diff --git "a/tmp-checkpoint-15416/trainer_state.json" "b/tmp-checkpoint-15416/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/tmp-checkpoint-15416/trainer_state.json"
@@ -0,0 +1,9427 @@
+{
+  "best_metric": 0.9808588624000549,
+  "best_model_checkpoint": "swinv2-base-patch4-window12-192-22k-ConcreteClassifier-PVT\\checkpoint-1927",
+  "epoch": 8.0,
+  "eval_steps": 500,
+  "global_step": 15416,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7298045320878742e-06,
+      "loss": 2.3066,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4596090641757485e-06,
+      "loss": 2.131,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1894135962636225e-06,
+      "loss": 2.0104,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.919218128351497e-06,
+      "loss": 1.8938,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.649022660439371e-06,
+      "loss": 1.7743,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0378827192527245e-05,
+      "loss": 1.73,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2108631724615119e-05,
+      "loss": 1.6552,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3838436256702994e-05,
+      "loss": 1.4119,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5568240788790867e-05,
+      "loss": 1.1716,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7298045320878743e-05,
+      "loss": 1.2431,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9027849852966615e-05,
+      "loss": 1.1115,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.075765438505449e-05,
+      "loss": 1.0365,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2487458917142365e-05,
+      "loss": 1.2179,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4217263449230237e-05,
+      "loss": 1.1192,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5947067981318112e-05,
+      "loss": 0.8697,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7676872513405988e-05,
+      "loss": 1.0286,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.940667704549386e-05,
+      "loss": 0.9186,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1136481577581735e-05,
+      "loss": 1.0909,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.286628610966961e-05,
+      "loss": 1.0519,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4596090641757486e-05,
+      "loss": 1.1043,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.632589517384536e-05,
+      "loss": 0.8634,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.805569970593323e-05,
+      "loss": 0.86,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.97855042380211e-05,
+      "loss": 0.8325,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.151530877010898e-05,
+      "loss": 1.1083,
+      "step": 240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.324511330219685e-05,
+      "loss": 1.0759,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.497491783428473e-05,
+      "loss": 0.9178,
+      "step": 260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.67047223663726e-05,
+      "loss": 1.1132,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8434526898460474e-05,
+      "loss": 1.0484,
+      "step": 280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.0164331430548346e-05,
+      "loss": 0.8629,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.1894135962636225e-05,
+      "loss": 0.6426,
+      "step": 300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.36239404947241e-05,
+      "loss": 1.0117,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.5353745026811976e-05,
+      "loss": 0.9769,
+      "step": 320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.708354955889985e-05,
+      "loss": 0.9459,
+      "step": 330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.881335409098772e-05,
+      "loss": 1.3208,
+      "step": 340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.054315862307559e-05,
+      "loss": 0.9631,
+      "step": 350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.227296315516347e-05,
+      "loss": 1.3533,
+      "step": 360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.400276768725135e-05,
+      "loss": 1.2228,
+      "step": 370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.573257221933921e-05,
+      "loss": 0.9195,
+      "step": 380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.746237675142709e-05,
+      "loss": 1.0466,
+      "step": 390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.919218128351497e-05,
+      "loss": 0.6537,
+      "step": 400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.092198581560285e-05,
+      "loss": 1.0766,
+      "step": 410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.265179034769071e-05,
+      "loss": 1.3196,
+      "step": 420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.438159487977858e-05,
+      "loss": 1.1902,
+      "step": 430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.611139941186646e-05,
+      "loss": 0.8251,
+      "step": 440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.784120394395434e-05,
+      "loss": 0.6196,
+      "step": 450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.95710084760422e-05,
+      "loss": 1.1941,
+      "step": 460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.130081300813008e-05,
+      "loss": 0.9509,
+      "step": 470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.303061754021796e-05,
+      "loss": 1.0669,
+      "step": 480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.476042207230583e-05,
+      "loss": 0.8493,
+      "step": 490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64902266043937e-05,
+      "loss": 1.0381,
+      "step": 500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.822003113648158e-05,
+      "loss": 0.6853,
+      "step": 510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.994983566856946e-05,
+      "loss": 1.0025,
+      "step": 520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.167964020065733e-05,
+      "loss": 0.9758,
+      "step": 530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.34094447327452e-05,
+      "loss": 1.2263,
+      "step": 540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.513924926483307e-05,
+      "loss": 1.0245,
+      "step": 550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.686905379692095e-05,
+      "loss": 0.9136,
+      "step": 560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.859885832900881e-05,
+      "loss": 1.0586,
+      "step": 570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00010032866286109669,
+      "loss": 1.029,
+      "step": 580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00010205846739318457,
+      "loss": 0.9973,
+      "step": 590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00010378827192527245,
+      "loss": 0.7652,
+      "step": 600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00010551807645736032,
+      "loss": 1.2198,
+      "step": 610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001072478809894482,
+      "loss": 1.3727,
+      "step": 620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00010897768552153607,
+      "loss": 0.9157,
+      "step": 630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011070749005362395,
+      "loss": 1.4063,
+      "step": 640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011243729458571182,
+      "loss": 0.9697,
+      "step": 650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001141670991177997,
+      "loss": 1.0678,
+      "step": 660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011589690364988756,
+      "loss": 1.1373,
+      "step": 670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011762670818197544,
+      "loss": 1.1127,
+      "step": 680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001193565127140633,
+      "loss": 0.8828,
+      "step": 690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00012108631724615118,
+      "loss": 1.0497,
+      "step": 700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00012281612177823906,
+      "loss": 1.0252,
+      "step": 710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00012454592631032694,
+      "loss": 1.0175,
+      "step": 720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00012627573084241482,
+      "loss": 1.1303,
+      "step": 730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001280055353745027,
+      "loss": 0.8751,
+      "step": 740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012973533990659055,
+      "loss": 0.9646,
+      "step": 750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013146514443867843,
+      "loss": 1.0412,
+      "step": 760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001331949489707663,
+      "loss": 1.3243,
+      "step": 770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013492475350285418,
+      "loss": 1.0182,
+      "step": 780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013665455803494206,
+      "loss": 1.3814,
+      "step": 790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013838436256702994,
+      "loss": 0.9131,
+      "step": 800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014011416709911782,
+      "loss": 1.0305,
+      "step": 810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001418439716312057,
+      "loss": 0.9296,
+      "step": 820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014357377616329355,
+      "loss": 1.1304,
+      "step": 830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00014530358069538143,
+      "loss": 0.964,
+      "step": 840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001470333852274693,
+      "loss": 0.9223,
+      "step": 850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00014876318975955716,
+      "loss": 0.8419,
+      "step": 860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00015049299429164504,
+      "loss": 0.7896,
+      "step": 870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00015222279882373292,
+      "loss": 1.2753,
+      "step": 880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001539526033558208,
+      "loss": 1.01,
+      "step": 890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00015568240788790867,
+      "loss": 0.8729,
+      "step": 900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00015741221241999653,
+      "loss": 1.2337,
+      "step": 910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001591420169520844,
+      "loss": 1.0598,
+      "step": 920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00016087182148417228,
+      "loss": 0.9216,
+      "step": 930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016260162601626016,
+      "loss": 1.2614,
+      "step": 940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00016433143054834804,
+      "loss": 0.9602,
+      "step": 950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00016606123508043592,
+      "loss": 1.1126,
+      "step": 960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001677910396125238,
+      "loss": 0.7103,
+      "step": 970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00016952084414461165,
+      "loss": 1.0462,
+      "step": 980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017125064867669953,
+      "loss": 0.9843,
+      "step": 990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001729804532087874,
+      "loss": 1.2419,
+      "step": 1000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017471025774087529,
+      "loss": 1.3823,
+      "step": 1010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017644006227296316,
+      "loss": 0.9726,
+      "step": 1020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017816986680505104,
+      "loss": 1.1238,
+      "step": 1030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017989967133713892,
+      "loss": 0.934,
+      "step": 1040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001816294758692268,
+      "loss": 1.5093,
+      "step": 1050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018335928040131465,
+      "loss": 1.0234,
+      "step": 1060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018508908493340253,
+      "loss": 0.7998,
+      "step": 1070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001868188894654904,
+      "loss": 1.0191,
+      "step": 1080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001885486939975783,
+      "loss": 1.1171,
+      "step": 1090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019027849852966614,
+      "loss": 1.1417,
+      "step": 1100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019200830306175402,
+      "loss": 0.9539,
+      "step": 1110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001937381075938419,
+      "loss": 1.0955,
+      "step": 1120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019546791212592978,
+      "loss": 1.3748,
+      "step": 1130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019719771665801763,
+      "loss": 1.1126,
+      "step": 1140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001989275211901055,
+      "loss": 1.0674,
+      "step": 1150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00020065732572219339,
+      "loss": 1.1236,
+      "step": 1160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020238713025428126,
+      "loss": 1.0824,
+      "step": 1170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00020411693478636914,
+      "loss": 1.1181,
+      "step": 1180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00020584673931845702,
+      "loss": 1.0881,
+      "step": 1190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002075765438505449,
+      "loss": 1.1874,
+      "step": 1200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00020930634838263278,
+      "loss": 1.0377,
+      "step": 1210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00021103615291472063,
+      "loss": 0.8027,
+      "step": 1220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002127659574468085,
+      "loss": 0.8027,
+      "step": 1230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002144957619788964,
+      "loss": 1.2303,
+      "step": 1240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00021622556651098427,
+      "loss": 1.2914,
+      "step": 1250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00021795537104307214,
+      "loss": 1.2018,
+      "step": 1260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00021968517557516002,
+      "loss": 0.994,
+      "step": 1270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002214149801072479,
+      "loss": 1.1917,
+      "step": 1280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00022314478463933578,
+      "loss": 0.9846,
+      "step": 1290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00022487458917142363,
+      "loss": 1.1561,
+      "step": 1300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002266043937035115,
+      "loss": 1.3345,
+      "step": 1310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002283341982355994,
+      "loss": 1.1663,
+      "step": 1320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023006400276768727,
+      "loss": 1.0486,
+      "step": 1330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023179380729977512,
+      "loss": 1.2478,
+      "step": 1340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000233523611831863,
+      "loss": 1.1334,
+      "step": 1350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023525341636395088,
+      "loss": 1.1248,
+      "step": 1360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023698322089603876,
+      "loss": 1.2183,
+      "step": 1370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002387130254281266,
+      "loss": 1.4618,
+      "step": 1380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00024044282996021449,
+      "loss": 1.5796,
+      "step": 1390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00024217263449230237,
+      "loss": 1.4877,
+      "step": 1400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00024390243902439024,
+      "loss": 0.9612,
+      "step": 1410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002456322435564781,
+      "loss": 1.3634,
+      "step": 1420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000247362048088566,
+      "loss": 1.1377,
+      "step": 1430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002490918526206539,
+      "loss": 1.1365,
+      "step": 1440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00025082165715274176,
+      "loss": 1.2433,
+      "step": 1450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00025255146168482964,
+      "loss": 1.2164,
+      "step": 1460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002542812662169175,
+      "loss": 1.0668,
+      "step": 1470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002560110707490054,
+      "loss": 1.174,
+      "step": 1480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002577408752810932,
+      "loss": 1.1307,
+      "step": 1490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002594706798131811,
+      "loss": 1.2406,
+      "step": 1500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000261200484345269,
+      "loss": 1.2542,
+      "step": 1510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00026293028887735685,
+      "loss": 0.9126,
+      "step": 1520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00026466009340944473,
+      "loss": 1.1105,
+      "step": 1530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002663898979415326,
+      "loss": 0.8491,
+      "step": 1540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002681197024736205,
+      "loss": 1.4498,
+      "step": 1550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00026984950700570837,
+      "loss": 1.3299,
+      "step": 1560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00027157931153779625,
+      "loss": 1.2222,
+      "step": 1570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002733091160698841,
+      "loss": 0.9812,
+      "step": 1580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000275038920601972,
+      "loss": 1.2824,
+      "step": 1590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002767687251340599,
+      "loss": 1.1105,
+      "step": 1600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00027849852966614776,
+      "loss": 1.0978,
+      "step": 1610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028022833419823564,
+      "loss": 1.1159,
+      "step": 1620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002819581387303235,
+      "loss": 1.1441,
+      "step": 1630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002836879432624114,
+      "loss": 1.2566,
+      "step": 1640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002854177477944992,
+      "loss": 1.1327,
+      "step": 1650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002871475523265871,
+      "loss": 1.232,
+      "step": 1660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000288877356858675,
+      "loss": 1.0974,
+      "step": 1670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029060716139076286,
+      "loss": 1.0083,
+      "step": 1680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029233696592285074,
+      "loss": 1.5513,
+      "step": 1690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002940667704549386,
+      "loss": 1.3731,
+      "step": 1700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029579657498702644,
+      "loss": 1.1877,
+      "step": 1710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002975263795191143,
+      "loss": 1.199,
+      "step": 1720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002992561840512022,
+      "loss": 1.4768,
+      "step": 1730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0003009859885832901,
+      "loss": 1.7352,
+      "step": 1740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00030271579311537796,
+      "loss": 1.3652,
+      "step": 1750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00030444559764746583,
+      "loss": 1.1335,
+      "step": 1760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003061754021795537,
+      "loss": 1.2259,
+      "step": 1770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0003079052067116416,
+      "loss": 1.3024,
+      "step": 1780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00030963501124372947,
+      "loss": 1.4433,
+      "step": 1790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00031136481577581735,
+      "loss": 1.1668,
+      "step": 1800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0003130946203079052,
+      "loss": 1.3298,
+      "step": 1810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00031482442483999305,
+      "loss": 1.2601,
+      "step": 1820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00031655422937208093,
+      "loss": 1.1204,
+      "step": 1830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0003182840339041688,
+      "loss": 1.2142,
+      "step": 1840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0003200138384362567,
+      "loss": 0.93,
+      "step": 1850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00032174364296834457,
+      "loss": 1.4335,
+      "step": 1860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00032347344750043245,
+      "loss": 1.3081,
+      "step": 1870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003252032520325203,
+      "loss": 1.2692,
+      "step": 1880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0003269330565646082,
+      "loss": 1.2172,
+      "step": 1890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0003286628610966961,
+      "loss": 1.2254,
+      "step": 1900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00033039266562878396,
+      "loss": 1.1448,
+      "step": 1910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00033212247016087184,
+      "loss": 1.2473,
+      "step": 1920
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": {
+        "accuracy": 0.5909208819714656
+      },
+      "eval_f1": {
+        "f1": 0.5910435702573883
+      },
+      "eval_loss": 0.9808588624000549,
+      "eval_precision": {
+        "precision": 0.6661654194259725
+      },
+      "eval_recall": {
+        "recall": 0.6016064778916844
+      },
+      "eval_runtime": 243.6104,
+      "eval_samples_per_second": 15.824,
+      "eval_steps_per_second": 7.914,
+      "step": 1927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0003338522746929597,
+      "loss": 1.1687,
+      "step": 1930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003355820792250476,
+      "loss": 1.4294,
+      "step": 1940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0003373118837571355,
+      "loss": 1.2866,
+      "step": 1950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003390416882892233,
+      "loss": 1.0858,
