diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5857 @@
+{
+  "best_metric": 0.1442219316959381,
+  "best_model_checkpoint": "/leonardo_work/IscrC_AGENT/PROFES2025/results/full_size/industry/checkpoint-6434",
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 9651,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9993783027665528e-05,
+      "loss": 0.3893,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9987566055331057e-05,
+      "loss": 0.2563,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9981349082996583e-05,
+      "loss": 0.2752,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997513211066211e-05,
+      "loss": 0.1981,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9968915138327635e-05,
+      "loss": 0.3212,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9962698165993164e-05,
+      "loss": 0.3368,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.995648119365869e-05,
+      "loss": 0.1943,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9950264221324216e-05,
+      "loss": 0.2304,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944047248989746e-05,
+      "loss": 0.2129,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.993783027665527e-05,
+      "loss": 0.2646,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9931613304320797e-05,
+      "loss": 0.2544,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925396331986323e-05,
+      "loss": 0.2028,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991917935965185e-05,
+      "loss": 0.3563,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991296238731738e-05,
+      "loss": 0.2753,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906745414982905e-05,
+      "loss": 0.3856,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990052844264843e-05,
+      "loss": 0.2214,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894311470313957e-05,
+      "loss": 0.2224,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9888094497979486e-05,
+      "loss": 0.1958,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881877525645012e-05,
+      "loss": 0.2366,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875660553310538e-05,
+      "loss": 0.2545,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9869443580976067e-05,
+      "loss": 0.2205,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9863226608641593e-05,
+      "loss": 0.1819,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.985700963630712e-05,
+      "loss": 0.3102,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9850792663972645e-05,
+      "loss": 0.2572,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844575691638175e-05,
+      "loss": 0.2034,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98383587193037e-05,
+      "loss": 0.293,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9832141746969227e-05,
+      "loss": 0.1886,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825924774634756e-05,
+      "loss": 0.2712,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819707802300282e-05,
+      "loss": 0.2588,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813490829965808e-05,
+      "loss": 0.2133,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9807273857631334e-05,
+      "loss": 0.3084,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.980105688529686e-05,
+      "loss": 0.3041,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.979483991296239e-05,
+      "loss": 0.2575,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9788622940627915e-05,
+      "loss": 0.2455,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9782405968293445e-05,
+      "loss": 0.208,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.977618899595897e-05,
+      "loss": 0.2808,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9769972023624497e-05,
+      "loss": 0.2144,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9763755051290022e-05,
+      "loss": 0.1937,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.975753807895555e-05,
+      "loss": 0.2377,
+      "step": 390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9751321106621078e-05,
+      "loss": 0.191,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9745104134286604e-05,
+      "loss": 0.1982,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9738887161952133e-05,
+      "loss": 0.2371,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.973267018961766e-05,
+      "loss": 0.2327,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9726453217283185e-05,
+      "loss": 0.1661,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.972023624494871e-05,
+      "loss": 0.269,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9714019272614237e-05,
+      "loss": 0.2122,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9707802300279766e-05,
+      "loss": 0.2669,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9701585327945292e-05,
+      "loss": 0.1997,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.969536835561082e-05,
+      "loss": 0.1665,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9689151383276344e-05,
+      "loss": 0.1998,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.968293441094187e-05,
+      "loss": 0.1435,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.96767174386074e-05,
+      "loss": 0.253,
+      "step": 520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9670500466272926e-05,
+      "loss": 0.3009,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9664283493938455e-05,
+      "loss": 0.2452,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.965806652160398e-05,
+      "loss": 0.2036,
+      "step": 550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9651849549269507e-05,
+      "loss": 0.3045,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9645632576935033e-05,
+      "loss": 0.1715,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.963941560460056e-05,
+      "loss": 0.2188,
+      "step": 580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9633198632266088e-05,
+      "loss": 0.221,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9626981659931614e-05,
+      "loss": 0.2364,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9620764687597144e-05,
+      "loss": 0.132,
+      "step": 610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.961454771526267e-05,
+      "loss": 0.3427,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9608330742928196e-05,
+      "loss": 0.1321,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.960211377059372e-05,
