step": 18300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.1287495786990224e-06,
+      "loss": 0.0017,
+      "step": 18310
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.100662846871138e-06,
+      "loss": 0.0006,
+      "step": 18320
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.072576115043253e-06,
+      "loss": 0.0541,
+      "step": 18330
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.044489383215369e-06,
+      "loss": 0.0727,
+      "step": 18340
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.016402651387484e-06,
+      "loss": 0.0088,
+      "step": 18350
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.9883159195596e-06,
+      "loss": 0.0026,
+      "step": 18360
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.960229187731715e-06,
+      "loss": 0.2229,
+      "step": 18370
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.932142455903831e-06,
+      "loss": 0.0124,
+      "step": 18380
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.904055724075947e-06,
+      "loss": 0.0039,
+      "step": 18390
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.875968992248062e-06,
+      "loss": 0.0892,
+      "step": 18400
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.847882260420178e-06,
+      "loss": 0.0013,
+      "step": 18410
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.819795528592293e-06,
+      "loss": 0.4419,
+      "step": 18420
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.791708796764409e-06,
+      "loss": 0.0012,
+      "step": 18430
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.7636220649365244e-06,
+      "loss": 0.0009,
+      "step": 18440
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.73553533310864e-06,
+      "loss": 0.0004,
+      "step": 18450
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.7074486012807553e-06,
+      "loss": 0.0003,
+      "step": 18460
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6793618694528708e-06,
+      "loss": 0.2601,
+      "step": 18470
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6512751376249862e-06,
+      "loss": 0.0004,
+      "step": 18480
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6231884057971017e-06,
+      "loss": 0.0024,
+      "step": 18490
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.595101673969217e-06,
+      "loss": 0.141,
+      "step": 18500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.5670149421413326e-06,
+      "loss": 0.0003,
+      "step": 18510
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.538928210313448e-06,
+      "loss": 0.0004,
+      "step": 18520
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.510841478485564e-06,
+      "loss": 0.0003,
+      "step": 18530
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4827547466576795e-06,
+      "loss": 0.0004,
+      "step": 18540
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.454668014829795e-06,
+      "loss": 0.0005,
+      "step": 18550
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4265812830019104e-06,
+      "loss": 0.0003,
+      "step": 18560
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.398494551174026e-06,
+      "loss": 0.0008,
+      "step": 18570
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.3704078193461413e-06,
+      "loss": 0.0005,
+      "step": 18580
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.342321087518257e-06,
+      "loss": 0.479,
+      "step": 18590
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.314234355690372e-06,
+      "loss": 0.0019,
+      "step": 18600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.2861476238624873e-06,
+      "loss": 0.0046,
+      "step": 18610
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.2580608920346028e-06,
+      "loss": 0.0034,
+      "step": 18620
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.2299741602067182e-06,
+      "loss": 0.0007,
+      "step": 18630
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.2018874283788337e-06,
+      "loss": 0.1667,
+      "step": 18640
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.173800696550949e-06,
+      "loss": 0.001,
+      "step": 18650
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.1457139647230646e-06,
+      "loss": 0.001,
+      "step": 18660
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.1176272328951805e-06,
+      "loss": 0.0005,
+      "step": 18670
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.089540501067296e-06,
+      "loss": 0.0011,
+      "step": 18680
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.0614537692394115e-06,
+      "loss": 0.0008,
+      "step": 18690
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.033367037411527e-06,
+      "loss": 0.0003,
+      "step": 18700
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.0052803055836424e-06,
+      "loss": 0.09,
+      "step": 18710
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.977193573755758e-06,
+      "loss": 0.0005,
+      "step": 18720
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9491068419278733e-06,
+      "loss": 0.0011,
+      "step": 18730
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.921020110099989e-06,
+      "loss": 0.0005,
+      "step": 18740
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.8929333782721043e-06,
+      "loss": 0.0013,
+      "step": 18750
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.8648466464442197e-06,
+      "loss": 0.0004,
+      "step": 18760
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.8367599146163352e-06,
+      "loss": 0.0548,
+      "step": 18770
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.808673182788451e-06,
+      "loss": 0.0007,
+      "step": 18780
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.7805864509605666e-06,
+      "loss": 0.1198,
+      "step": 18790
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.752499719132682e-06,
+      "loss": 0.0011,
+      "step": 18800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.7244129873047975e-06,
+      "loss": 0.0016,
+      "step": 18810
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.696326255476913e-06,
+      "loss": 0.0004,
+      "step": 18820
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.6682395236490284e-06,