+      "step": 1960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0003407714928213112,
+      "loss": 1.1465,
+      "step": 1970
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00034250129735339906,
+      "loss": 1.2876,
+      "step": 1980
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00034423110188548694,
+      "loss": 1.0266,
+      "step": 1990
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003459609064175748,
+      "loss": 1.1494,
+      "step": 2000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0003476907109496627,
+      "loss": 1.5705,
+      "step": 2010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00034942051548175057,
+      "loss": 0.9916,
+      "step": 2020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00035115032001383845,
+      "loss": 1.2183,
+      "step": 2030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00035288012454592633,
+      "loss": 1.1186,
+      "step": 2040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0003546099290780142,
+      "loss": 1.1386,
+      "step": 2050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0003563397336101021,
+      "loss": 1.413,
+      "step": 2060
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00035806953814218997,
+      "loss": 1.4434,
+      "step": 2070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00035979934267427784,
+      "loss": 1.3144,
+      "step": 2080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0003615291472063657,
+      "loss": 1.4108,
+      "step": 2090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003632589517384536,
+      "loss": 1.2305,
+      "step": 2100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0003649887562705415,
+      "loss": 1.2335,
+      "step": 2110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0003667185608026293,
+      "loss": 1.3617,
+      "step": 2120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0003684483653347172,
+      "loss": 1.1457,
+      "step": 2130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00037017816986680506,
+      "loss": 1.0253,
+      "step": 2140
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00037190797439889294,
+      "loss": 1.1486,
+      "step": 2150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0003736377789309808,
+      "loss": 1.2603,
+      "step": 2160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003753675834630687,
+      "loss": 1.25,
+      "step": 2170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0003770973879951566,
+      "loss": 1.5531,
+      "step": 2180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003788271925272444,
+      "loss": 1.3442,
+      "step": 2190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0003805569970593323,
+      "loss": 1.1317,
+      "step": 2200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00038228680159142016,
+      "loss": 1.3271,
+      "step": 2210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00038401660612350804,
+      "loss": 1.3165,
+      "step": 2220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003857464106555959,
+      "loss": 1.2377,
+      "step": 2230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0003874762151876838,
+      "loss": 1.2677,
+      "step": 2240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0003892060197197717,
+      "loss": 1.2756,
+      "step": 2250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00039093582425185955,
+      "loss": 1.0193,
+      "step": 2260
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039266562878394743,
+      "loss": 1.239,
+      "step": 2270
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00039439543331603526,
+      "loss": 1.6044,
+      "step": 2280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00039612523784812313,
+      "loss": 1.2396,
+      "step": 2290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000397855042380211,
+      "loss": 1.0345,
+      "step": 2300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0003995848469122989,
+      "loss": 1.3125,
+      "step": 2310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00040131465144438677,
+      "loss": 1.4382,
+      "step": 2320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00040304445597647465,
+      "loss": 1.1232,
+      "step": 2330
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00040477426050856253,
+      "loss": 1.3415,
+      "step": 2340
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0004065040650406504,
+      "loss": 1.3517,
+      "step": 2350
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0004082338695727383,
+      "loss": 1.1871,
+      "step": 2360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00040996367410482616,
+      "loss": 1.541,
+      "step": 2370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00041169347863691404,
+      "loss": 1.0871,
+      "step": 2380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0004134232831690019,
+      "loss": 1.0727,
+      "step": 2390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0004151530877010898,
+      "loss": 1.2664,
+      "step": 2400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0004168828922331777,
+      "loss": 0.9518,
+      "step": 2410
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00041861269676526556,
+      "loss": 1.5851,
+      "step": 2420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0004203425012973534,
+      "loss": 1.3433,
+      "step": 2430
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00042207230582944126,
+      "loss": 1.1357,
+      "step": 2440
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00042380211036152914,
+      "loss": 1.3124,
+      "step": 2450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000425531914893617,
+      "loss": 1.3198,
+      "step": 2460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0004272617194257049,
+      "loss": 1.31,
+      "step": 2470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0004289915239577928,
+      "loss": 1.2563,
+      "step": 2480
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00043072132848988065,
+      "loss": 1.2553,
+      "step": 2490
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00043245113302196853,
+      "loss": 1.3314,
+      "step": 2500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0004341809375540564,
+      "loss": 1.8077,
+      "step": 2510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0004359107420861443,
+      "loss": 1.7739,
+      "step": 2520
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00043764054661823217,
+      "loss": 1.3835,
+      "step": 2530
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00043937035115032005,
+      "loss": 1.3689,
+      "step": 2540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0004411001556824079,
+      "loss": 1.4259,
+      "step": 2550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0004428299602144958,
+      "loss": 1.0997,
+      "step": 2560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0004445597647465837,
+      "loss": 0.944,
+      "step": 2570
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00044628956927867156,
+      "loss": 1.431,
+      "step": 2580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0004480193738107594,
+      "loss": 1.492,
+      "step": 2590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00044974917834284726,
+      "loss": 1.0205,
+      "step": 2600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00045147898287493514,
+      "loss": 1.2334,
+      "step": 2610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000453208787407023,
+      "loss": 1.0434,
+      "step": 2620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0004549385919391109,
+      "loss": 1.4912,
+      "step": 2630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0004566683964711988,
+      "loss": 1.2939,
+      "step": 2640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00045839820100328666,
+      "loss": 1.0389,
+      "step": 2650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00046012800553537454,
+      "loss": 1.1925,
+      "step": 2660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0004618578100674624,
+      "loss": 1.5455,
+      "step": 2670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00046358761459955024,
+      "loss": 1.2916,
+      "step": 2680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0004653174191316381,
+      "loss": 1.253,
+      "step": 2690
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000467047223663726,
+      "loss": 1.377,
+      "step": 2700
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0004687770281958139,
+      "loss": 1.4184,
+      "step": 2710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00047050683272790175,
+      "loss": 1.3838,
+      "step": 2720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00047223663725998963,
+      "loss": 1.1912,
+      "step": 2730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0004739664417920775,
+      "loss": 1.0304,
+      "step": 2740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00047569624632416534,
+      "loss": 1.3583,
+      "step": 2750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0004774260508562532,
+      "loss": 1.3856,
+      "step": 2760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0004791558553883411,
+      "loss": 1.5254,
+      "step": 2770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00048088565992042897,
+      "loss": 1.2068,
+      "step": 2780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00048261546445251685,
+      "loss": 1.3795,
+      "step": 2790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00048434526898460473,
+      "loss": 1.5103,
+      "step": 2800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004860750735166926,
+      "loss": 1.6531,
+      "step": 2810
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0004878048780487805,
+      "loss": 1.3675,
+      "step": 2820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0004895346825808684,
+      "loss": 1.6244,
+      "step": 2830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0004912644871129562,
+      "loss": 1.8395,
+      "step": 2840
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0004929942916450441,
+      "loss": 1.6552,
+      "step": 2850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000494724096177132,
+      "loss": 1.6432,
+      "step": 2860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0004964539007092199,
+      "loss": 1.4839,
+      "step": 2870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0004981837052413078,
+      "loss": 1.403,
+      "step": 2880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0004999135097733956,
+      "loss": 1.8624,
+      "step": 2890
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005016433143054835,
+      "loss": 1.4894,
+      "step": 2900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005033731188375714,
+      "loss": 2.1144,
+      "step": 2910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005051029233696593,
+      "loss": 1.7605,
+      "step": 2920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000506832727901747,
+      "loss": 1.6352,
+      "step": 2930
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000508562532433835,
+      "loss": 1.6826,
+      "step": 2940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005102923369659228,
+      "loss": 1.8023,
+      "step": 2950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005120221414980108,
+      "loss": 1.6957,
+      "step": 2960
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005137519460300986,
+      "loss": 2.1667,
+      "step": 2970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005154817505621864,
+      "loss": 2.1662,
+      "step": 2980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005172115550942743,
+      "loss": 1.8724,
+      "step": 2990
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005189413596263622,
+      "loss": 1.7312,
+      "step": 3000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005206711641584501,
+      "loss": 2.0953,
+      "step": 3010
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000522400968690538,
+      "loss": 1.8218,
+      "step": 3020
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005241307732226258,
+      "loss": 1.9732,
+      "step": 3030
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005258605777547137,
+      "loss": 2.229,
+      "step": 3040
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005275903822868016,
+      "loss": 1.9946,
+      "step": 3050
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005293201868188895,
+      "loss": 2.162,
+      "step": 3060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005310499913509773,
+      "loss": 2.0647,
+      "step": 3070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005327797958830652,
+      "loss": 2.0291,
+      "step": 3080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005345096004151531,
+      "loss": 2.0622,
+      "step": 3090
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000536239404947241,
+      "loss": 1.9414,
+      "step": 3100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005379692094793289,
+      "loss": 2.0906,
+      "step": 3110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005396990140114167,
+      "loss": 2.0018,
+      "step": 3120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005414288185435046,
+      "loss": 1.9754,
+      "step": 3130
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005431586230755925,
+      "loss": 1.949,
+      "step": 3140
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005448884276076803,
+      "loss": 1.918,
+      "step": 3150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005466182321397683,
+      "loss": 2.1205,
+      "step": 3160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000548348036671856,
+      "loss": 1.8933,
+      "step": 3170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000550077841203944,
+      "loss": 1.9666,
+      "step": 3180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005518076457360318,
+      "loss": 1.9286,
+      "step": 3190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0005535374502681198,
+      "loss": 1.9417,
+      "step": 3200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005552672548002075,
+      "loss": 1.8296,
+      "step": 3210
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0005569970593322955,
+      "loss": 1.9005,
+      "step": 3220
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005587268638643833,
+      "loss": 1.954,
+      "step": 3230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0005604566683964713,
+      "loss": 1.9252,
+      "step": 3240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000562186472928559,
+      "loss": 1.9827,
+      "step": 3250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000563916277460647,
+      "loss": 1.7931,
+      "step": 3260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005656460819927348,
+      "loss": 1.8842,
+      "step": 3270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0005673758865248228,
+      "loss": 1.8838,
+      "step": 3280
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005691056910569106,
+      "loss": 1.7902,
+      "step": 3290
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0005708354955889984,
+      "loss": 1.9054,
+      "step": 3300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005725653001210863,
+      "loss": 1.972,
+      "step": 3310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0005742951046531742,
+      "loss": 1.999,
+      "step": 3320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0005760249091852621,
+      "loss": 1.9566,
+      "step": 3330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00057775471371735,
+      "loss": 1.8273,
+      "step": 3340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0005794845182494378,
+      "loss": 1.9554,
+      "step": 3350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0005812143227815257,
+      "loss": 2.0318,
+      "step": 3360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0005829441273136136,
+      "loss": 1.9838,
+      "step": 3370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0005846739318457015,
+      "loss": 1.9402,
+      "step": 3380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0005864037363777894,
+      "loss": 1.9904,
+      "step": 3390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0005881335409098772,
+      "loss": 1.9198,
+      "step": 3400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0005898633454419651,
+      "loss": 1.9405,
+      "step": 3410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0005915931499740529,
+      "loss": 1.8445,
+      "step": 3420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0005933229545061409,
+      "loss": 1.9201,
+      "step": 3430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0005950527590382286,
+      "loss": 1.8237,
+      "step": 3440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0005967825635703165,
+      "loss": 1.7255,
+      "step": 3450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0005985123681024044,
+      "loss": 1.844,
+      "step": 3460
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0006002421726344923,
+      "loss": 2.0433,
+      "step": 3470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0006019719771665802,
+      "loss": 1.7753,
+      "step": 3480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000603701781698668,
+      "loss": 1.858,
+      "step": 3490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006054315862307559,
+      "loss": 1.9348,
+      "step": 3500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0006071613907628438,
+      "loss": 1.8196,
+      "step": 3510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006088911952949317,
+      "loss": 2.0001,
+      "step": 3520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0006106209998270195,
+      "loss": 1.806,
+      "step": 3530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006123508043591074,
+      "loss": 1.8369,
+      "step": 3540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0006140806088911953,
+      "loss": 1.8824,
+      "step": 3550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006158104134232832,
+      "loss": 1.8376,
+      "step": 3560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0006175402179553711,
+      "loss": 1.7554,
+      "step": 3570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006192700224874589,
+      "loss": 1.835,
+      "step": 3580
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0006209998270195468,
+      "loss": 1.7473,
+      "step": 3590
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006227296315516347,
+      "loss": 1.8827,
+      "step": 3600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0006244594360837226,
+      "loss": 1.7536,
+      "step": 3610
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006261892406158103,
+      "loss": 1.8997,
+      "step": 3620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0006279190451478983,
+      "loss": 1.813,
+      "step": 3630
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006296488496799861,
+      "loss": 1.8474,
+      "step": 3640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0006313786542120741,
+      "loss": 1.8263,
+      "step": 3650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006331084587441619,
+      "loss": 1.8817,
+      "step": 3660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0006348382632762499,
+      "loss": 1.9347,
+      "step": 3670
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006365680678083376,
+      "loss": 1.786,
+      "step": 3680
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0006382978723404256,
+      "loss": 2.197,
+      "step": 3690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0006400276768725134,
+      "loss": 1.9695,
+      "step": 3700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006417574814046014,
+      "loss": 1.8939,
+      "step": 3710
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0006434872859366891,
+      "loss": 1.8662,
+      "step": 3720
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006452170904687771,
+      "loss": 1.8054,
+      "step": 3730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0006469468950008649,
+      "loss": 1.8351,
+      "step": 3740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006486766995329529,
+      "loss": 1.9102,
+      "step": 3750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0006504065040650406,
+      "loss": 1.8462,
+      "step": 3760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006521363085971285,
+      "loss": 1.894,
+      "step": 3770
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0006538661131292164,
+      "loss": 1.7559,
+      "step": 3780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006555959176613043,
+      "loss": 1.8255,
+      "step": 3790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0006573257221933922,
+      "loss": 1.7118,
+      "step": 3800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00065905552672548,
+      "loss": 1.8048,
+      "step": 3810
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0006607853312575679,
+      "loss": 1.7759,
+      "step": 3820
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006625151357896558,
+      "loss": 1.7974,
+      "step": 3830
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0006642449403217437,
+      "loss": 1.7673,
+      "step": 3840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006659747448538316,
+      "loss": 1.7959,
+      "step": 3850
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": {
+        "accuracy": 0.26848249027237353
+      },
+      "eval_f1": {
+        "f1": 0.17635435227079216
+      },
+      "eval_loss": 1.8579649925231934,
+      "eval_precision": {
+        "precision": 0.15600300783137744
+      },
+      "eval_recall": {
+        "recall": 0.2748312652287413
+      },
+      "eval_runtime": 115.151,
+      "eval_samples_per_second": 33.478,
+      "eval_steps_per_second": 16.743,
+      "step": 3854
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0006677045493859194,
+      "loss": 1.9251,
+      "step": 3860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006694343539180073,
+      "loss": 1.7743,
+      "step": 3870
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0006711641584500952,
+      "loss": 1.8658,
+      "step": 3880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000672893962982183,
+      "loss": 1.8682,
+      "step": 3890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000674623767514271,
+      "loss": 1.9279,