+      "loss": 0.283,
+      "step": 640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9595896798259247e-05,
+      "loss": 0.2523,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9589679825924777e-05,
+      "loss": 0.178,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9583462853590303e-05,
+      "loss": 0.323,
+      "step": 670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9577245881255832e-05,
+      "loss": 0.1116,
+      "step": 680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9571028908921358e-05,
+      "loss": 0.1701,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9564811936586884e-05,
+      "loss": 0.1643,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.955859496425241e-05,
+      "loss": 0.3486,
+      "step": 710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9552377991917936e-05,
+      "loss": 0.2046,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9546161019583465e-05,
+      "loss": 0.1578,
+      "step": 730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.953994404724899e-05,
+      "loss": 0.155,
+      "step": 740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.953372707491452e-05,
+      "loss": 0.2874,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9527510102580047e-05,
+      "loss": 0.1739,
+      "step": 760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9521293130245573e-05,
+      "loss": 0.1395,
+      "step": 770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95150761579111e-05,
+      "loss": 0.1622,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508859185576625e-05,
+      "loss": 0.2987,
+      "step": 790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9502642213242154e-05,
+      "loss": 0.1992,
+      "step": 800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949642524090768e-05,
+      "loss": 0.1582,
+      "step": 810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9490208268573206e-05,
+      "loss": 0.2756,
+      "step": 820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9483991296238732e-05,
+      "loss": 0.1518,
+      "step": 830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9477774323904258e-05,
+      "loss": 0.1327,
+      "step": 840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9471557351569787e-05,
+      "loss": 0.2362,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9465340379235313e-05,
+      "loss": 0.2683,
+      "step": 860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9459123406900843e-05,
+      "loss": 0.2734,
+      "step": 870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945290643456637e-05,
+      "loss": 0.2063,
+      "step": 880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9446689462231895e-05,
+      "loss": 0.2737,
+      "step": 890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.944047248989742e-05,
+      "loss": 0.1777,
+      "step": 900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9434255517562946e-05,
+      "loss": 0.1622,
+      "step": 910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9428038545228476e-05,
+      "loss": 0.2168,
+      "step": 920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9421821572894002e-05,
+      "loss": 0.2234,
+      "step": 930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.941560460055953e-05,
+      "loss": 0.1816,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9409387628225057e-05,
+      "loss": 0.1772,
+      "step": 950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9403170655890583e-05,
+      "loss": 0.2137,
+      "step": 960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.939695368355611e-05,
+      "loss": 0.1457,
+      "step": 970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9390736711221635e-05,
+      "loss": 0.2507,
+      "step": 980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9384519738887164e-05,
+      "loss": 0.1794,
+      "step": 990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.937830276655269e-05,
+      "loss": 0.2282,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.937208579421822e-05,
+      "loss": 0.2025,
+      "step": 1010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9365868821883746e-05,
+      "loss": 0.1462,
+      "step": 1020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.935965184954927e-05,
+      "loss": 0.1568,
+      "step": 1030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9353434877214798e-05,
+      "loss": 0.2411,
+      "step": 1040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9347217904880324e-05,
+      "loss": 0.1666,
+      "step": 1050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9341000932545853e-05,
+      "loss": 0.2755,
+      "step": 1060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.933478396021138e-05,
+      "loss": 0.1768,
+      "step": 1070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9328566987876905e-05,
+      "loss": 0.1834,
+      "step": 1080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9322350015542434e-05,
+      "loss": 0.2204,
+      "step": 1090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9316133043207957e-05,
+      "loss": 0.218,
+      "step": 1100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9309916070873486e-05,
+      "loss": 0.1357,
+      "step": 1110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9303699098539012e-05,
+      "loss": 0.224,
+      "step": 1120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.929748212620454e-05,
+      "loss": 0.2346,
+      "step": 1130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9291265153870068e-05,
+      "loss": 0.1351,
+      "step": 1140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9285048181535594e-05,
+      "loss": 0.184,
+      "step": 1150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.927883120920112e-05,
+      "loss": 0.2442,
+      "step": 1160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9272614236866645e-05,
+      "loss": 0.1969,
+      "step": 1170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9266397264532175e-05,
+      "loss": 0.2406,
+      "step": 1180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.92601802921977e-05,
+      "loss": 0.1033,
+      "step": 1190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.925396331986323e-05,
+      "loss": 0.2369,
+      "step": 1200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9247746347528756e-05,
+      "loss": 0.1821,