+      "loss": 0.0325,
+      "step": 18830
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.640152791821144e-06,
+      "loss": 0.2583,
+      "step": 18840
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.612066059993259e-06,
+      "loss": 0.0008,
+      "step": 18850
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.583979328165375e-06,
+      "loss": 0.0003,
+      "step": 18860
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.5558925963374903e-06,
+      "loss": 0.0089,
+      "step": 18870
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.5278058645096058e-06,
+      "loss": 0.0003,
+      "step": 18880
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.4997191326817213e-06,
+      "loss": 0.0003,
+      "step": 18890
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.4716324008538367e-06,
+      "loss": 0.2502,
+      "step": 18900
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.443545669025952e-06,
+      "loss": 0.0594,
+      "step": 18910
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.4154589371980677e-06,
+      "loss": 0.229,
+      "step": 18920
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.387372205370183e-06,
+      "loss": 0.0014,
+      "step": 18930
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.3592854735422986e-06,
+      "loss": 0.0005,
+      "step": 18940
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.331198741714414e-06,
+      "loss": 0.0651,
+      "step": 18950
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.30311200988653e-06,
+      "loss": 0.3065,
+      "step": 18960
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.2750252780586454e-06,
+      "loss": 0.1241,
+      "step": 18970
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.246938546230761e-06,
+      "loss": 0.087,
+      "step": 18980
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.2188518144028764e-06,
+      "loss": 0.0004,
+      "step": 18990
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.190765082574992e-06,
+      "loss": 0.0005,
+      "step": 19000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1626783507471073e-06,
+      "loss": 0.2476,
+      "step": 19010
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1345916189192223e-06,
+      "loss": 0.0054,
+      "step": 19020
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.106504887091338e-06,
+      "loss": 0.0019,
+      "step": 19030
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0784181552634537e-06,
+      "loss": 0.0012,
+      "step": 19040
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.050331423435569e-06,
+      "loss": 0.0218,
+      "step": 19050
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0222446916076846e-06,
+      "loss": 0.0829,
+      "step": 19060
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.9941579597798e-06,
+      "loss": 0.0009,
+      "step": 19070
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.9660712279519156e-06,
+      "loss": 0.0005,
+      "step": 19080
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.937984496124031e-06,
+      "loss": 0.0635,
+      "step": 19090
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.9098977642961465e-06,
+      "loss": 0.0069,
+      "step": 19100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.8818110324682622e-06,
+      "loss": 0.2412,
+      "step": 19110
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.8537243006403776e-06,
+      "loss": 0.0004,
+      "step": 19120
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.8256375688124931e-06,
+      "loss": 0.0008,
+      "step": 19130
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7975508369846086e-06,
+      "loss": 0.0013,
+      "step": 19140
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.769464105156724e-06,
+      "loss": 0.0005,
+      "step": 19150
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7413773733288397e-06,
+      "loss": 0.0062,
+      "step": 19160
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.7132906415009552e-06,
+      "loss": 0.0003,
+      "step": 19170
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6852039096730707e-06,
+      "loss": 0.0336,
+      "step": 19180
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.657117177845186e-06,
+      "loss": 0.0005,
+      "step": 19190
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6290304460173014e-06,
+      "loss": 0.001,
+      "step": 19200
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6009437141894169e-06,
+      "loss": 0.0005,
+      "step": 19210
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.5728569823615323e-06,
+      "loss": 0.0005,
+      "step": 19220
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.544770250533648e-06,
+      "loss": 0.2423,
+      "step": 19230
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.5166835187057635e-06,
+      "loss": 0.08,
+      "step": 19240
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.488596786877879e-06,
+      "loss": 0.0607,
+      "step": 19250
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4605100550499944e-06,
+      "loss": 0.0013,
+      "step": 19260
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4324233232221099e-06,
+      "loss": 0.0005,
+      "step": 19270
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4043365913942256e-06,
+      "loss": 0.0004,
+      "step": 19280
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.376249859566341e-06,
+      "loss": 0.0045,
+      "step": 19290
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3481631277384565e-06,
+      "loss": 0.0728,
+      "step": 19300
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.320076395910572e-06,
+      "loss": 0.0004,
+      "step": 19310
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2919896640826874e-06,
+      "loss": 0.001,
+      "step": 19320
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2639029322548029e-06,
+      "loss": 0.0007,
+      "step": 19330
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2358162004269184e-06,
+      "loss": 0.001,
+      "step": 19340
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2077294685990338e-06,
+      "loss": 0.0007,
+      "step": 19350