+      "step": 3900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006763535720463587,
+      "loss": 2.0145,
+      "step": 3910
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0006780833765784466,
+      "loss": 2.0323,
+      "step": 3920
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0006798131811105345,
+      "loss": 1.9978,
+      "step": 3930
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0006815429856426224,
+      "loss": 1.9258,
+      "step": 3940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0006832727901747102,
+      "loss": 1.9777,
+      "step": 3950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0006850025947067981,
+      "loss": 1.9442,
+      "step": 3960
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000686732399238886,
+      "loss": 1.9772,
+      "step": 3970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0006884622037709739,
+      "loss": 1.9368,
+      "step": 3980
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0006901920083030618,
+      "loss": 1.8948,
+      "step": 3990
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0006919218128351496,
+      "loss": 1.9485,
+      "step": 4000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0006936516173672375,
+      "loss": 1.8472,
+      "step": 4010
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0006953814218993254,
+      "loss": 1.7604,
+      "step": 4020
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0006971112264314133,
+      "loss": 1.8371,
+      "step": 4030
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0006988410309635011,
+      "loss": 1.7566,
+      "step": 4040
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000700570835495589,
+      "loss": 1.9021,
+      "step": 4050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0007023006400276769,
+      "loss": 1.744,
+      "step": 4060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0007040304445597648,
+      "loss": 1.8639,
+      "step": 4070
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0007057602490918527,
+      "loss": 1.857,
+      "step": 4080
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0007074900536239404,
+      "loss": 1.8229,
+      "step": 4090
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0007092198581560284,
+      "loss": 1.8733,
+      "step": 4100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0007109496626881162,
+      "loss": 1.8442,
+      "step": 4110
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0007126794672202042,
+      "loss": 1.8209,
+      "step": 4120
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0007144092717522919,
+      "loss": 1.7018,
+      "step": 4130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0007161390762843799,
+      "loss": 1.8013,
+      "step": 4140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0007178688808164677,
+      "loss": 1.9447,
+      "step": 4150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0007195986853485557,
+      "loss": 1.8768,
+      "step": 4160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0007213284898806435,
+      "loss": 1.822,
+      "step": 4170
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0007230582944127314,
+      "loss": 1.77,
+      "step": 4180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0007247880989448192,
+      "loss": 1.8653,
+      "step": 4190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0007265179034769072,
+      "loss": 1.7695,
+      "step": 4200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000728247708008995,
+      "loss": 1.8426,
+      "step": 4210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.000729977512541083,
+      "loss": 1.8379,
+      "step": 4220
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0007317073170731707,
+      "loss": 1.8384,
+      "step": 4230
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0007334371216052586,
+      "loss": 1.9245,
+      "step": 4240
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0007351669261373465,
+      "loss": 1.8017,
+      "step": 4250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0007368967306694344,
+      "loss": 1.8151,
+      "step": 4260
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0007386265352015222,
+      "loss": 1.8356,
+      "step": 4270
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0007403563397336101,
+      "loss": 1.9064,
+      "step": 4280
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000742086144265698,
+      "loss": 1.8325,
+      "step": 4290
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0007438159487977859,
+      "loss": 1.8079,
+      "step": 4300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0007455457533298738,
+      "loss": 1.7581,
+      "step": 4310
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0007472755578619616,
+      "loss": 1.9346,
+      "step": 4320
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0007490053623940495,
+      "loss": 1.8706,
+      "step": 4330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0007507351669261374,
+      "loss": 1.8314,
+      "step": 4340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0007524649714582253,
+      "loss": 1.8881,
+      "step": 4350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0007541947759903132,
+      "loss": 1.8785,
+      "step": 4360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000755924580522401,
+      "loss": 1.7916,
+      "step": 4370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0007576543850544888,
+      "loss": 1.8089,
+      "step": 4380
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0007593841895865767,
+      "loss": 1.7736,
+      "step": 4390
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0007611139941186646,
+      "loss": 1.8184,
+      "step": 4400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0007628437986507524,
+      "loss": 1.8084,
+      "step": 4410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0007645736031828403,
+      "loss": 2.0347,
+      "step": 4420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0007663034077149282,
+      "loss": 1.7984,
+      "step": 4430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0007680332122470161,
+      "loss": 1.7799,
+      "step": 4440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000769763016779104,
+      "loss": 1.7603,
+      "step": 4450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0007714928213111918,
+      "loss": 1.6663,
+      "step": 4460
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0007732226258432797,
+      "loss": 1.9232,
+      "step": 4470
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0007749524303753676,
+      "loss": 1.8086,
+      "step": 4480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0007766822349074555,
+      "loss": 1.7616,
+      "step": 4490
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0007784120394395433,
+      "loss": 1.8018,
+      "step": 4500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0007801418439716312,
+      "loss": 1.8413,
+      "step": 4510
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0007818716485037191,
+      "loss": 2.1126,
+      "step": 4520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000783601453035807,
+      "loss": 1.9105,
+      "step": 4530
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0007853312575678949,
+      "loss": 1.8485,
+      "step": 4540
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0007870610620999827,
+      "loss": 1.8835,
+      "step": 4550
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0007887908666320705,
+      "loss": 1.7881,
+      "step": 4560
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0007905206711641585,
+      "loss": 1.7013,
+      "step": 4570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0007922504756962463,
+      "loss": 1.7657,
+      "step": 4580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0007939802802283343,
+      "loss": 1.9326,
+      "step": 4590
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.000795710084760422,
+      "loss": 1.6888,
+      "step": 4600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00079743988929251,
+      "loss": 1.8114,
+      "step": 4610
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0007991696938245978,
+      "loss": 1.9245,
+      "step": 4620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0008008994983566858,
+      "loss": 1.8287,
+      "step": 4630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0008026293028887735,
+      "loss": 1.8763,
+      "step": 4640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0008043591074208615,
+      "loss": 1.6896,
+      "step": 4650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0008060889119529493,
+      "loss": 1.9996,
+      "step": 4660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0008078187164850373,
+      "loss": 1.7345,
+      "step": 4670
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0008095485210171251,
+      "loss": 1.7908,
+      "step": 4680
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000811278325549213,
+      "loss": 1.8212,
+      "step": 4690
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0008130081300813008,
+      "loss": 1.8434,
+      "step": 4700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0008147379346133887,
+      "loss": 1.9086,
+      "step": 4710
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0008164677391454766,
+      "loss": 1.9098,
+      "step": 4720
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0008181975436775644,
+      "loss": 1.7695,
+      "step": 4730
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0008199273482096523,
+      "loss": 1.7415,
+      "step": 4740
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0008216571527417402,
+      "loss": 1.8291,
+      "step": 4750
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0008233869572738281,
+      "loss": 1.7842,
+      "step": 4760
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000825116761805916,
+      "loss": 1.8745,
+      "step": 4770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0008268465663380038,
+      "loss": 1.7075,
+      "step": 4780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0008285763708700917,
+      "loss": 1.8133,
+      "step": 4790
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0008303061754021796,
+      "loss": 1.7318,
+      "step": 4800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0008320359799342675,
+      "loss": 1.874,
+      "step": 4810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0008337657844663554,
+      "loss": 1.8895,
+      "step": 4820
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0008354955889984432,
+      "loss": 1.8485,
+      "step": 4830
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0008372253935305311,
+      "loss": 1.7643,
+      "step": 4840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000838955198062619,
+      "loss": 1.8477,
+      "step": 4850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0008406850025947068,
+      "loss": 1.8611,
+      "step": 4860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0008424148071267946,
+      "loss": 1.8908,
+      "step": 4870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0008441446116588825,
+      "loss": 1.8017,
+      "step": 4880
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0008458744161909704,
+      "loss": 1.9186,
+      "step": 4890
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0008476042207230583,
+      "loss": 1.8141,
+      "step": 4900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0008493340252551462,
+      "loss": 1.8298,
+      "step": 4910
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000851063829787234,
+      "loss": 1.7711,
+      "step": 4920
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0008527936343193219,
+      "loss": 1.8265,
+      "step": 4930
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0008545234388514098,
+      "loss": 1.7649,
+      "step": 4940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0008562532433834977,
+      "loss": 1.9159,
+      "step": 4950
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0008579830479155855,
+      "loss": 1.7968,
+      "step": 4960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0008597128524476734,
+      "loss": 1.807,
+      "step": 4970
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0008614426569797613,
+      "loss": 1.7633,
+      "step": 4980
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0008631724615118492,
+      "loss": 1.8019,
+      "step": 4990
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0008649022660439371,
+      "loss": 1.8697,
+      "step": 5000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0008666320705760249,
+      "loss": 1.8348,
+      "step": 5010
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0008683618751081128,
+      "loss": 1.805,
+      "step": 5020
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0008700916796402006,
+      "loss": 1.7346,
+      "step": 5030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0008718214841722886,
+      "loss": 1.9219,
+      "step": 5040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0008735512887043763,
+      "loss": 2.0954,
+      "step": 5050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0008752810932364643,
+      "loss": 1.9262,
+      "step": 5060
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0008770108977685521,
+      "loss": 1.8491,
+      "step": 5070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0008787407023006401,
+      "loss": 2.0113,
+      "step": 5080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0008804705068327279,
+      "loss": 1.9341,
+      "step": 5090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0008822003113648159,
+      "loss": 1.8968,
+      "step": 5100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0008839301158969036,
+      "loss": 2.0009,
+      "step": 5110
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0008856599204289916,
+      "loss": 1.9956,
+      "step": 5120
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0008873897249610794,
+      "loss": 1.9083,
+      "step": 5130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0008891195294931674,
+      "loss": 1.9274,
+      "step": 5140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0008908493340252551,
+      "loss": 1.8263,
+      "step": 5150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0008925791385573431,
+      "loss": 1.8235,
+      "step": 5160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0008943089430894309,
+      "loss": 1.8786,
+      "step": 5170
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0008960387476215188,
+      "loss": 1.8952,
+      "step": 5180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0008977685521536067,
+      "loss": 1.7448,
+      "step": 5190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0008994983566856945,
+      "loss": 1.9518,
+      "step": 5200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009012281612177824,
+      "loss": 1.8636,
+      "step": 5210
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009029579657498703,
+      "loss": 1.9309,
+      "step": 5220
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009046877702819582,
+      "loss": 1.8714,
+      "step": 5230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000906417574814046,
+      "loss": 1.7624,
+      "step": 5240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009081473793461339,
+      "loss": 1.9421,
+      "step": 5250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009098771838782218,
+      "loss": 1.9204,
+      "step": 5260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009116069884103097,
+      "loss": 1.8561,
+      "step": 5270
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0009133367929423976,
+      "loss": 1.8766,
+      "step": 5280
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0009150665974744854,
+      "loss": 1.8905,
+      "step": 5290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0009167964020065733,
+      "loss": 1.8466,
+      "step": 5300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0009185262065386612,
+      "loss": 1.8735,
+      "step": 5310
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0009202560110707491,
+      "loss": 2.0758,
+      "step": 5320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0009219858156028368,
+      "loss": 1.9582,
+      "step": 5330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0009237156201349248,
+      "loss": 1.9317,
+      "step": 5340
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0009254454246670126,
+      "loss": 1.8719,
+      "step": 5350
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0009271752291991005,
+      "loss": 1.8377,
+      "step": 5360
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0009289050337311884,
+      "loss": 1.9785,
+      "step": 5370
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0009306348382632762,
+      "loss": 1.9259,
+      "step": 5380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0009323646427953641,
+      "loss": 1.9097,
+      "step": 5390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000934094447327452,
+      "loss": 1.9034,
+      "step": 5400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0009358242518595399,
+      "loss": 1.8983,
+      "step": 5410
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0009375540563916278,
+      "loss": 1.8115,
+      "step": 5420
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0009392838609237156,
+      "loss": 1.8606,
+      "step": 5430
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0009410136654558035,
+      "loss": 1.8883,
+      "step": 5440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0009427434699878914,
+      "loss": 1.7868,
+      "step": 5450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0009444732745199793,
+      "loss": 1.9107,
+      "step": 5460
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0009462030790520671,
+      "loss": 1.9427,
+      "step": 5470
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000947932883584155,
+      "loss": 1.8159,
+      "step": 5480
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0009496626881162429,
+      "loss": 1.909,
+      "step": 5490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0009513924926483307,
+      "loss": 1.8716,
+      "step": 5500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0009531222971804187,
+      "loss": 1.9116,
+      "step": 5510
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0009548521017125064,
+      "loss": 1.8846,
+      "step": 5520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0009565819062445944,
+      "loss": 1.8776,
+      "step": 5530
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0009583117107766822,
+      "loss": 1.828,
+      "step": 5540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0009600415153087702,
+      "loss": 1.9392,
+      "step": 5550
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0009617713198408579,
+      "loss": 1.8823,
+      "step": 5560
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0009635011243729459,
+      "loss": 1.8782,
+      "step": 5570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0009652309289050337,
+      "loss": 1.8898,
+      "step": 5580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0009669607334371217,
+      "loss": 1.8107,
+      "step": 5590
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0009686905379692095,
+      "loss": 1.8026,
+      "step": 5600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0009704203425012974,
+      "loss": 1.8616,
+      "step": 5610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0009721501470333852,
+      "loss": 1.8443,
+      "step": 5620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0009738799515654732,
+      "loss": 1.8465,
+      "step": 5630
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000975609756097561,
+      "loss": 1.8806,
+      "step": 5640
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0009773395606296489,
+      "loss": 1.8441,
+      "step": 5650
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0009790693651617367,
+      "loss": 1.8194,
+      "step": 5660
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0009807991696938246,
+      "loss": 1.9228,
+      "step": 5670
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0009825289742259125,
+      "loss": 1.792,
+      "step": 5680
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0009842587787580004,
+      "loss": 1.8211,
+      "step": 5690
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0009859885832900882,
+      "loss": 1.81,
+      "step": 5700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0009877183878221761,
+      "loss": 1.8862,
+      "step": 5710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000989448192354264,
+      "loss": 1.8373,
+      "step": 5720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0009911779968863519,
+      "loss": 1.8639,
+      "step": 5730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0009929078014184398,
+      "loss": 1.8718,
+      "step": 5740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0009946376059505276,
+      "loss": 1.8258,
+      "step": 5750
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0009963674104826155,
+      "loss": 1.8564,
+      "step": 5760
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0009980972150147034,
+      "loss": 1.9462,
+      "step": 5770
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0009998270195467913,
+      "loss": 1.8437,
+      "step": 5780
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": {
+        "accuracy": 0.1743190661478599
+      },
+      "eval_f1": {
+        "f1": 0.1111482743173735
+      },
+      "eval_loss": 1.9242523908615112,
+      "eval_precision": {
+        "precision": 0.1516816437642854
+      },
+      "eval_recall": {
+        "recall": 0.1887752678966561
+      },
+      "eval_runtime": 118.0197,
+      "eval_samples_per_second": 32.664,
+      "eval_steps_per_second": 16.336,
+      "step": 5781
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000999999926169841,
+      "loss": 1.8482,
+      "step": 5790
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0009999996709545039,
+      "loss": 1.8564,
+      "step": 5800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0009999992334425987,
+      "loss": 1.8474,
+      "step": 5810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0009999986136342847,
+      "loss": 1.8997,
+      "step": 5820
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000999997811529788,
+      "loss": 1.8401,
+      "step": 5830
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000999996827129401,
+      "loss": 1.8567,