+      "step": 1210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9241529375194282e-05,
+      "loss": 0.1517,
+      "step": 1220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9235312402859808e-05,
+      "loss": 0.1349,
+      "step": 1230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9229095430525334e-05,
+      "loss": 0.2878,
+      "step": 1240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9222878458190863e-05,
+      "loss": 0.248,
+      "step": 1250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.921666148585639e-05,
+      "loss": 0.2228,
+      "step": 1260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9210444513521915e-05,
+      "loss": 0.151,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9204227541187445e-05,
+      "loss": 0.1618,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.919801056885297e-05,
+      "loss": 0.2238,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9191793596518497e-05,
+      "loss": 0.2239,
+      "step": 1300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9185576624184023e-05,
+      "loss": 0.1518,
+      "step": 1310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9179359651849552e-05,
+      "loss": 0.1446,
+      "step": 1320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9173142679515078e-05,
+      "loss": 0.1952,
+      "step": 1330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9166925707180604e-05,
+      "loss": 0.2233,
+      "step": 1340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9160708734846133e-05,
+      "loss": 0.1755,
+      "step": 1350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.915449176251166e-05,
+      "loss": 0.2177,
+      "step": 1360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9148274790177185e-05,
+      "loss": 0.2145,
+      "step": 1370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.914205781784271e-05,
+      "loss": 0.1288,
+      "step": 1380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.913584084550824e-05,
+      "loss": 0.255,
+      "step": 1390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9129623873173767e-05,
+      "loss": 0.1894,
+      "step": 1400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9123406900839293e-05,
+      "loss": 0.2308,
+      "step": 1410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.911718992850482e-05,
+      "loss": 0.2879,
+      "step": 1420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9110972956170344e-05,
+      "loss": 0.1954,
+      "step": 1430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9104755983835874e-05,
+      "loss": 0.16,
+      "step": 1440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.90985390115014e-05,
+      "loss": 0.1336,
+      "step": 1450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9092322039166926e-05,
+      "loss": 0.2585,
+      "step": 1460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9086105066832455e-05,
+      "loss": 0.184,
+      "step": 1470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.907988809449798e-05,
+      "loss": 0.1759,
+      "step": 1480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9073671122163507e-05,
+      "loss": 0.2084,
+      "step": 1490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9067454149829033e-05,
+      "loss": 0.2503,
+      "step": 1500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9061237177494562e-05,
+      "loss": 0.1733,
+      "step": 1510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.905502020516009e-05,
+      "loss": 0.1497,
+      "step": 1520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9048803232825614e-05,
+      "loss": 0.2807,
+      "step": 1530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9042586260491144e-05,
+      "loss": 0.2014,
+      "step": 1540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.903636928815667e-05,
+      "loss": 0.188,
+      "step": 1550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9030152315822196e-05,
+      "loss": 0.1558,
+      "step": 1560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.902393534348772e-05,
+      "loss": 0.2582,
+      "step": 1570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.901771837115325e-05,
+      "loss": 0.1809,
+      "step": 1580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9011501398818777e-05,
+      "loss": 0.1817,
+      "step": 1590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9005284426484303e-05,
+      "loss": 0.2176,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8999067454149832e-05,
+      "loss": 0.2987,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8992850481815358e-05,
+      "loss": 0.1145,
+      "step": 1620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986633509480884e-05,
+      "loss": 0.1415,
+      "step": 1630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898041653714641e-05,
+      "loss": 0.0793,
+      "step": 1640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974199564811936e-05,
+      "loss": 0.2155,
+      "step": 1650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8967982592477466e-05,
+      "loss": 0.1325,
+      "step": 1660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896176562014299e-05,
+      "loss": 0.2684,
+      "step": 1670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895554864780852e-05,
+      "loss": 0.1618,
+      "step": 1680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8949331675474047e-05,
+      "loss": 0.2058,
+      "step": 1690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8943114703139573e-05,
+      "loss": 0.1678,
+      "step": 1700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.89368977308051e-05,
+      "loss": 0.1904,
+      "step": 1710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8930680758470625e-05,
+      "loss": 0.1515,
+      "step": 1720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8924463786136154e-05,
+      "loss": 0.1311,
+      "step": 1730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.891824681380168e-05,
+      "loss": 0.2086,
+      "step": 1740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8912029841467206e-05,
+      "loss": 0.205,
+      "step": 1750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8905812869132732e-05,
+      "loss": 0.2312,
+      "step": 1760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.889959589679826e-05,
+      "loss": 0.1127,
+      "step": 1770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8893378924463787e-05,