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1796427367711493e-06,
+      "loss": 0.055,
+      "step": 19360
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.151556004943265e-06,
+      "loss": 0.4369,
+      "step": 19370
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.1234692731153804e-06,
+      "loss": 0.0044,
+      "step": 19380
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.095382541287496e-06,
+      "loss": 0.0012,
+      "step": 19390
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.0672958094596112e-06,
+      "loss": 0.0353,
+      "step": 19400
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.0392090776317268e-06,
+      "loss": 0.0095,
+      "step": 19410
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.0111223458038423e-06,
+      "loss": 0.0496,
+      "step": 19420
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.830356139759578e-07,
+      "loss": 0.001,
+      "step": 19430
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.549488821480732e-07,
+      "loss": 0.0006,
+      "step": 19440
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.268621503201888e-07,
+      "loss": 0.0007,
+      "step": 19450
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.987754184923043e-07,
+      "loss": 0.0226,
+      "step": 19460
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.706886866644199e-07,
+      "loss": 0.1195,
+      "step": 19470
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 8.426019548365353e-07,
+      "loss": 0.001,
+      "step": 19480
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 8.145152230086507e-07,
+      "loss": 0.0008,
+      "step": 19490
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.864284911807662e-07,
+      "loss": 0.0012,
+      "step": 19500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.583417593528817e-07,
+      "loss": 0.0007,
+      "step": 19510
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.302550275249972e-07,
+      "loss": 0.3778,
+      "step": 19520
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.021682956971128e-07,
+      "loss": 0.0937,
+      "step": 19530
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.740815638692282e-07,
+      "loss": 0.0005,
+      "step": 19540
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.459948320413437e-07,
+      "loss": 0.0004,
+      "step": 19550
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.179081002134592e-07,
+      "loss": 0.0007,
+      "step": 19560
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.898213683855746e-07,
+      "loss": 0.0006,
+      "step": 19570
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.617346365576902e-07,
+      "loss": 0.0004,
+      "step": 19580
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.336479047298056e-07,
+      "loss": 0.0008,
+      "step": 19590
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.055611729019212e-07,
+      "loss": 0.0003,
+      "step": 19600
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.774744410740366e-07,
+      "loss": 0.0019,
+      "step": 19610
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.4938770924615215e-07,
+      "loss": 0.0007,
+      "step": 19620
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.2130097741826767e-07,
+      "loss": 0.0006,
+      "step": 19630
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.932142455903831e-07,
+      "loss": 0.0007,
+      "step": 19640
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.651275137624986e-07,
+      "loss": 0.0143,
+      "step": 19650
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.370407819346141e-07,
+      "loss": 0.0079,
+      "step": 19660
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.089540501067296e-07,
+      "loss": 0.2126,
+      "step": 19670
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.808673182788451e-07,
+      "loss": 0.0069,
+      "step": 19680
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.527805864509606e-07,
+      "loss": 0.0018,
+      "step": 19690
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.2469385462307607e-07,
+      "loss": 0.0003,
+      "step": 19700
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9660712279519154e-07,
+      "loss": 0.0004,
+      "step": 19710
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6852039096730706e-07,
+      "loss": 0.0003,
+      "step": 19720
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.4043365913942256e-07,
+      "loss": 0.0628,
+      "step": 19730
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.1234692731153804e-07,
+      "loss": 0.1076,
+      "step": 19740
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 8.426019548365353e-08,
+      "loss": 0.0037,
+      "step": 19750
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 5.617346365576902e-08,
+      "loss": 0.0018,
+      "step": 19760
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.808673182788451e-08,
+      "loss": 0.0625,
+      "step": 19770
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0,
+      "loss": 0.0012,
+      "step": 19780
+    },
+    {
+      "epoch": 4.2,
+      "eval_accuracy": 0.8205392852535157,
+      "eval_loss": 0.9528893828392029,
+      "eval_runtime": 1277.8645,
+      "eval_samples_per_second": 6.066,
+      "eval_steps_per_second": 3.033,
+      "step": 19780
+    },
+    {
+      "epoch": 4.2,
+      "step": 19780,
+      "total_flos": 4.933783690868294e+19,
+      "train_loss": 0.8840404472118776,
+      "train_runtime": 19817.2516,
+      "train_samples_per_second": 1.996,
+      "train_steps_per_second": 0.998
+    },
+    {
+      "epoch": 4.2,
+      "eval_accuracy": 0.805377888004177,
+      "eval_loss": 1.100114107131958,
+      "eval_runtime": 1295.6967,
+      "eval_samples_per_second": 5.913,
+      "eval_steps_per_second": 2.957,
+      "step": 19780
+    },
+    {
+      "epoch": 4.2,
+      "eval_accuracy": 0.805377888004177,
+      "eval_loss": 1.100114107131958,
+      "eval_runtime": 1245.1415,
+      "eval_samples_per_second": 6.153,
+      "eval_steps_per_second": 3.077,
+      "step": 19780
+    }
+  ],
+  "max_steps": 19780,
+  "num_train_epochs": 9223372036854775807,
+  "total_flos": 4.933783690868294e+19,
+  "trial_name": null,
+  "trial_params": null
+}