+      "step": 5840
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0009999956604334825,
+      "loss": 1.9804,
+      "step": 5850
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000999994311442458,
+      "loss": 1.8427,
+      "step": 5860
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0009999927801568192,
+      "loss": 1.8681,
+      "step": 5870
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0009999910665771247,
+      "loss": 1.8561,
+      "step": 5880
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000999989170703999,
+      "loss": 1.9098,
+      "step": 5890
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0009999870925381333,
+      "loss": 1.935,
+      "step": 5900
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0009999848320802853,
+      "loss": 2.0045,
+      "step": 5910
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0009999823893312794,
+      "loss": 1.9466,
+      "step": 5920
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0009999797642920058,
+      "loss": 1.9047,
+      "step": 5930
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0009999769569634218,
+      "loss": 1.9901,
+      "step": 5940
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000999973967346551,
+      "loss": 1.9159,
+      "step": 5950
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0009999707954424836,
+      "loss": 1.9826,
+      "step": 5960
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0009999674412523754,
+      "loss": 1.9556,
+      "step": 5970
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0009999639047774498,
+      "loss": 1.9591,
+      "step": 5980
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000999960186018996,
+      "loss": 1.9551,
+      "step": 5990
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0009999562849783699,
+      "loss": 1.937,
+      "step": 6000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0009999522016569938,
+      "loss": 1.9167,
+      "step": 6010
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0009999479360563565,
+      "loss": 1.9296,
+      "step": 6020
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.000999943488178013,
+      "loss": 1.9602,
+      "step": 6030
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0009999388580235853,
+      "loss": 1.9229,
+      "step": 6040
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0009999340455947613,
+      "loss": 1.9508,
+      "step": 6050
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0009999290508932957,
+      "loss": 1.9451,
+      "step": 6060
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0009999238739210093,
+      "loss": 1.9379,
+      "step": 6070
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00099991851467979,
+      "loss": 1.9622,
+      "step": 6080
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0009999129731715912,
+      "loss": 1.9257,
+      "step": 6090
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0009999072493984337,
+      "loss": 1.936,
+      "step": 6100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0009999013433624041,
+      "loss": 1.939,
+      "step": 6110
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.000999895255065656,
+      "loss": 1.9445,
+      "step": 6120
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0009998889845104092,
+      "loss": 1.945,
+      "step": 6130
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0009998825316989493,
+      "loss": 1.989,
+      "step": 6140
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0009998758966336297,
+      "loss": 1.9469,
+      "step": 6150
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0009998690793168688,
+      "loss": 1.9297,
+      "step": 6160
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0009998620797511525,
+      "loss": 1.9461,
+      "step": 6170
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0009998548979390331,
+      "loss": 1.9486,
+      "step": 6180
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0009998475338831286,
+      "loss": 1.9473,
+      "step": 6190
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.000999839987586124,
+      "loss": 1.9557,
+      "step": 6200
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0009998322590507706,
+      "loss": 1.9447,
+      "step": 6210
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0009998243482798863,
+      "loss": 1.9628,
+      "step": 6220
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0009998162552763551,
+      "loss": 1.9514,
+      "step": 6230
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0009998079800431282,
+      "loss": 1.9163,
+      "step": 6240
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000999799522583222,
+      "loss": 1.9278,
+      "step": 6250
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00099979088289972,
+      "loss": 1.9194,
+      "step": 6260
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0009997820609957732,
+      "loss": 1.9528,
+      "step": 6270
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0009997730568745966,
+      "loss": 1.9465,
+      "step": 6280
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0009997638705394741,
+      "loss": 1.9105,
+      "step": 6290
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0009997545019937546,
+      "loss": 1.9432,
+      "step": 6300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0009997449512408538,
+      "loss": 1.9348,
+      "step": 6310
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0009997352182842538,
+      "loss": 1.9364,
+      "step": 6320
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0009997253031275035,
+      "loss": 1.9734,
+      "step": 6330
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0009997152057742173,
+      "loss": 1.9606,
+      "step": 6340
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0009997049262280772,
+      "loss": 1.9483,
+      "step": 6350
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0009996944644928307,
+      "loss": 1.9196,
+      "step": 6360
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0009996838205722923,
+      "loss": 1.9586,
+      "step": 6370
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0009996729944703427,
+      "loss": 1.9599,
+      "step": 6380
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0009996619861909288,
+      "loss": 1.9236,
+      "step": 6390
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0009996507957380643,
+      "loss": 1.9563,
+      "step": 6400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000999639423115829,
+      "loss": 1.9324,
+      "step": 6410
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0009996278683283696,
+      "loss": 1.9386,
+      "step": 6420
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0009996161313798985,
+      "loss": 1.9506,
+      "step": 6430
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0009996042122746955,
+      "loss": 1.9184,
+      "step": 6440
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0009995921110171058,
+      "loss": 1.955,
+      "step": 6450
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0009995798276115414,
+      "loss": 1.9215,
+      "step": 6460
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0009995673620624807,
+      "loss": 1.9362,
+      "step": 6470
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0009995547143744688,
+      "loss": 1.939,
+      "step": 6480
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.000999541884552117,
+      "loss": 1.8658,
+      "step": 6490
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0009995288726001028,
+      "loss": 1.9591,
+      "step": 6500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0009995156785231702,
+      "loss": 1.9814,
+      "step": 6510
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0009995023023261298,
+      "loss": 1.9623,
+      "step": 6520
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0009994887440138588,
+      "loss": 1.9069,
+      "step": 6530
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0009994750035912997,
+      "loss": 1.8249,
+      "step": 6540
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0009994610810634627,
+      "loss": 1.8952,
+      "step": 6550
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.000999446976435424,
+      "loss": 1.9932,
+      "step": 6560
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0009994326897123256,
+      "loss": 1.8771,
+      "step": 6570
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0009994182208993764,
+      "loss": 2.0021,
+      "step": 6580
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.000999403570001852,
+      "loss": 1.9962,
+      "step": 6590
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0009993887370250939,
+      "loss": 1.9772,
+      "step": 6600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00099937372197451,
+      "loss": 1.951,
+      "step": 6610
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0009993585248555745,
+      "loss": 1.9398,
+      "step": 6620
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0009993431456738284,
+      "loss": 1.9512,
+      "step": 6630
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.000999327584434879,
+      "loss": 1.9954,
+      "step": 6640
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0009993118411443995,
+      "loss": 1.9426,
+      "step": 6650
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00099929591580813,
+      "loss": 1.9368,
+      "step": 6660
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0009992798084318766,
+      "loss": 1.9514,
+      "step": 6670
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0009992635190215122,
+      "loss": 1.9406,
+      "step": 6680
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0009992470475829755,
+      "loss": 1.9547,
+      "step": 6690
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0009992303941222722,
+      "loss": 1.9348,
+      "step": 6700
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.000999213558645474,
+      "loss": 1.9717,
+      "step": 6710
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0009991965411587187,
+      "loss": 1.9505,
+      "step": 6720
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000999179341668211,
+      "loss": 1.925,
+      "step": 6730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0009991619601802217,
+      "loss": 1.9599,
+      "step": 6740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000999144396701088,
+      "loss": 1.9617,
+      "step": 6750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0009991266512372132,
+      "loss": 1.9616,
+      "step": 6760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0009991087237950676,
+      "loss": 1.9458,
+      "step": 6770
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0009990906143811868,
+      "loss": 1.9476,
+      "step": 6780
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0009990723230021742,
+      "loss": 1.919,
+      "step": 6790
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.000999053849664698,
+      "loss": 1.9499,
+      "step": 6800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.000999035194375494,
+      "loss": 1.9444,
+      "step": 6810
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0009990163571413633,
+      "loss": 1.9498,
+      "step": 6820
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0009989973379691739,
+      "loss": 1.9729,
+      "step": 6830
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0009989781368658604,
+      "loss": 2.0205,
+      "step": 6840
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0009989587538384232,
+      "loss": 1.9512,
+      "step": 6850
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0009989391888939295,
+      "loss": 1.9329,
+      "step": 6860
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000998919442039512,
+      "loss": 1.9252,
+      "step": 6870
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0009988995132823705,
+      "loss": 1.9392,
+      "step": 6880
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0009988794026297711,
+      "loss": 1.9263,
+      "step": 6890
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000998859110089046,
+      "loss": 1.9332,
+      "step": 6900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0009988386356675932,
+      "loss": 1.9138,
+      "step": 6910
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.000998817979372878,
+      "loss": 1.9538,
+      "step": 6920
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0009987971412124318,
+      "loss": 1.9492,
+      "step": 6930
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0009987761211938514,
+      "loss": 1.9498,
+      "step": 6940
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.000998754919324801,
+      "loss": 1.9517,
+      "step": 6950
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0009987335356130107,
+      "loss": 1.916,
+      "step": 6960
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0009987119700662767,
+      "loss": 1.9625,
+      "step": 6970
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0009986902226924615,
+      "loss": 1.9502,
+      "step": 6980
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.000998668293499494,
+      "loss": 1.9368,
+      "step": 6990
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0009986461824953701,
+      "loss": 1.9518,
+      "step": 7000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0009986238896881508,
+      "loss": 1.9244,
+      "step": 7010
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0009986014150859637,
+      "loss": 1.9348,
+      "step": 7020
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0009985787586970034,
+      "loss": 1.9646,
+      "step": 7030
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0009985559205295301,
+      "loss": 1.9216,
+      "step": 7040
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0009985329005918703,
+      "loss": 1.9065,
+      "step": 7050
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0009985096988924169,
+      "loss": 1.9527,
+      "step": 7060
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0009984863154396293,
+      "loss": 1.9143,
+      "step": 7070
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0009984627502420329,
+      "loss": 1.9082,
+      "step": 7080
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0009984390033082194,
+      "loss": 1.9009,
+      "step": 7090
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0009984150746468465,
+      "loss": 1.766,
+      "step": 7100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0009983909642666388,
+      "loss": 1.8211,
+      "step": 7110
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.000998366672176387,
+      "loss": 1.8446,
+      "step": 7120
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000998342198384947,
+      "loss": 1.8342,
+      "step": 7130
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0009983175429012425,
+      "loss": 1.8967,
+      "step": 7140
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0009982927057342623,
+      "loss": 1.8948,
+      "step": 7150
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0009982676868930623,
+      "loss": 1.9191,
+      "step": 7160
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0009982424863867639,
+      "loss": 1.9185,
+      "step": 7170
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0009982171042245552,
+      "loss": 1.8958,
+      "step": 7180
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0009981915404156901,
+      "loss": 1.9371,
+      "step": 7190
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0009981657949694892,
+      "loss": 1.8752,
+      "step": 7200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.000998139867895339,
+      "loss": 1.8434,
+      "step": 7210
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0009981137592026926,
+      "loss": 1.8518,
+      "step": 7220
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0009980874689010687,
+      "loss": 1.7907,
+      "step": 7230
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000998060997000053,
+      "loss": 1.8275,
+      "step": 7240
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0009980343435092966,
+      "loss": 1.9524,
+      "step": 7250
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0009980075084385174,
+      "loss": 1.8303,
+      "step": 7260
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.000997980491797499,
+      "loss": 1.8564,
+      "step": 7270
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.000997953293596092,
+      "loss": 1.8884,
+      "step": 7280
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.000997925913844212,
+      "loss": 1.862,
+      "step": 7290
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0009978983525518422,
+      "loss": 1.8021,
+      "step": 7300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0009978706097290308,
+      "loss": 1.9126,
+      "step": 7310
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0009978426853858929,
+      "loss": 1.9132,
+      "step": 7320
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0009978145795326093,
+      "loss": 1.8305,
+      "step": 7330
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0009977862921794272,
+      "loss": 1.871,
+      "step": 7340
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0009977578233366603,
+      "loss": 1.8928,
+      "step": 7350
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0009977291730146876,
+      "loss": 1.8817,
+      "step": 7360
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0009977003412239556,
+      "loss": 1.8307,
+      "step": 7370
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0009976713279749754,
+      "loss": 1.8866,
+      "step": 7380
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0009976421332783256,
+      "loss": 1.8442,
+      "step": 7390
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00099761275714465,
+      "loss": 1.8367,
+      "step": 7400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0009975831995846591,
+      "loss": 1.8936,
+      "step": 7410
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0009975534606091296,
+      "loss": 1.807,
+      "step": 7420
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0009975235402289039,
+      "loss": 1.8231,
+      "step": 7430
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0009974934384548906,
+      "loss": 1.8122,
+      "step": 7440
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.000997463155298065,
+      "loss": 1.8028,
+      "step": 7450
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.000997432690769468,
+      "loss": 1.8177,
+      "step": 7460
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0009974020448802066,
+      "loss": 1.9069,
+      "step": 7470
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0009973712176414543,
+      "loss": 1.8245,
+      "step": 7480
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0009973402090644504,
+      "loss": 1.8876,
+      "step": 7490
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0009973090191605005,
+      "loss": 1.8971,
+      "step": 7500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.000997277647940976,
+      "loss": 1.7917,
+      "step": 7510
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.000997246095417315,
+      "loss": 1.8608,
+      "step": 7520
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.000997214361601021,
+      "loss": 1.7863,
+      "step": 7530
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0009971824465036642,
+      "loss": 1.8487,
+      "step": 7540
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0009971503501368805,
+      "loss": 1.8698,
+      "step": 7550
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0009971180725123722,
+      "loss": 1.8424,
+      "step": 7560
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.000997085613641907,
+      "loss": 1.9235,
+      "step": 7570
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0009970529735373198,
+      "loss": 1.7802,
+      "step": 7580
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0009970201522105106,
+      "loss": 1.8938,
+      "step": 7590
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.000996987149673446,
+      "loss": 1.7626,
+      "step": 7600
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0009969539659381583,
+      "loss": 1.8598,
+      "step": 7610
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0009969206010167465,
+      "loss": 1.8503,
+      "step": 7620
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0009968870549213749,
+      "loss": 1.9692,
+      "step": 7630
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0009968533276642742,
+      "loss": 1.8035,
+      "step": 7640
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.000996819419257741,
+      "loss": 1.8945,
+      "step": 7650
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0009967853297141384,
+      "loss": 1.8432,
+      "step": 7660
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0009967510590458953,
+      "loss": 1.8781,
+      "step": 7670
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.000996716607265506,
+      "loss": 1.8669,
+      "step": 7680
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0009966819743855318,
+      "loss": 1.8776,
+      "step": 7690
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0009966471604185997,
+      "loss": 1.803,
+      "step": 7700
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": {
+        "accuracy": 0.1846952010376135
+      },
+      "eval_f1": {
+        "f1": 0.1316379146314846
+      },
+      "eval_loss": 1.9003419876098633,
+      "eval_precision": {
+        "precision": 0.1460659091286986
+      },
+      "eval_recall": {
+        "recall": 0.1928584239028949
+      },
+      "eval_runtime": 112.0693,
+      "eval_samples_per_second": 34.398,
+      "eval_steps_per_second": 17.204,
+      "step": 7708
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0009966121653774023,
+      "loss": 1.8044,
+      "step": 7710
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.000996576989274699,
+      "loss": 1.856,
+      "step": 7720
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0009965416321233143,
+      "loss": 1.7691,
+      "step": 7730
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0009965060939361393,
+      "loss": 1.8381,
+      "step": 7740
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0009964703747261312,
+      "loss": 1.9571,
+      "step": 7750
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.000996434474506313,
+      "loss": 1.8362,
+      "step": 7760
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0009963983932897735,
+      "loss": 1.776,