+      "loss": 0.1162,
+      "step": 1780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8887161952129313e-05,
+      "loss": 0.1994,
+      "step": 1790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8880944979794843e-05,
+      "loss": 0.2825,
+      "step": 1800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.887472800746037e-05,
+      "loss": 0.1908,
+      "step": 1810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8868511035125895e-05,
+      "loss": 0.1636,
+      "step": 1820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.886229406279142e-05,
+      "loss": 0.169,
+      "step": 1830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.885607709045695e-05,
+      "loss": 0.2464,
+      "step": 1840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8849860118122476e-05,
+      "loss": 0.1444,
+      "step": 1850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8843643145788002e-05,
+      "loss": 0.2067,
+      "step": 1860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.883742617345353e-05,
+      "loss": 0.2296,
+      "step": 1870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8831209201119057e-05,
+      "loss": 0.1626,
+      "step": 1880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8824992228784583e-05,
+      "loss": 0.1875,
+      "step": 1890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.881877525645011e-05,
+      "loss": 0.1162,
+      "step": 1900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8812558284115635e-05,
+      "loss": 0.203,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8806341311781165e-05,
+      "loss": 0.1937,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.880012433944669e-05,
+      "loss": 0.2276,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.879390736711222e-05,
+      "loss": 0.128,
+      "step": 1940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8787690394777746e-05,
+      "loss": 0.1819,
+      "step": 1950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8781473422443272e-05,
+      "loss": 0.1333,
+      "step": 1960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8775256450108798e-05,
+      "loss": 0.1531,
+      "step": 1970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8769039477774324e-05,
+      "loss": 0.1907,
+      "step": 1980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8762822505439853e-05,
+      "loss": 0.1259,
+      "step": 1990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.875660553310538e-05,
+      "loss": 0.159,
+      "step": 2000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.875038856077091e-05,
+      "loss": 0.2116,
+      "step": 2010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8744171588436434e-05,
+      "loss": 0.2144,
+      "step": 2020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.873795461610196e-05,
+      "loss": 0.2221,
+      "step": 2030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8731737643767486e-05,
+      "loss": 0.1156,
+      "step": 2040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8725520671433012e-05,
+      "loss": 0.1401,
+      "step": 2050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8719303699098542e-05,
+      "loss": 0.1724,
+      "step": 2060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8713086726764068e-05,
+      "loss": 0.1084,
+      "step": 2070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8706869754429594e-05,
+      "loss": 0.2258,
+      "step": 2080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.870065278209512e-05,
+      "loss": 0.2928,
+      "step": 2090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8694435809760646e-05,
+      "loss": 0.1396,
+      "step": 2100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8688218837426175e-05,
+      "loss": 0.2783,
+      "step": 2110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.86820018650917e-05,
+      "loss": 0.1927,
+      "step": 2120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.867578489275723e-05,
+      "loss": 0.2052,
+      "step": 2130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8669567920422756e-05,
+      "loss": 0.1972,
+      "step": 2140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8663350948088282e-05,
+      "loss": 0.225,
+      "step": 2150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8657133975753808e-05,
+      "loss": 0.2785,
+      "step": 2160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8650917003419334e-05,
+      "loss": 0.2944,
+      "step": 2170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8644700031084864e-05,
+      "loss": 0.1658,
+      "step": 2180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.863848305875039e-05,
+      "loss": 0.2138,
+      "step": 2190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.863226608641592e-05,
+      "loss": 0.1792,
+      "step": 2200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8626049114081445e-05,
+      "loss": 0.153,
+      "step": 2210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.861983214174697e-05,
+      "loss": 0.1495,
+      "step": 2220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8613615169412497e-05,
+      "loss": 0.262,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8607398197078023e-05,
+      "loss": 0.1946,
+      "step": 2240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8601181224743552e-05,
+      "loss": 0.1768,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8594964252409078e-05,
+      "loss": 0.2519,
+      "step": 2260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8588747280074607e-05,
+      "loss": 0.1583,
+      "step": 2270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8582530307740133e-05,
+      "loss": 0.1101,
+      "step": 2280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.857631333540566e-05,
+      "loss": 0.1953,
+      "step": 2290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8570096363071185e-05,
+      "loss": 0.1953,
+      "step": 2300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.856387939073671e-05,
+      "loss": 0.1676,
+      "step": 2310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.855766241840224e-05,
+      "loss": 0.1976,
+      "step": 2320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8551445446067767e-05,
+      "loss": 0.1804,
+      "step": 2330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8545228473733293e-05,