+      "step": 7770
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0009963621310896678,
+      "loss": 1.8468,
+      "step": 7780
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0009963256879192165,
+      "loss": 1.7727,
+      "step": 7790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0009962890637917072,
+      "loss": 1.8783,
+      "step": 7800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0009962522587204923,
+      "loss": 1.8325,
+      "step": 7810
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0009962152727189908,
+      "loss": 1.8977,
+      "step": 7820
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0009961781058006876,
+      "loss": 1.7941,
+      "step": 7830
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0009961407579791334,
+      "loss": 1.7944,
+      "step": 7840
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0009961032292679453,
+      "loss": 1.7098,
+      "step": 7850
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0009960655196808055,
+      "loss": 1.9654,
+      "step": 7860
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0009960276292314632,
+      "loss": 1.755,
+      "step": 7870
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0009959895579337325,
+      "loss": 1.7779,
+      "step": 7880
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0009959513058014945,
+      "loss": 2.0262,
+      "step": 7890
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.000995912872848695,
+      "loss": 1.9572,
+      "step": 7900
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.000995874259089347,
+      "loss": 1.7426,
+      "step": 7910
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0009958354645375287,
+      "loss": 1.7812,
+      "step": 7920
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.000995796489207384,
+      "loss": 1.8882,
+      "step": 7930
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0009957573331131236,
+      "loss": 1.8841,
+      "step": 7940
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.000995717996269023,
+      "loss": 1.8111,
+      "step": 7950
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0009956784786894243,
+      "loss": 1.7847,
+      "step": 7960
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0009956387803887355,
+      "loss": 1.8502,
+      "step": 7970
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0009955989013814303,
+      "loss": 1.8171,
+      "step": 7980
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0009955588416820482,
+      "loss": 1.9437,
+      "step": 7990
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0009955186013051948,
+      "loss": 1.8288,
+      "step": 8000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0009954781802655418,
+      "loss": 1.7936,
+      "step": 8010
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0009954375785778257,
+      "loss": 1.7904,
+      "step": 8020
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0009953967962568501,
+      "loss": 1.743,
+      "step": 8030
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0009953558333174839,
+      "loss": 1.7847,
+      "step": 8040
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0009953146897746618,
+      "loss": 1.8605,
+      "step": 8050
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0009952733656433848,
+      "loss": 1.7898,
+      "step": 8060
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.000995231860938719,
+      "loss": 1.9566,
+      "step": 8070
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.000995190175675797,
+      "loss": 1.8443,
+      "step": 8080
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0009951483098698168,
+      "loss": 1.8683,
+      "step": 8090
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0009951062635360425,
+      "loss": 1.786,
+      "step": 8100
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.000995064036689804,
+      "loss": 1.8314,
+      "step": 8110
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0009950216293464966,
+      "loss": 1.8071,
+      "step": 8120
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.000994979041521582,
+      "loss": 1.7505,
+      "step": 8130
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0009949362732305875,
+      "loss": 1.7527,
+      "step": 8140
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0009948933244891059,
+      "loss": 1.91,
+      "step": 8150
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0009948501953127961,
+      "loss": 1.7183,
+      "step": 8160
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0009948068857173829,
+      "loss": 1.9057,
+      "step": 8170
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0009947633957186567,
+      "loss": 1.7652,
+      "step": 8180
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0009947197253324731,
+      "loss": 1.8544,
+      "step": 8190
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0009946758745747548,
+      "loss": 1.9037,
+      "step": 8200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.000994631843461489,
+      "loss": 1.8158,
+      "step": 8210
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0009945876320087293,
+      "loss": 1.8054,
+      "step": 8220
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.000994543240232595,
+      "loss": 1.9452,
+      "step": 8230
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0009944986681492708,
+      "loss": 1.8333,
+      "step": 8240
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0009944539157750077,
+      "loss": 1.7858,
+      "step": 8250
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0009944089831261218,
+      "loss": 1.8569,
+      "step": 8260
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0009943638702189954,
+      "loss": 1.9051,
+      "step": 8270
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0009943185770700767,
+      "loss": 1.8262,
+      "step": 8280
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0009942731036958784,
+      "loss": 1.7873,
+      "step": 8290
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0009942274501129808,
+      "loss": 1.9258,
+      "step": 8300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0009941816163380284,
+      "loss": 1.7442,
+      "step": 8310
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0009941356023877317,
+      "loss": 1.7348,
+      "step": 8320
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0009940894082788674,
+      "loss": 1.8439,
+      "step": 8330
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0009940430340282776,
+      "loss": 1.9081,
+      "step": 8340
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0009939964796528698,
+      "loss": 1.8382,
+      "step": 8350
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0009939497451696175,
+      "loss": 1.754,
+      "step": 8360
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00099390283059556,
+      "loss": 1.7955,
+      "step": 8370
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0009938557359478017,
+      "loss": 1.6966,
+      "step": 8380
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.000993808461243513,
+      "loss": 1.7841,
+      "step": 8390
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0009937610064999305,
+      "loss": 1.9377,
+      "step": 8400
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0009937133717343553,
+      "loss": 2.106,
+      "step": 8410
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0009936655569641548,
+      "loss": 1.874,
+      "step": 8420
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0009936175622067623,
+      "loss": 1.9231,
+      "step": 8430
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0009935693874796759,
+      "loss": 1.8391,
+      "step": 8440
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00099352103280046,
+      "loss": 1.8697,
+      "step": 8450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0009934724981867447,
+      "loss": 1.8482,
+      "step": 8460
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0009934237836562248,
+      "loss": 1.7992,
+      "step": 8470
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.000993374889226662,
+      "loss": 1.8635,
+      "step": 8480
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.000993325814915882,
+      "loss": 1.9378,
+      "step": 8490
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0009932765607417778,
+      "loss": 1.7518,
+      "step": 8500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0009932271267223064,
+      "loss": 1.8512,
+      "step": 8510
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.000993177512875492,
+      "loss": 1.8,
+      "step": 8520
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0009931277192194228,
+      "loss": 1.9404,
+      "step": 8530
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0009930777457722534,
+      "loss": 1.8171,
+      "step": 8540
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.000993027592552204,
+      "loss": 1.853,
+      "step": 8550
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0009929772595775597,
+      "loss": 1.7074,
+      "step": 8560
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.000992926746866672,
+      "loss": 1.886,
+      "step": 8570
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0009928760544379571,
+      "loss": 1.786,
+      "step": 8580
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0009928251823098976,
+      "loss": 1.7788,
+      "step": 8590
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0009927741305010408,
+      "loss": 1.7721,
+      "step": 8600
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0009927228990299999,
+      "loss": 1.8738,
+      "step": 8610
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0009926714879154538,
+      "loss": 1.8784,
+      "step": 8620
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0009926198971761462,
+      "loss": 1.7806,
+      "step": 8630
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.000992568126830887,
+      "loss": 1.7354,
+      "step": 8640
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0009925161768985515,
+      "loss": 1.8456,
+      "step": 8650
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00099246404739808,
+      "loss": 1.9295,
+      "step": 8660
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0009924117383484788,
+      "loss": 1.7736,
+      "step": 8670
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0009923592497688193,
+      "loss": 1.9097,
+      "step": 8680
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0009923065816782382,
+      "loss": 1.7617,
+      "step": 8690
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0009922537340959386,
+      "loss": 1.8676,
+      "step": 8700
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.000992200707041188,
+      "loss": 1.8388,
+      "step": 8710
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0009921475005333195,
+      "loss": 1.8154,
+      "step": 8720
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0009920941145917322,
+      "loss": 1.9333,
+      "step": 8730
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00099204054923589,
+      "loss": 1.843,
+      "step": 8740
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0009919868044853228,
+      "loss": 1.9432,
+      "step": 8750
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0009919328803596255,
+      "loss": 1.7557,
+      "step": 8760
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0009918787768784581,
+      "loss": 1.805,
+      "step": 8770
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0009918244940615468,
+      "loss": 1.8165,
+      "step": 8780
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0009917700319286827,
+      "loss": 1.8896,
+      "step": 8790
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.000991715390499722,
+      "loss": 1.8423,
+      "step": 8800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.000991660569794587,
+      "loss": 1.9545,
+      "step": 8810
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0009916055698332646,
+      "loss": 1.8081,
+      "step": 8820
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0009915503906358079,
+      "loss": 1.7959,
+      "step": 8830
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0009914950322223344,
+      "loss": 1.7619,
+      "step": 8840
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0009914394946130276,
+      "loss": 1.6856,
+      "step": 8850
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0009913837778281363,
+      "loss": 1.8222,
+      "step": 8860
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.000991327881887974,
+      "loss": 1.7221,
+      "step": 8870
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0009912718068129207,
+      "loss": 1.7537,
+      "step": 8880
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0009912155526234203,
+      "loss": 1.7016,
+      "step": 8890
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0009911591193399831,
+      "loss": 1.84,
+      "step": 8900
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0009911025069831842,
+      "loss": 1.8604,
+      "step": 8910
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0009910457155736642,
+      "loss": 1.7591,
+      "step": 8920
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0009909887451321287,
+      "loss": 1.8894,
+      "step": 8930
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.000990931595679349,
+      "loss": 1.8903,
+      "step": 8940
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.000990874267236161,
+      "loss": 1.8954,
+      "step": 8950
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0009908167598234666,
+      "loss": 1.7949,
+      "step": 8960
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0009907590734622325,
+      "loss": 1.8333,
+      "step": 8970
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.000990701208173491,
+      "loss": 1.8182,
+      "step": 8980
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.000990643163978339,
+      "loss": 1.8232,
+      "step": 8990
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.000990584940897939,
+      "loss": 1.6914,
+      "step": 9000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0009905265389535192,
+      "loss": 1.8622,
+      "step": 9010
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0009904679581663723,
+      "loss": 1.8321,
+      "step": 9020
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0009904091985578564,
+      "loss": 1.8133,
+      "step": 9030
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0009903502601493952,
+      "loss": 1.825,
+      "step": 9040
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.000990291142962477,
+      "loss": 1.8074,
+      "step": 9050
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0009902318470186554,
+      "loss": 1.7792,
+      "step": 9060
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0009901723723395495,
+      "loss": 1.8017,
+      "step": 9070
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0009901127189468433,
+      "loss": 1.8355,
+      "step": 9080
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0009900528868622862,
+      "loss": 1.7781,
+      "step": 9090
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0009899928761076923,
+      "loss": 1.7382,
+      "step": 9100
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0009899326867049414,
+      "loss": 1.7697,
+      "step": 9110
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.000989872318675978,
+      "loss": 1.7791,
+      "step": 9120
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0009898117720428118,
+      "loss": 1.853,
+      "step": 9130
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0009897510468275179,
+      "loss": 1.8826,
+      "step": 9140
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0009896901430522362,
+      "loss": 1.9778,
+      "step": 9150
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0009896290607391718,
+      "loss": 1.9321,
+      "step": 9160
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.000989567799910595,
+      "loss": 1.9429,
+      "step": 9170
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.000989506360588841,
+      "loss": 1.9588,
+      "step": 9180
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0009894447427963103,
+      "loss": 1.9601,
+      "step": 9190
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0009893829465554681,
+      "loss": 1.9501,
+      "step": 9200
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0009893209718888454,
+      "loss": 1.9414,
+      "step": 9210
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0009892588188190371,
+      "loss": 2.0087,
+      "step": 9220
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0009891964873687041,
+      "loss": 1.9263,
+      "step": 9230
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.000989133977560572,
+      "loss": 1.9342,
+      "step": 9240
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0009890712894174316,
+      "loss": 1.9505,
+      "step": 9250
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0009890084229621384,
+      "loss": 1.9658,
+      "step": 9260
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0009889453782176132,
+      "loss": 1.9293,
+      "step": 9270
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0009888821552068417,
+      "loss": 1.946,
+      "step": 9280
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0009888187539528746,
+      "loss": 2.0008,
+      "step": 9290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0009887551744788272,
+      "loss": 1.9339,
+      "step": 9300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0009886914168078808,
+      "loss": 1.9194,
+      "step": 9310
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0009886274809632807,
+      "loss": 1.9403,
+      "step": 9320
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0009885633669683375,
+      "loss": 1.9332,
+      "step": 9330
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0009884990748464267,
+      "loss": 1.9838,
+      "step": 9340
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0009884346046209887,
+      "loss": 1.9355,
+      "step": 9350
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0009883699563155292,
+      "loss": 1.9767,
+      "step": 9360
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0009883051299536183,
+      "loss": 1.9497,
+      "step": 9370
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0009882401255588912,
+      "loss": 1.9496,
+      "step": 9380
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0009881749431550485,
+      "loss": 1.9549,
+      "step": 9390
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.000988109582765855,
+      "loss": 1.937,
+      "step": 9400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0009880440444151404,
+      "loss": 1.9494,
+      "step": 9410
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0009879783281267999,
+      "loss": 1.9397,
+      "step": 9420
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.000987912433924793,
+      "loss": 1.9257,
+      "step": 9430
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0009878463618331445,
+      "loss": 1.8816,
+      "step": 9440
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0009877801118759438,
+      "loss": 1.9802,
+      "step": 9450
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0009877136840773452,
+      "loss": 1.9163,
+      "step": 9460
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0009876470784615677,
+      "loss": 1.8752,
+      "step": 9470
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0009875802950528954,
+      "loss": 1.9372,
+      "step": 9480
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.000987513333875677,
+      "loss": 1.8707,
+      "step": 9490
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0009874461949543262,
+      "loss": 1.7964,
+      "step": 9500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0009873788783133213,
+      "loss": 1.8374,
+      "step": 9510
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0009873113839772058,
+      "loss": 1.7562,
+      "step": 9520
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0009872437119705871,
+      "loss": 1.7843,
+      "step": 9530
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0009871758623181384,
+      "loss": 1.7316,
+      "step": 9540
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0009871078350445971,
+      "loss": 1.8084,
+      "step": 9550
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0009870396301747658,
+      "loss": 1.8954,
+      "step": 9560
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.000986971247733511,
+      "loss": 1.7907,
+      "step": 9570
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0009869026877457649,
+      "loss": 1.9981,
+      "step": 9580
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0009868339502365237,
+      "loss": 1.8354,
+      "step": 9590
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.000986765035230849,
+      "loss": 1.8247,
+      "step": 9600
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0009866959427538666,
+      "loss": 1.7769,
+      "step": 9610
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.000986626672830767,
+      "loss": 1.8071,
+      "step": 9620
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0009865572254868058,
+      "loss": 1.8121,
+      "step": 9630
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": {
+        "accuracy": 0.33229571984435796
+      },
+      "eval_f1": {
+        "f1": 0.24134878979912178
+      },
+      "eval_loss": 1.7385518550872803,
+      "eval_precision": {
+        "precision": 0.2476900084001273
+      },
+      "eval_recall": {
+        "recall": 0.3378007007551934
+      },
+      "eval_runtime": 102.5366,
+      "eval_samples_per_second": 37.596,
+      "eval_steps_per_second": 18.803,
+      "step": 9635
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.000986487600747303,
+      "loss": 1.842,
+      "step": 9640
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0009864177986376428,
+      "loss": 1.6668,
+      "step": 9650
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0009863478191832755,
+      "loss": 1.7779,
+      "step": 9660
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0009862776624097146,
+      "loss": 1.9484,
+      "step": 9670
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0009862073283425387,
+      "loss": 1.7389,
+      "step": 9680
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0009861368170073915,
+      "loss": 1.7946,
+      "step": 9690
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0009860661284299807,
+      "loss": 1.9001,
+      "step": 9700
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.000985995262636079,
+      "loss": 1.8473,
+      "step": 9710
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0009859242196515235,
+      "loss": 1.8383,
+      "step": 9720
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0009858529995022162,
+      "loss": 1.7071,
+      "step": 9730
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0009857816022141232,
+      "loss": 1.7298,
+      "step": 9740
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0009857100278132759,
+      "loss": 1.7207,
+      "step": 9750
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0009856382763257694,
+      "loss": 1.7967,
+      "step": 9760
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0009855663477777638,
+      "loss": 1.8998,
+      "step": 9770
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0009854942421954843,
+      "loss": 1.8676,
+      "step": 9780
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0009854219596052197,
+      "loss": 1.9189,
+      "step": 9790
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000985349500033324,
+      "loss": 1.7721,
+      "step": 9800
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000985276863506215,
+      "loss": 1.8258,
+      "step": 9810
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0009852040500503761,
+      "loss": 1.7623,
+      "step": 9820
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0009851310596923541,
+      "loss": 1.7993,
+      "step": 9830
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0009850578924587613,
+      "loss": 1.7903,
+      "step": 9840
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0009849845483762736,
+      "loss": 1.8658,
+      "step": 9850
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.000984911027471632,
+      "loss": 1.8133,
+      "step": 9860
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0009848373297716413,
+      "loss": 1.842,
+      "step": 9870
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.000984763455303172,
+      "loss": 1.7339,
+      "step": 9880
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0009846894040931573,
+      "loss": 1.7764,
+      "step": 9890
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0009846151761685965,
+      "loss": 1.832,
+      "step": 9900
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0009845407715565523,
+      "loss": 1.8243,
+      "step": 9910
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0009844661902841522,
+      "loss": 1.8842,
+      "step": 9920
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0009843914323785878,
+      "loss": 1.8539,
+      "step": 9930
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0009843164978671159,
+      "loss": 1.7264,
+      "step": 9940
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0009842413867770566,
+      "loss": 1.8868,
+      "step": 9950
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.000984166099135795,
+      "loss": 1.7069,
+      "step": 9960
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0009840906349707808,
+      "loss": 1.7897,
+      "step": 9970
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0009840149943095274,
+      "loss": 1.7677,
+      "step": 9980
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0009839391771796129,
+      "loss": 1.643,
+      "step": 9990
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0009838631836086797,
+      "loss": 1.7819,
+      "step": 10000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.000983787013624435,
+      "loss": 1.8601,
+      "step": 10010
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0009837106672546492,
+      "loss": 1.7895,
+      "step": 10020
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0009836341445271582,
+      "loss": 1.8085,
+      "step": 10030
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0009835574454698612,
+      "loss": 1.787,
+      "step": 10040
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0009834805701107229,
+      "loss": 1.8519,
+      "step": 10050
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0009834035184777708,
+      "loss": 1.7236,
+      "step": 10060
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0009833262905990978,
+      "loss": 1.8543,
+      "step": 10070
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0009832488865028606,
+      "loss": 1.6843,
+      "step": 10080
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0009831713062172802,
+      "loss": 1.6678,
+      "step": 10090
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.000983093549770642,
+      "loss": 1.9228,
+      "step": 10100
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.000983015617191295,
+      "loss": 1.7595,
+      "step": 10110
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0009829375085076534,
+      "loss": 1.7518,
+      "step": 10120
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0009828592237481953,
+      "loss": 1.7682,
+      "step": 10130
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0009827807629414623,
+      "loss": 1.7051,
+      "step": 10140
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0009827021261160607,
+      "loss": 1.7019,
+      "step": 10150
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0009826233133006612,
+      "loss": 1.6837,
+      "step": 10160
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0009825443245239985,
+      "loss": 1.7594,
+      "step": 10170
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.000982465159814871,
+      "loss": 1.867,
+      "step": 10180
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.000982385819202142,
+      "loss": 1.6618,
+      "step": 10190
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0009823063027147383,
+      "loss": 1.6971,
+      "step": 10200
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0009822266103816516,
+      "loss": 1.9753,
+      "step": 10210
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0009821467422319364,
+      "loss": 1.7991,
+      "step": 10220
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0009820666982947126,
+      "loss": 1.8229,
+      "step": 10230
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0009819864785991638,
+      "loss": 1.8833,
+      "step": 10240
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0009819060831745372,
+      "loss": 1.8254,
+      "step": 10250
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0009818255120501447,
+      "loss": 1.7899,
+      "step": 10260
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0009817447652553618,
+      "loss": 1.7448,
+      "step": 10270
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0009816638428196286,
+      "loss": 1.7687,
+      "step": 10280
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0009815827447724483,
+      "loss": 1.7017,
+      "step": 10290
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0009815014711433891,
+      "loss": 1.6916,
+      "step": 10300
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.000981420021962083,
+      "loss": 1.6695,
+      "step": 10310
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0009813383972582254,
+      "loss": 1.916,
+      "step": 10320
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0009812565970615766,
+      "loss": 1.8299,
+      "step": 10330
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0009811746214019599,
+      "loss": 1.7731,
+      "step": 10340
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0009810924703092634,
+      "loss": 1.8146,
+      "step": 10350
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0009810101438134388,
+      "loss": 1.794,
+      "step": 10360
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0009809276419445018,
+      "loss": 1.7443,
+      "step": 10370
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.000980844964732532,
+      "loss": 1.6457,
+      "step": 10380
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0009807621122076729,
+      "loss": 1.7514,
+      "step": 10390
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.000980679084400132,
+      "loss": 1.7748,
+      "step": 10400
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0009805958813401809,
+      "loss": 1.8627,
+      "step": 10410
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0009805125030581546,
+      "loss": 1.9099,
+      "step": 10420
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0009804289495844525,
+      "loss": 1.8258,
+      "step": 10430
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0009803452209495374,
+      "loss": 1.8714,
+      "step": 10440
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0009802613171839363,
+      "loss": 1.7998,
+      "step": 10450
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0009801772383182402,
+      "loss": 1.7625,
+      "step": 10460
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0009800929843831034,
+      "loss": 1.8561,
+      "step": 10470
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0009800085554092443,
+      "loss": 1.7429,
+      "step": 10480
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0009799239514274455,
+      "loss": 1.8117,
+      "step": 10490
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0009798391724685526,
+      "loss": 1.5906,
+      "step": 10500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0009797542185634757,
+      "loss": 1.7367,
+      "step": 10510
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0009796690897431886,
+      "loss": 1.8763,
+      "step": 10520
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0009795837860387285,
+      "loss": 1.7159,
+      "step": 10530
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0009794983074811962,
+      "loss": 1.6452,
+      "step": 10540
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0009794126541017574,
+      "loss": 1.7642,
+      "step": 10550
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0009793268259316401,
+      "loss": 1.7403,
+      "step": 10560
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0009792408230021369,
+      "loss": 1.8482,
+      "step": 10570
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.000979154645344604,
+      "loss": 1.8208,
+      "step": 10580
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0009790682929904611,
+      "loss": 1.8134,
+      "step": 10590
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0009789817659711917,
+      "loss": 1.7651,
+      "step": 10600
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.000978895064318343,
+      "loss": 1.7863,
+      "step": 10610
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.000978808188063526,
+      "loss": 1.7277,
+      "step": 10620
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0009787211372384147,
+      "loss": 1.777,
+      "step": 10630
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0009786339118747478,
+      "loss": 1.7441,
+      "step": 10640
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.000978546512004327,
+      "loss": 1.9068,
+      "step": 10650
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0009784589376590175,
+      "loss": 1.7786,
+      "step": 10660
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0009783711888707485,
+      "loss": 1.7149,
+      "step": 10670
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0009782832656715123,
+      "loss": 1.7479,
+      "step": 10680
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0009781951680933657,
+      "loss": 1.7153,
+      "step": 10690
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0009781068961684282,
+      "loss": 1.8753,
+      "step": 10700
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.000978018449928883,
+      "loss": 1.7972,
+      "step": 10710
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.000977929829406977,
+      "loss": 1.6859,
+      "step": 10720
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0009778410346350209,
+      "loss": 1.8016,
+      "step": 10730
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0009777520656453886,
+      "loss": 1.6856,
+      "step": 10740
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0009776629224705175,
+      "loss": 1.7482,
+      "step": 10750
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.000977573605142909,
+      "loss": 1.7562,
+      "step": 10760
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.000977484113695127,
+      "loss": 1.7686,
+      "step": 10770
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0009773944481598,
+      "loss": 1.8105,
+      "step": 10780
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0009773046085696192,
+      "loss": 1.8326,
+      "step": 10790
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0009772145949573394,
+      "loss": 1.7717,
+      "step": 10800
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0009771244073557793,
+      "loss": 1.8094,
+      "step": 10810
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0009770340457978205,
+      "loss": 1.7605,
+      "step": 10820
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0009769435103164084,
+      "loss": 1.7474,
+      "step": 10830
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0009768528009445512,
+      "loss": 1.7731,
+      "step": 10840
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0009767619177153215,
+      "loss": 1.8068,
+      "step": 10850
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0009766708606618543,
+      "loss": 1.8979,
+      "step": 10860
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0009765796298173487,
+      "loss": 1.7847,
+      "step": 10870
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0009764882252150666,
+      "loss": 1.8401,
+      "step": 10880
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0009763966468883338,
+      "loss": 1.5723,
+      "step": 10890
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0009763048948705389,
+      "loss": 1.7514,
+      "step": 10900
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0009762129691951343,
+      "loss": 1.7905,
+      "step": 10910
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0009761208698956352,
+      "loss": 1.907,
+      "step": 10920
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0009760285970056205,
+      "loss": 1.8021,
+      "step": 10930
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0009759361505587324,
+      "loss": 1.6455,
+      "step": 10940
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0009758435305886763,
+      "loss": 1.8394,
+      "step": 10950
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0009757507371292207,
+      "loss": 1.7937,
+      "step": 10960
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0009756577702141973,
+      "loss": 1.7578,
+      "step": 10970
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0009755646298775016,
+      "loss": 1.7708,
+      "step": 10980
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0009754713161530919,
+      "loss": 1.8778,
+      "step": 10990
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0009753778290749895,
+      "loss": 1.843,
+      "step": 11000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0009752841686772793,
+      "loss": 1.8345,
+      "step": 11010
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0009751903349941094,
+      "loss": 1.6698,
+      "step": 11020
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0009750963280596907,
+      "loss": 1.7422,
+      "step": 11030
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0009750021479082977,
+      "loss": 1.8949,
+      "step": 11040
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0009749077945742677,
+      "loss": 1.806,
+      "step": 11050
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0009748132680920014,
+      "loss": 1.7548,
+      "step": 11060
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0009747185684959625,
+      "loss": 1.795,
+      "step": 11070
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0009746236958206779,
+      "loss": 1.7781,
+      "step": 11080
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0009745286501007376,
+      "loss": 1.787,
+      "step": 11090
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0009744334313707944,
+      "loss": 1.8119,
+      "step": 11100
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0009743380396655646,
+      "loss": 1.7497,
+      "step": 11110
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0009742424750198273,
+      "loss": 1.7348,
+      "step": 11120
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0009741467374684247,
+      "loss": 1.7012,
+      "step": 11130
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0009740508270462625,
+      "loss": 1.816,
+      "step": 11140
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0009739547437883084,
+      "loss": 1.7649,
+      "step": 11150
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.000973858487729594,
+      "loss": 1.8383,
+      "step": 11160
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0009737620589052136,
+      "loss": 1.7866,
+      "step": 11170
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0009736654573503247,
+      "loss": 1.7176,
+      "step": 11180
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0009735686831001473,
+      "loss": 1.7426,
+      "step": 11190
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0009734717361899646,
+      "loss": 1.7987,
+      "step": 11200
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0009733746166551232,
+      "loss": 1.7344,
+      "step": 11210
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0009732773245310318,
+      "loss": 1.809,
+      "step": 11220
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0009731798598531628,
+      "loss": 1.7074,
+      "step": 11230
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0009730822226570509,
+      "loss": 1.6944,
+      "step": 11240
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0009729844129782942,
+      "loss": 1.7243,
+      "step": 11250
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0009728864308525534,
+      "loss": 1.7516,
+      "step": 11260
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.000972788276315552,
+      "loss": 1.7105,
+      "step": 11270
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0009726899494030768,
+      "loss": 1.6982,
+      "step": 11280
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0009725914501509767,
+      "loss": 1.7656,
+      "step": 11290
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0009724927785951643,
+      "loss": 1.8973,
+      "step": 11300
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0009723939347716143,
+      "loss": 1.9031,
+      "step": 11310
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0009722949187163646,
+      "loss": 1.8425,
+      "step": 11320
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0009721957304655159,
+      "loss": 1.8458,
+      "step": 11330
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0009720963700552315,
+      "loss": 1.7749,
+      "step": 11340
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0009719968375217374,
+      "loss": 1.9048,
+      "step": 11350
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0009718971329013229,
+      "loss": 1.7444,
+      "step": 11360
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0009717972562303391,
+      "loss": 1.6944,
+      "step": 11370
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0009716972075452009,
+      "loss": 1.7349,
+      "step": 11380
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0009715969868823849,
+      "loss": 1.8686,
+      "step": 11390
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0009714965942784313,
+      "loss": 1.6679,
+      "step": 11400
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0009713960297699424,
+      "loss": 1.7048,
+      "step": 11410
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0009712952933935832,
+      "loss": 1.8649,
+      "step": 11420
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0009711943851860819,
+      "loss": 1.7515,
+      "step": 11430
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0009710933051842288,
+      "loss": 1.9001,
+      "step": 11440
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0009709920534248769,
+      "loss": 1.8789,
+      "step": 11450
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0009708906299449418,
+      "loss": 1.9457,
+      "step": 11460
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0009707890347814022,
+      "loss": 1.8666,
+      "step": 11470
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0009706872679712986,
+      "loss": 1.7263,
+      "step": 11480
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0009705853295517348,
+      "loss": 1.8461,
+      "step": 11490
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0009704832195598769,
+      "loss": 1.7918,
+      "step": 11500
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0009703809380329534,
+      "loss": 1.8426,
+      "step": 11510
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0009702784850082553,
+      "loss": 1.7251,
+      "step": 11520
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0009701758605231368,
+      "loss": 1.6913,
+      "step": 11530
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0009700730646150135,
+      "loss": 1.5939,
+      "step": 11540
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0009699700973213645,
+      "loss": 1.8592,
+      "step": 11550
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0009698669586797311,
+      "loss": 1.7984,
+      "step": 11560
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": {
+        "accuracy": 0.3014267185473411
+      },
+      "eval_f1": {
+        "f1": 0.2074421859419775
+      },
+      "eval_loss": 1.7321857213974,
+      "eval_precision": {
+        "precision": 0.17325128260235972
+      },
+      "eval_recall": {
+        "recall": 0.31234162186318526
+      },
+      "eval_runtime": 103.9228,
+      "eval_samples_per_second": 37.095,
+      "eval_steps_per_second": 18.552,
+      "step": 11562
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0009697636487277166,
+      "loss": 1.7792,
+      "step": 11570
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0009696601675029873,
+      "loss": 1.6447,
+      "step": 11580
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0009695565150432717,
+      "loss": 1.8134,
+      "step": 11590
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0009694526913863608,
+      "loss": 1.7022,
+      "step": 11600
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0009693486965701082,
+      "loss": 1.8088,
+      "step": 11610