+      "loss": 0.2403,
+      "step": 2340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.853901150139882e-05,
+      "loss": 0.1236,
+      "step": 2350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8532794529064345e-05,
+      "loss": 0.1579,
+      "step": 2360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8526577556729874e-05,
+      "loss": 0.1378,
+      "step": 2370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.85203605843954e-05,
+      "loss": 0.2052,
+      "step": 2380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.851414361206093e-05,
+      "loss": 0.2143,
+      "step": 2390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8507926639726455e-05,
+      "loss": 0.2167,
+      "step": 2400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.850170966739198e-05,
+      "loss": 0.233,
+      "step": 2410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8495492695057507e-05,
+      "loss": 0.162,
+      "step": 2420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8489275722723033e-05,
+      "loss": 0.1778,
+      "step": 2430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8483058750388563e-05,
+      "loss": 0.2168,
+      "step": 2440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.847684177805409e-05,
+      "loss": 0.1423,
+      "step": 2450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8470624805719618e-05,
+      "loss": 0.1304,
+      "step": 2460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8464407833385144e-05,
+      "loss": 0.2245,
+      "step": 2470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.845819086105067e-05,
+      "loss": 0.1728,
+      "step": 2480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8451973888716196e-05,
+      "loss": 0.1373,
+      "step": 2490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8445756916381722e-05,
+      "loss": 0.1906,
+      "step": 2500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.843953994404725e-05,
+      "loss": 0.2493,
+      "step": 2510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8433322971712777e-05,
+      "loss": 0.1949,
+      "step": 2520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8427105999378307e-05,
+      "loss": 0.1458,
+      "step": 2530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8420889027043832e-05,
+      "loss": 0.2337,
+      "step": 2540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.841467205470936e-05,
+      "loss": 0.1528,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8408455082374884e-05,
+      "loss": 0.1116,
+      "step": 2560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.840223811004041e-05,
+      "loss": 0.1922,
+      "step": 2570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.839602113770594e-05,
+      "loss": 0.2645,
+      "step": 2580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8389804165371466e-05,
+      "loss": 0.1877,
+      "step": 2590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8383587193036995e-05,
+      "loss": 0.1848,
+      "step": 2600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837737022070252e-05,
+      "loss": 0.2055,
+      "step": 2610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8371153248368047e-05,
+      "loss": 0.2675,
+      "step": 2620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8364936276033573e-05,
+      "loss": 0.146,
+      "step": 2630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.83587193036991e-05,
+      "loss": 0.1186,
+      "step": 2640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.835250233136463e-05,
+      "loss": 0.1855,
+      "step": 2650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8346285359030154e-05,
+      "loss": 0.157,
+      "step": 2660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.834006838669568e-05,
+      "loss": 0.1632,
+      "step": 2670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8333851414361206e-05,
+      "loss": 0.2215,
+      "step": 2680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8327634442026732e-05,
+      "loss": 0.1797,
+      "step": 2690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.832141746969226e-05,
+      "loss": 0.1637,
+      "step": 2700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8315200497357788e-05,
+      "loss": 0.1872,
+      "step": 2710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8308983525023317e-05,
+      "loss": 0.1608,
+      "step": 2720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8302766552688843e-05,
+      "loss": 0.1395,
+      "step": 2730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.829654958035437e-05,
+      "loss": 0.206,
+      "step": 2740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8290332608019895e-05,
+      "loss": 0.1329,
+      "step": 2750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.828411563568542e-05,
+      "loss": 0.2381,
+      "step": 2760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.827789866335095e-05,
+      "loss": 0.1259,
+      "step": 2770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8271681691016476e-05,
+      "loss": 0.1355,
+      "step": 2780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8265464718682006e-05,
+      "loss": 0.1409,
+      "step": 2790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.825924774634753e-05,
+      "loss": 0.2008,
+      "step": 2800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8253030774013057e-05,
+      "loss": 0.1917,
+      "step": 2810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8246813801678583e-05,
+      "loss": 0.1932,
+      "step": 2820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.824059682934411e-05,
+      "loss": 0.149,
+      "step": 2830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.823437985700964e-05,
+      "loss": 0.163,
+      "step": 2840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8228162884675165e-05,
+      "loss": 0.1699,
+      "step": 2850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.822194591234069e-05,
+      "loss": 0.1113,
+      "step": 2860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.821572894000622e-05,
+      "loss": 0.1921,
+      "step": 2870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8209511967671746e-05,
+      "loss": 0.1468,
+      "step": 2880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8203294995337272e-05,
+      "loss": 0.229,
+      "step": 2890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8197078023002798e-05,