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0009692445306324296,
+      "loss": 1.852,
+      "step": 11620
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0009691401936113032,
+      "loss": 1.8435,
+      "step": 11630
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0009690356855447695,
+      "loss": 1.8285,
+      "step": 11640
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0009689310064709316,
+      "loss": 1.7503,
+      "step": 11650
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0009688261564279548,
+      "loss": 1.8343,
+      "step": 11660
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0009687211354540664,
+      "loss": 1.7397,
+      "step": 11670
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0009686159435875569,
+      "loss": 1.8581,
+      "step": 11680
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0009685105808667782,
+      "loss": 1.9103,
+      "step": 11690
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0009684050473301449,
+      "loss": 1.7767,
+      "step": 11700
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0009682993430161339,
+      "loss": 1.7691,
+      "step": 11710
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0009681934679632842,
+      "loss": 1.8959,
+      "step": 11720
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0009680874222101972,
+      "loss": 1.7943,
+      "step": 11730
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0009679812057955365,
+      "loss": 1.8252,
+      "step": 11740
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0009678748187580278,
+      "loss": 1.8658,
+      "step": 11750
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0009677682611364593,
+      "loss": 1.9736,
+      "step": 11760
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.000967661532969681,
+      "loss": 1.7876,
+      "step": 11770
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0009675546342966054,
+      "loss": 1.7609,
+      "step": 11780
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0009674475651562071,
+      "loss": 1.8479,
+      "step": 11790
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0009673403255875225,
+      "loss": 1.756,
+      "step": 11800
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0009672329156296508,
+      "loss": 1.7295,
+      "step": 11810
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0009671253353217528,
+      "loss": 2.0492,
+      "step": 11820
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0009670175847030515,
+      "loss": 1.7373,
+      "step": 11830
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0009669096638128322,
+      "loss": 1.9517,
+      "step": 11840
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.000966801572690442,
+      "loss": 1.7751,
+      "step": 11850
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0009666933113752905,
+      "loss": 1.8541,
+      "step": 11860
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0009665848799068486,
+      "loss": 1.8422,
+      "step": 11870
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0009664762783246501,
+      "loss": 1.7967,
+      "step": 11880
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0009663675066682902,
+      "loss": 1.761,
+      "step": 11890
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0009662585649774264,
+      "loss": 1.7239,
+      "step": 11900
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.000966149453291778,
+      "loss": 1.8602,
+      "step": 11910
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0009660401716511266,
+      "loss": 1.7893,
+      "step": 11920
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0009659307200953156,
+      "loss": 1.7777,
+      "step": 11930
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0009658210986642499,
+      "loss": 1.8514,
+      "step": 11940
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0009657113073978972,
+      "loss": 1.7697,
+      "step": 11950
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0009656013463362865,
+      "loss": 1.7849,
+      "step": 11960
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0009654912155195088,
+      "loss": 1.7234,
+      "step": 11970
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0009653809149877172,
+      "loss": 1.7891,
+      "step": 11980
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0009652704447811265,
+      "loss": 1.6848,
+      "step": 11990
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0009651598049400132,
+      "loss": 1.8787,
+      "step": 12000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0009650489955047161,
+      "loss": 1.8163,
+      "step": 12010
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0009649380165156356,
+      "loss": 1.6971,
+      "step": 12020
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0009648268680132338,
+      "loss": 1.7703,
+      "step": 12030
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0009647155500380348,
+      "loss": 1.8627,
+      "step": 12040
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0009646040626306243,
+      "loss": 1.831,
+      "step": 12050
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0009644924058316498,
+      "loss": 1.6914,
+      "step": 12060
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0009643805796818208,
+      "loss": 1.8321,
+      "step": 12070
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0009642685842219084,
+      "loss": 1.7376,
+      "step": 12080
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0009641564194927451,
+      "loss": 1.7676,
+      "step": 12090
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0009640440855352257,
+      "loss": 1.88,
+      "step": 12100
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0009639315823903065,
+      "loss": 1.7956,
+      "step": 12110
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.000963818910099005,
+      "loss": 1.8877,
+      "step": 12120
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0009637060687024012,
+      "loss": 1.8099,
+      "step": 12130
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.000963593058241636,
+      "loss": 1.7973,
+      "step": 12140
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0009634798787579124,
+      "loss": 1.7283,
+      "step": 12150
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.000963366530292495,
+      "loss": 1.8744,
+      "step": 12160
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0009632530128867099,
+      "loss": 1.7395,
+      "step": 12170
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0009631393265819444,
+      "loss": 1.7538,
+      "step": 12180
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0009630254714196483,
+      "loss": 1.8393,
+      "step": 12190
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.000962911447441332,
+      "loss": 1.7348,
+      "step": 12200
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0009627972546885682,
+      "loss": 1.8174,
+      "step": 12210
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0009626828932029907,
+      "loss": 1.7659,
+      "step": 12220
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0009625683630262951,
+      "loss": 1.7505,
+      "step": 12230
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0009624536642002379,
+      "loss": 1.7804,
+      "step": 12240
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0009623387967666382,
+      "loss": 1.71,
+      "step": 12250
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0009622237607673752,
+      "loss": 1.6895,
+      "step": 12260
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0009621085562443908,
+      "loss": 1.8456,
+      "step": 12270
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0009619931832396873,
+      "loss": 1.8019,
+      "step": 12280
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0009618776417953294,
+      "loss": 1.6919,
+      "step": 12290
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0009617619319534428,
+      "loss": 1.7653,
+      "step": 12300
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.000961646053756214,
+      "loss": 1.7869,
+      "step": 12310
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0009615300072458919,
+      "loss": 1.8052,
+      "step": 12320
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0009614137924647861,
+      "loss": 1.7572,
+      "step": 12330
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0009612974094552678,
+      "loss": 1.6216,
+      "step": 12340
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0009611808582597694,
+      "loss": 1.8323,
+      "step": 12350
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0009610641389207848,
+      "loss": 1.8704,
+      "step": 12360
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.000960947251480869,
+      "loss": 1.8894,
+      "step": 12370
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0009608301959826384,
+      "loss": 1.8785,
+      "step": 12380
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0009607129724687708,
+      "loss": 1.8482,
+      "step": 12390
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0009605955809820049,
+      "loss": 1.904,
+      "step": 12400
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.000960478021565141,
+      "loss": 1.8867,
+      "step": 12410
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0009603602942610403,
+      "loss": 1.907,
+      "step": 12420
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0009602423991126258,
+      "loss": 1.9181,
+      "step": 12430
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0009601243361628809,
+      "loss": 1.8593,
+      "step": 12440
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0009600061054548506,
+      "loss": 1.7883,
+      "step": 12450
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0009598877070316412,
+      "loss": 1.7691,
+      "step": 12460
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0009597691409364199,
+      "loss": 1.8437,
+      "step": 12470
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0009596504072124153,
+      "loss": 1.8041,
+      "step": 12480
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0009595315059029166,
+      "loss": 1.8488,
+      "step": 12490
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0009594124370512746,
+      "loss": 1.8085,
+      "step": 12500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0009592932007009009,
+      "loss": 1.7688,
+      "step": 12510
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0009591737968952685,
+      "loss": 1.8884,
+      "step": 12520
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0009590542256779111,
+      "loss": 1.9053,
+      "step": 12530
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0009589344870924237,
+      "loss": 1.9115,
+      "step": 12540
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.000958814581182462,
+      "loss": 1.8114,
+      "step": 12550
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0009586945079917429,
+      "loss": 1.8891,
+      "step": 12560
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0009585742675640445,
+      "loss": 1.8384,
+      "step": 12570
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0009584538599432056,
+      "loss": 1.859,
+      "step": 12580
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0009583332851731258,
+      "loss": 1.7339,
+      "step": 12590
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0009582125432977661,
+      "loss": 1.8824,
+      "step": 12600
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0009580916343611481,
+      "loss": 1.751,
+      "step": 12610
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0009579705584073544,
+      "loss": 1.6991,
+      "step": 12620
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0009578493154805285,
+      "loss": 1.7719,
+      "step": 12630
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0009577279056248749,
+      "loss": 1.8559,
+      "step": 12640
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0009576063288846585,
+      "loss": 1.7681,
+      "step": 12650
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0009574845853042057,
+      "loss": 1.7717,
+      "step": 12660
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0009573626749279032,
+      "loss": 1.8828,
+      "step": 12670
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0009572405978001988,
+      "loss": 1.7839,
+      "step": 12680
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.000957118353965601,
+      "loss": 1.714,
+      "step": 12690
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0009569959434686792,
+      "loss": 1.8501,
+      "step": 12700
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0009568733663540634,
+      "loss": 1.7382,
+      "step": 12710
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0009567506226664441,
+      "loss": 1.8327,
+      "step": 12720
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0009566277124505733,
+      "loss": 1.8203,
+      "step": 12730
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.000956504635751263,
+      "loss": 1.7329,
+      "step": 12740
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0009563813926133862,
+      "loss": 1.7699,
+      "step": 12750
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0009562579830818765,
+      "loss": 1.8813,
+      "step": 12760
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0009561344072017283,
+      "loss": 1.9075,
+      "step": 12770
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0009560106650179964,
+      "loss": 1.7317,
+      "step": 12780
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0009558867565757965,
+      "loss": 1.6895,
+      "step": 12790
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0009557626819203047,
+      "loss": 1.6617,
+      "step": 12800
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0009556384410967578,
+      "loss": 1.7724,
+      "step": 12810
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0009555140341504534,
+      "loss": 1.8285,
+      "step": 12820
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.000955389461126749,
+      "loss": 1.8063,
+      "step": 12830
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0009552647220710636,
+      "loss": 1.725,
+      "step": 12840
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.000955139817028876,
+      "loss": 1.6705,
+      "step": 12850
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0009550147460457256,
+      "loss": 1.7254,
+      "step": 12860
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0009548895091672127,
+      "loss": 1.8107,
+      "step": 12870
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0009547641064389978,
+      "loss": 1.6768,
+      "step": 12880
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0009546385379068018,
+      "loss": 1.7145,
+      "step": 12890
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0009545128036164064,
+      "loss": 1.7159,
+      "step": 12900
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.000954386903613653,
+      "loss": 1.7503,
+      "step": 12910
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0009542608379444444,
+      "loss": 1.7583,
+      "step": 12920
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.000954134606654743,
+      "loss": 1.8885,
+      "step": 12930
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.000954008209790572,
+      "loss": 1.6543,
+      "step": 12940
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0009538816473980149,
+      "loss": 1.8198,
+      "step": 12950
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0009537549195232156,
+      "loss": 1.7566,
+      "step": 12960
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.000953628026212378,
+      "loss": 1.7466,
+      "step": 12970
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0009535009675117665,
+      "loss": 1.6681,
+      "step": 12980
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0009533737434677063,
+      "loss": 1.8596,
+      "step": 12990
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0009532463541265822,
+      "loss": 1.8017,
+      "step": 13000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0009531187995348394,
+      "loss": 1.7382,
+      "step": 13010
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0009529910797389835,
+      "loss": 1.7205,
+      "step": 13020
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0009528631947855803,
+      "loss": 1.9236,
+      "step": 13030
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.000952735144721256,
+      "loss": 1.596,
+      "step": 13040
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0009526069295926966,
+      "loss": 1.8309,
+      "step": 13050
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0009524785494466485,
+      "loss": 1.7308,
+      "step": 13060
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0009523500043299184,
+      "loss": 1.7496,
+      "step": 13070
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0009522212942893728,
+      "loss": 1.639,
+      "step": 13080
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0009520924193719387,
+      "loss": 1.8481,
+      "step": 13090
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0009519633796246028,
+      "loss": 1.5754,
+      "step": 13100
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0009518341750944124,
+      "loss": 1.91,
+      "step": 13110
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0009517048058284745,
+      "loss": 1.7873,
+      "step": 13120
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0009515752718739564,
+      "loss": 1.9317,
+      "step": 13130
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0009514455732780851,
+      "loss": 1.8516,
+      "step": 13140
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.000951315710088148,
+      "loss": 1.8624,
+      "step": 13150
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0009511856823514923,
+      "loss": 1.7967,
+      "step": 13160
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0009510554901155252,
+      "loss": 1.6538,
+      "step": 13170
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0009509251334277141,
+      "loss": 1.7719,
+      "step": 13180
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.000950794612335586,
+      "loss": 1.7984,
+      "step": 13190
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0009506639268867283,
+      "loss": 1.9263,
+      "step": 13200
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0009505330771287877,
+      "loss": 1.838,
+      "step": 13210
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0009504020631094715,
+      "loss": 1.7548,
+      "step": 13220
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.000950270884876546,
+      "loss": 1.8724,
+      "step": 13230
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0009501395424778385,
+      "loss": 1.7411,
+      "step": 13240
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0009500080359612353,
+      "loss": 1.7446,
+      "step": 13250
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0009498763653746829,
+      "loss": 1.7895,
+      "step": 13260
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0009497445307661875,
+      "loss": 1.6295,
+      "step": 13270
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.000949612532183815,
+      "loss": 1.6633,
+      "step": 13280
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0009494803696756913,
+      "loss": 2.1007,
+      "step": 13290
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0009493480432900021,
+      "loss": 1.7611,
+      "step": 13300
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0009492155530749925,
+      "loss": 1.847,
+      "step": 13310
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0009490828990789677,
+      "loss": 1.7704,
+      "step": 13320
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0009489500813502923,
+      "loss": 1.7631,
+      "step": 13330
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0009488170999373911,
+      "loss": 1.8335,
+      "step": 13340
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0009486839548887479,
+      "loss": 1.852,
+      "step": 13350
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0009485506462529067,
+      "loss": 1.609,
+      "step": 13360
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0009484171740784708,
+      "loss": 1.7785,
+      "step": 13370
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0009482835384141034,
+      "loss": 1.7499,
+      "step": 13380
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0009481497393085272,
+      "loss": 1.8266,
+      "step": 13390
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0009480157768105242,
+      "loss": 1.7917,
+      "step": 13400
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0009478816509689367,
+      "loss": 1.753,
+      "step": 13410
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0009477473618326656,
+      "loss": 1.7479,
+      "step": 13420
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0009476129094506722,
+      "loss": 1.7644,
+      "step": 13430
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0009474782938719769,
+      "loss": 1.6127,
+      "step": 13440
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0009473435151456593,
+      "loss": 1.817,
+      "step": 13450
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0009472085733208592,
+      "loss": 1.6814,
+      "step": 13460
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0009470734684467754,
+      "loss": 1.6956,
+      "step": 13470
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0009469382005726661,
+      "loss": 1.7385,
+      "step": 13480
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": {
+        "accuracy": 0.3120622568093385
+      },
+      "eval_f1": {
+        "f1": 0.2220020856519651
+      },
+      "eval_loss": 1.6964426040649414,
+      "eval_precision": {
+        "precision": 0.24821744277779434
+      },
+      "eval_recall": {
+        "recall": 0.3239827067963708
+      },
+      "eval_runtime": 102.2041,
+      "eval_samples_per_second": 37.719,
+      "eval_steps_per_second": 18.864,
+      "step": 13489
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0009468027697478494,
+      "loss": 1.867,
+      "step": 13490
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0009466671760217023,
+      "loss": 1.7677,
+      "step": 13500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0009465314194436611,
+      "loss": 1.8369,
+      "step": 13510
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0009463955000632222,
+      "loss": 1.8289,
+      "step": 13520