+      "loss": 0.1813,
+      "step": 2900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8190861050668327e-05,
+      "loss": 0.098,
+      "step": 2910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8184644078333853e-05,
+      "loss": 0.1549,
+      "step": 2920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.817842710599938e-05,
+      "loss": 0.2878,
+      "step": 2930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.817221013366491e-05,
+      "loss": 0.1384,
+      "step": 2940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8165993161330435e-05,
+      "loss": 0.223,
+      "step": 2950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.815977618899596e-05,
+      "loss": 0.2422,
+      "step": 2960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8153559216661487e-05,
+      "loss": 0.108,
+      "step": 2970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8147342244327016e-05,
+      "loss": 0.11,
+      "step": 2980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8141125271992542e-05,
+      "loss": 0.3302,
+      "step": 2990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8134908299658068e-05,
+      "loss": 0.2265,
+      "step": 3000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8128691327323594e-05,
+      "loss": 0.1773,
+      "step": 3010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.812247435498912e-05,
+      "loss": 0.1203,
+      "step": 3020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.811625738265465e-05,
+      "loss": 0.214,
+      "step": 3030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8110040410320175e-05,
+      "loss": 0.1317,
+      "step": 3040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.81038234379857e-05,
+      "loss": 0.2677,
+      "step": 3050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.809760646565123e-05,
+      "loss": 0.2044,
+      "step": 3060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8091389493316756e-05,
+      "loss": 0.2198,
+      "step": 3070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8085172520982282e-05,
+      "loss": 0.1562,
+      "step": 3080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.807895554864781e-05,
+      "loss": 0.2986,
+      "step": 3090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8072738576313338e-05,
+      "loss": 0.1326,
+      "step": 3100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8066521603978864e-05,
+      "loss": 0.155,
+      "step": 3110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.806030463164439e-05,
+      "loss": 0.1297,
+      "step": 3120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.805408765930992e-05,
+      "loss": 0.1509,
+      "step": 3130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8047870686975445e-05,
+      "loss": 0.1781,
+      "step": 3140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.804165371464097e-05,
+      "loss": 0.1042,
+      "step": 3150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8035436742306497e-05,
+      "loss": 0.1696,
+      "step": 3160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8029219769972026e-05,
+      "loss": 0.1539,
+      "step": 3170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8023002797637552e-05,
+      "loss": 0.1496,
+      "step": 3180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8016785825303078e-05,
+      "loss": 0.1115,
+      "step": 3190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8010568852968608e-05,
+      "loss": 0.1542,
+      "step": 3200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8004351880634134e-05,
+      "loss": 0.1694,
+      "step": 3210
+    },
+    {
+      "epoch": 1.0,
+      "eval_FN": 167,
+      "eval_FP": 282,
+      "eval_TN": 5556,
+      "eval_TP": 428,
+      "eval_accuracy": 0.9302036374941707,
+      "eval_f1": 0.6559386973180077,
+      "eval_loss": 0.15284980833530426,
+      "eval_precision": 0.6028169014084507,
+      "eval_recall": 0.719327731092437,
+      "eval_runtime": 44.2385,
+      "eval_samples_per_second": 145.416,
+      "eval_steps_per_second": 9.11,
+      "step": 3217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.799813490829966e-05,
+      "loss": 0.1347,
+      "step": 3220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7991917935965186e-05,
+      "loss": 0.1658,
+      "step": 3230
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.798570096363071e-05,
+      "loss": 0.1627,
+      "step": 3240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.797948399129624e-05,
+      "loss": 0.1611,
+      "step": 3250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7973267018961767e-05,
+      "loss": 0.0682,
+      "step": 3260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7967050046627296e-05,
+      "loss": 0.1272,
+      "step": 3270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7960833074292822e-05,
+      "loss": 0.1918,
+      "step": 3280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7954616101958348e-05,
+      "loss": 0.1376,
+      "step": 3290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7948399129623874e-05,
+      "loss": 0.1097,
+      "step": 3300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.79421821572894e-05,
+      "loss": 0.1886,
+      "step": 3310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.793596518495493e-05,
+      "loss": 0.1053,
+      "step": 3320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7929748212620455e-05,
+      "loss": 0.0787,
+      "step": 3330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.792353124028598e-05,
+      "loss": 0.1485,
+      "step": 3340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7917314267951507e-05,
+      "loss": 0.2014,
+      "step": 3350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7911097295617037e-05,
+      "loss": 0.1432,
+      "step": 3360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7904880323282563e-05,
+      "loss": 0.1306,
+      "step": 3370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.789866335094809e-05,
+      "loss": 0.1445,
+      "step": 3380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7892446378613618e-05,
+      "loss": 0.2115,
+      "step": 3390
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7886229406279144e-05,
+      "loss": 0.2041,
+      "step": 3400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.788001243394467e-05,
+      "loss": 0.1273,
+      "step": 3410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7873795461610196e-05,
+      "loss": 0.1711,
+      "step": 3420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7867578489275722e-05,
+      "loss": 0.1312,
+      "step": 3430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.786136151694125e-05,
+      "loss": 0.1883,
+      "step": 3440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7855144544606777e-05,
+      "loss": 0.1525,
+      "step": 3450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7848927572272307e-05,
+      "loss": 0.247,
+      "step": 3460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7842710599937833e-05,
+      "loss": 0.1988,
+      "step": 3470
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.783649362760336e-05,
+      "loss": 0.0987,
+      "step": 3480
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7830276655268885e-05,
+      "loss": 0.2392,
+      "step": 3490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.782405968293441e-05,
+      "loss": 0.1028,
+      "step": 3500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.781784271059994e-05,
+      "loss": 0.126,
+      "step": 3510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7811625738265466e-05,
+      "loss": 0.2144,
+      "step": 3520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7805408765930995e-05,
+      "loss": 0.1192,
+      "step": 3530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.779919179359652e-05,
+      "loss": 0.1401,
+      "step": 3540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7792974821262047e-05,
+      "loss": 0.1596,
+      "step": 3550
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7786757848927573e-05,
+      "loss": 0.1066,
+      "step": 3560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.77805408765931e-05,
+      "loss": 0.2651,
+      "step": 3570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.777432390425863e-05,
+      "loss": 0.1078,
+      "step": 3580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7768106931924154e-05,
+      "loss": 0.1002,
+      "step": 3590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.776188995958968e-05,
+      "loss": 0.1789,
+      "step": 3600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7755672987255206e-05,
+      "loss": 0.1257,
+      "step": 3610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7749456014920732e-05,
+      "loss": 0.2421,
+      "step": 3620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7743239042586262e-05,
+      "loss": 0.1032,
+      "step": 3630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7737022070251788e-05,
+      "loss": 0.2197,
+      "step": 3640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7730805097917317e-05,
+      "loss": 0.1614,
+      "step": 3650
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7724588125582843e-05,
+      "loss": 0.1161,
+      "step": 3660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.771837115324837e-05,
+      "loss": 0.1665,
+      "step": 3670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7712154180913895e-05,
+      "loss": 0.2234,
+      "step": 3680
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.770593720857942e-05,
+      "loss": 0.1737,
+      "step": 3690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.769972023624495e-05,
+      "loss": 0.1686,
+      "step": 3700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7693503263910476e-05,
+      "loss": 0.1488,
+      "step": 3710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7687286291576006e-05,
+      "loss": 0.1327,
+      "step": 3720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.768106931924153e-05,
+      "loss": 0.2055,
+      "step": 3730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7674852346907058e-05,
+      "loss": 0.0838,
+      "step": 3740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7668635374572584e-05,
+      "loss": 0.1591,
+      "step": 3750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.766241840223811e-05,
+      "loss": 0.1303,
+      "step": 3760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.765620142990364e-05,
+      "loss": 0.1705,
+      "step": 3770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7649984457569165e-05,
+      "loss": 0.2133,
+      "step": 3780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7643767485234694e-05,
+      "loss": 0.1598,
+      "step": 3790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.763755051290022e-05,
+      "loss": 0.1383,
+      "step": 3800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7631333540565746e-05,
+      "loss": 0.1023,
+      "step": 3810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7625116568231272e-05,
+      "loss": 0.1213,
+      "step": 3820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7618899595896798e-05,
+      "loss": 0.1929,
+      "step": 3830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7612682623562328e-05,
+      "loss": 0.0804,
+      "step": 3840
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7606465651227853e-05,
+      "loss": 0.1408,
+      "step": 3850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7600248678893383e-05,
+      "loss": 0.2302,
+      "step": 3860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.759403170655891e-05,
+      "loss": 0.1419,
+      "step": 3870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7587814734224435e-05,
+      "loss": 0.1511,
+      "step": 3880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.758159776188996e-05,
+      "loss": 0.2258,
+      "step": 3890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7575380789555487e-05,
+      "loss": 0.0988,
+      "step": 3900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7569163817221016e-05,
+      "loss": 0.1295,
+      "step": 3910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7562946844886542e-05,
+      "loss": 0.1557,
+      "step": 3920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7556729872552068e-05,
+      "loss": 0.1884,
+      "step": 3930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7550512900217594e-05,
+      "loss": 0.0964,
+      "step": 3940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.754429592788312e-05,
+      "loss": 0.088,
+      "step": 3950
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.753807895554865e-05,
+      "loss": 0.182,
+      "step": 3960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7531861983214175e-05,