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0009462594179299406,
+      "loss": 1.819,
+      "step": 13530
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0009461231730934312,
+      "loss": 1.8158,
+      "step": 13540
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0009459867656033676,
+      "loss": 1.7489,
+      "step": 13550
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0009458501955094835,
+      "loss": 1.7089,
+      "step": 13560
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0009457134628615712,
+      "loss": 1.9333,
+      "step": 13570
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0009455765677094825,
+      "loss": 1.8292,
+      "step": 13580
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0009454395101031283,
+      "loss": 1.8648,
+      "step": 13590
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0009453022900924795,
+      "loss": 1.7783,
+      "step": 13600
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0009451649077275651,
+      "loss": 1.7176,
+      "step": 13610
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0009450273630584738,
+      "loss": 1.6828,
+      "step": 13620
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0009448896561353536,
+      "loss": 1.8494,
+      "step": 13630
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.0009447517870084116,
+      "loss": 1.7528,
+      "step": 13640
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.000944613755727914,
+      "loss": 1.7617,
+      "step": 13650
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0009444755623441858,
+      "loss": 1.8195,
+      "step": 13660
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0009443372069076118,
+      "loss": 1.8131,
+      "step": 13670
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.000944198689468635,
+      "loss": 1.6968,
+      "step": 13680
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0009440600100777583,
+      "loss": 1.7565,
+      "step": 13690
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.0009439211687855433,
+      "loss": 1.7608,
+      "step": 13700
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.0009437821656426104,
+      "loss": 1.7052,
+      "step": 13710
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0009436430006996392,
+      "loss": 1.8568,
+      "step": 13720
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0009435036740073687,
+      "loss": 1.9427,
+      "step": 13730
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0009433641856165961,
+      "loss": 1.7566,
+      "step": 13740
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0009432245355781782,
+      "loss": 1.7976,
+      "step": 13750
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0009430847239430302,
+      "loss": 1.7797,
+      "step": 13760
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0009429447507621268,
+      "loss": 1.6913,
+      "step": 13770
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0009428046160865009,
+      "loss": 1.7375,
+      "step": 13780
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0009426643199672451,
+      "loss": 1.7487,
+      "step": 13790
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0009425238624555101,
+      "loss": 1.7797,
+      "step": 13800
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0009423832436025062,
+      "loss": 1.9005,
+      "step": 13810
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0009422424634595017,
+      "loss": 1.7843,
+      "step": 13820
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0009421015220778242,
+      "loss": 1.7961,
+      "step": 13830
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00094196041950886,
+      "loss": 1.741,
+      "step": 13840
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.0009418191558040543,
+      "loss": 1.724,
+      "step": 13850
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.000941677731014911,
+      "loss": 1.7876,
+      "step": 13860
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0009415361451929922,
+      "loss": 1.7415,
+      "step": 13870
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0009413943983899194,
+      "loss": 1.6685,
+      "step": 13880
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0009412524906573726,
+      "loss": 1.6868,
+      "step": 13890
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0009411104220470903,
+      "loss": 1.8981,
+      "step": 13900
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00094096819261087,
+      "loss": 1.7498,
+      "step": 13910
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.0009408258024005674,
+      "loss": 1.8692,
+      "step": 13920
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0009406832514680971,
+      "loss": 1.8906,
+      "step": 13930
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0009405405398654321,
+      "loss": 1.7531,
+      "step": 13940
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0009403976676446044,
+      "loss": 1.7559,
+      "step": 13950
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0009402546348577039,
+      "loss": 1.7564,
+      "step": 13960
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0009401114415568797,
+      "loss": 1.8532,
+      "step": 13970
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0009399680877943389,
+      "loss": 1.8019,
+      "step": 13980
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0009398245736223476,
+      "loss": 1.8442,
+      "step": 13990
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0009396808990932298,
+      "loss": 1.8377,
+      "step": 14000
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0009395370642593686,
+      "loss": 1.8058,
+      "step": 14010
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.000939393069173205,
+      "loss": 1.6893,
+      "step": 14020
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0009392489138872386,
+      "loss": 1.7234,
+      "step": 14030
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0009391045984540278,
+      "loss": 1.7425,
+      "step": 14040
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.0009389601229261888,
+      "loss": 1.7706,
+      "step": 14050
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0009388154873563964,
+      "loss": 1.8947,
+      "step": 14060
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.000938670691797384,
+      "loss": 1.8376,
+      "step": 14070
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0009385257363019428,
+      "loss": 1.6913,
+      "step": 14080
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0009383806209229226,
+      "loss": 1.7429,
+      "step": 14090
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0009382353457132319,
+      "loss": 1.7054,
+      "step": 14100
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0009380899107258367,
+      "loss": 1.7288,
+      "step": 14110
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0009379443160137617,
+      "loss": 1.7682,
+      "step": 14120
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00093779856163009,
+      "loss": 1.7532,
+      "step": 14130
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0009376526476279625,
+      "loss": 1.6605,
+      "step": 14140
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0009375065740605785,
+      "loss": 1.8331,
+      "step": 14150
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0009373603409811954,
+      "loss": 1.7324,
+      "step": 14160
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0009372139484431288,
+      "loss": 1.7655,
+      "step": 14170
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0009370673964997527,
+      "loss": 1.8016,
+      "step": 14180
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0009369206852044985,
+      "loss": 1.7753,
+      "step": 14190
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0009367738146108566,
+      "loss": 1.6695,
+      "step": 14200
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0009366267847723751,
+      "loss": 1.7622,
+      "step": 14210
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0009364795957426596,
+      "loss": 1.6401,
+      "step": 14220
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0009363322475753746,
+      "loss": 1.7551,
+      "step": 14230
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0009361847403242423,
+      "loss": 1.78,
+      "step": 14240
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0009360370740430428,
+      "loss": 1.7698,
+      "step": 14250
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0009358892487856142,
+      "loss": 1.7255,
+      "step": 14260
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0009357412646058526,
+      "loss": 1.7899,
+      "step": 14270
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.000935593121557712,
+      "loss": 1.717,
+      "step": 14280
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0009354448196952045,
+      "loss": 1.7246,
+      "step": 14290
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0009352963590723999,
+      "loss": 1.7834,
+      "step": 14300
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0009351477397434261,
+      "loss": 1.9131,
+      "step": 14310
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0009349989617624685,
+      "loss": 1.6523,
+      "step": 14320
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.000934850025183771,
+      "loss": 1.771,
+      "step": 14330
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.0009347009300616343,
+      "loss": 1.6129,
+      "step": 14340
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0009345516764504179,
+      "loss": 1.7177,
+      "step": 14350
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0009344022644045384,
+      "loss": 1.7367,
+      "step": 14360
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0009342526939784706,
+      "loss": 1.6732,
+      "step": 14370
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.000934102965226747,
+      "loss": 1.6573,
+      "step": 14380
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.0009339530782039575,
+      "loss": 1.7023,
+      "step": 14390
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00093380303296475,
+      "loss": 1.7721,
+      "step": 14400
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00093365282956383,
+      "loss": 1.8415,
+      "step": 14410
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0009335024680559608,
+      "loss": 1.7607,
+      "step": 14420
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.000933351948495963,
+      "loss": 1.8485,
+      "step": 14430
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.000933201270938715,
+      "loss": 1.7508,
+      "step": 14440
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0009330504354391531,
+      "loss": 1.8016,
+      "step": 14450
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0009328994420522707,
+      "loss": 1.7386,
+      "step": 14460
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.000932748290833119,
+      "loss": 1.8136,
+      "step": 14470
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.000932596981836807,
+      "loss": 1.7424,
+      "step": 14480
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0009324455151185008,
+      "loss": 1.6428,
+      "step": 14490
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0009322938907334238,
+      "loss": 1.7389,
+      "step": 14500
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0009321421087368578,
+      "loss": 1.6418,
+      "step": 14510
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0009319901691841413,
+      "loss": 1.774,
+      "step": 14520
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0009318380721306703,
+      "loss": 1.7014,
+      "step": 14530
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0009316858176318985,
+      "loss": 1.6742,
+      "step": 14540
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.000931533405743337,
+      "loss": 1.6953,
+      "step": 14550
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.000931380836520554,
+      "loss": 1.6567,
+      "step": 14560
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0009312281100191752,
+      "loss": 1.7334,
+      "step": 14570
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0009310752262948838,
+      "loss": 1.8719,
+      "step": 14580
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0009309221854034202,
+      "loss": 1.8223,
+      "step": 14590
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.000930768987400582,
+      "loss": 1.8442,
+      "step": 14600
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0009306156323422244,
+      "loss": 1.7612,
+      "step": 14610
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.0009304621202842593,
+      "loss": 1.8426,
+      "step": 14620
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.0009303084512826562,
+      "loss": 1.7055,
+      "step": 14630
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.0009301546253934422,
+      "loss": 1.6697,
+      "step": 14640
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.0009300006426727008,
+      "loss": 1.7494,
+      "step": 14650
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.0009298465031765732,
+      "loss": 1.6976,
+      "step": 14660
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.0009296922069612578,
+      "loss": 1.8057,
+      "step": 14670
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0009295377540830097,
+      "loss": 1.9329,
+      "step": 14680
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0009293831445981415,
+      "loss": 1.7457,
+      "step": 14690
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.000929228378563023,
+      "loss": 1.7327,
+      "step": 14700
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.0009290734560340806,
+      "loss": 1.8215,
+      "step": 14710
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.0009289183770677981,
+      "loss": 1.5827,
+      "step": 14720
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.0009287631417207163,
+      "loss": 1.54,
+      "step": 14730
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.000928607750049433,
+      "loss": 1.7683,
+      "step": 14740
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.000928452202110603,
+      "loss": 1.8053,
+      "step": 14750
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.0009282964979609379,
+      "loss": 1.8536,
+      "step": 14760
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.0009281406376572066,
+      "loss": 1.7002,
+      "step": 14770
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0009279846212562344,
+      "loss": 1.7629,
+      "step": 14780
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.0009278284488149042,
+      "loss": 1.677,
+      "step": 14790
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.0009276721203901555,
+      "loss": 1.7235,
+      "step": 14800
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0009275156360389842,
+      "loss": 1.6409,
+      "step": 14810
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0009273589958184437,
+      "loss": 1.7063,
+      "step": 14820
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.000927202199785644,
+      "loss": 1.7886,
+      "step": 14830
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.0009270452479977519,
+      "loss": 1.9833,
+      "step": 14840
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.0009268881405119909,
+      "loss": 1.7168,
+      "step": 14850
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.0009267308773856413,
+      "loss": 1.7221,
+      "step": 14860
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0009265734586760405,
+      "loss": 1.7475,
+      "step": 14870
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0009264158844405818,
+      "loss": 1.6445,
+      "step": 14880
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.0009262581547367164,
+      "loss": 1.7001,
+      "step": 14890
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.000926100269621951,
+      "loss": 1.8752,
+      "step": 14900
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.0009259422291538496,
+      "loss": 1.6439,
+      "step": 14910
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.0009257840333900325,
+      "loss": 1.7411,
+      "step": 14920
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.0009256256823881772,
+      "loss": 1.5836,
+      "step": 14930
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.0009254671762060171,
+      "loss": 1.7718,
+      "step": 14940
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0009253085149013429,
+      "loss": 1.7117,
+      "step": 14950
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0009251496985320011,
+      "loss": 1.8424,
+      "step": 14960
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.0009249907271558953,
+      "loss": 1.7725,
+      "step": 14970
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.0009248316008309854,
+      "loss": 1.6359,
+      "step": 14980
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.0009246723196152876,
+      "loss": 1.8313,
+      "step": 14990
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.000924512883566875,
+      "loss": 1.8976,
+      "step": 15000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.0009243532927438767,
+      "loss": 1.7728,
+      "step": 15010
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.0009241935472044787,
+      "loss": 1.8352,
+      "step": 15020
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.0009240336470069231,
+      "loss": 1.7677,
+      "step": 15030
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.0009238735922095083,
+      "loss": 1.9427,
+      "step": 15040
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0009237133828705893,
+      "loss": 1.8624,
+      "step": 15050
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0009235530190485776,
+      "loss": 1.8505,
+      "step": 15060
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0009233925008019405,
+      "loss": 1.827,
+      "step": 15070
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.000923231828189202,
+      "loss": 1.8163,
+      "step": 15080
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.0009230710012689421,
+      "loss": 1.8301,
+      "step": 15090
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.0009229100200997976,
+      "loss": 1.7922,
+      "step": 15100
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.0009227488847404608,
+      "loss": 1.7587,
+      "step": 15110
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.0009225875952496807,
+      "loss": 1.9375,
+      "step": 15120
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.0009224261516862625,
+      "loss": 1.8747,
+      "step": 15130
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0009222645541090673,
+      "loss": 1.9106,
+      "step": 15140
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0009221028025770126,
+      "loss": 1.8543,
+      "step": 15150
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0009219408971490719,
+      "loss": 1.8744,
+      "step": 15160
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0009217788378842748,
+      "loss": 1.8751,
+      "step": 15170
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0009216166248417072,
+      "loss": 1.7944,
+      "step": 15180
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0009214542580805108,
+      "loss": 1.8621,
+      "step": 15190
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0009212917376598832,
+      "loss": 1.8727,
+      "step": 15200
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0009211290636390787,
+      "loss": 1.8287,
+      "step": 15210
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0009209662360774071,
+      "loss": 1.8699,
+      "step": 15220
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.000920803255034234,
+      "loss": 1.8996,
+      "step": 15230
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0009206401205689811,
+      "loss": 1.8532,
+      "step": 15240
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0009204768327411267,
+      "loss": 1.8261,
+      "step": 15250
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.000920313391610204,
+      "loss": 1.8188,
+      "step": 15260
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.0009201497972358028,
+      "loss": 1.7931,
+      "step": 15270
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.0009199860496775683,
+      "loss": 1.6533,
+      "step": 15280
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.0009198221489952019,
+      "loss": 1.705,
+      "step": 15290
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0009196580952484607,
+      "loss": 1.6474,
+      "step": 15300
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0009194938884971576,
+      "loss": 1.7114,
+      "step": 15310
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0009193295288011614,
+      "loss": 1.9186,
+      "step": 15320
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.0009191650162203963,
+      "loss": 1.795,
+      "step": 15330
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.0009190003508148428,
+      "loss": 1.6633,
+      "step": 15340
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0009188355326445366,
+      "loss": 1.7603,
+      "step": 15350
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0009186705617695695,
+      "loss": 1.5689,
+      "step": 15360
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0009185054382500884,
+      "loss": 1.8437,
+      "step": 15370
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0009183401621462966,
+      "loss": 1.8561,
+      "step": 15380
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0009181747335184527,
+      "loss": 1.754,
+      "step": 15390
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0009180091524268707,
+      "loss": 1.7165,
+      "step": 15400
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0009178434189319205,
+      "loss": 1.578,
+      "step": 15410
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": {
+        "accuracy": 0.2827496757457847
+      },
+      "eval_f1": {
+        "f1": 0.20468330816120264
+      },
+      "eval_loss": 1.6264866590499878,
+      "eval_precision": {
+        "precision": 0.23605832742279897
+      },
+      "eval_recall": {
+        "recall": 0.2983260879145022
+      },
+      "eval_runtime": 106.6464,
+      "eval_samples_per_second": 36.148,
+      "eval_steps_per_second": 18.078,
+      "step": 15416
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 57810,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 500,
+  "total_flos": 7.111502715796586e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}