+      "loss": 0.1338,
+      "step": 3970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7525645010879705e-05,
+      "loss": 0.1663,
+      "step": 3980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.751942803854523e-05,
+      "loss": 0.1579,
+      "step": 3990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7513211066210757e-05,
+      "loss": 0.1745,
+      "step": 4000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7506994093876283e-05,
+      "loss": 0.1991,
+      "step": 4010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.750077712154181e-05,
+      "loss": 0.174,
+      "step": 4020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7494560149207338e-05,
+      "loss": 0.187,
+      "step": 4030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7488343176872864e-05,
+      "loss": 0.1624,
+      "step": 4040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7482126204538393e-05,
+      "loss": 0.2008,
+      "step": 4050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.747590923220392e-05,
+      "loss": 0.1045,
+      "step": 4060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7469692259869445e-05,
+      "loss": 0.1736,
+      "step": 4070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746347528753497e-05,
+      "loss": 0.2203,
+      "step": 4080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7457258315200497e-05,
+      "loss": 0.1772,
+      "step": 4090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7451041342866027e-05,
+      "loss": 0.1489,
+      "step": 4100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7444824370531552e-05,
+      "loss": 0.147,
+      "step": 4110
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7438607398197082e-05,
+      "loss": 0.1397,
+      "step": 4120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7432390425862608e-05,
+      "loss": 0.2557,
+      "step": 4130
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7426173453528134e-05,
+      "loss": 0.1262,
+      "step": 4140
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741995648119366e-05,
+      "loss": 0.1594,
+      "step": 4150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7413739508859186e-05,
+      "loss": 0.0806,
+      "step": 4160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7407522536524715e-05,
+      "loss": 0.0893,
+      "step": 4170
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.740130556419024e-05,
+      "loss": 0.1777,
+      "step": 4180
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7395088591855767e-05,
+      "loss": 0.2322,
+      "step": 4190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7388871619521296e-05,
+      "loss": 0.1417,
+      "step": 4200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7382654647186822e-05,
+      "loss": 0.0811,
+      "step": 4210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.737643767485235e-05,
+      "loss": 0.148,
+      "step": 4220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7370220702517874e-05,
+      "loss": 0.1863,
+      "step": 4230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7364003730183404e-05,
+      "loss": 0.1356,
+      "step": 4240
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.735778675784893e-05,
+      "loss": 0.1015,
+      "step": 4250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7351569785514456e-05,
+      "loss": 0.1622,
+      "step": 4260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.734535281317998e-05,
+      "loss": 0.1272,
+      "step": 4270
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7339135840845508e-05,
+      "loss": 0.1485,
+      "step": 4280
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7332918868511037e-05,
+      "loss": 0.1127,
+      "step": 4290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7326701896176563e-05,
+      "loss": 0.0915,
+      "step": 4300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7320484923842092e-05,
+      "loss": 0.1608,
+      "step": 4310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7314267951507618e-05,
+      "loss": 0.1627,
+      "step": 4320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7308050979173144e-05,
+      "loss": 0.1943,
+      "step": 4330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.730183400683867e-05,
+      "loss": 0.1469,
+      "step": 4340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7295617034504196e-05,
+      "loss": 0.0973,
+      "step": 4350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7289400062169726e-05,
+      "loss": 0.1963,
+      "step": 4360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.728318308983525e-05,
+      "loss": 0.0907,
+      "step": 4370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7276966117500777e-05,
+      "loss": 0.1354,
+      "step": 4380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7270749145166307e-05,
+      "loss": 0.2008,
+      "step": 4390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7264532172831833e-05,
+      "loss": 0.0663,
+      "step": 4400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.725831520049736e-05,
+      "loss": 0.1875,
+      "step": 4410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7252098228162885e-05,
+      "loss": 0.1751,
+      "step": 4420
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7245881255828414e-05,
+      "loss": 0.1679,
+      "step": 4430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.723966428349394e-05,
+      "loss": 0.2197,
+      "step": 4440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7233447311159466e-05,
+      "loss": 0.1957,
+      "step": 4450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7227230338824995e-05,
+      "loss": 0.0846,
+      "step": 4460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.722101336649052e-05,
+      "loss": 0.1534,
+      "step": 4470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7214796394156047e-05,
+      "loss": 0.1676,
+      "step": 4480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7208579421821573e-05,
+      "loss": 0.1989,
+      "step": 4490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7202362449487103e-05,
+      "loss": 0.1825,
+      "step": 4500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.719614547715263e-05,
+      "loss": 0.1496,
+      "step": 4510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7189928504818155e-05,
+      "loss": 0.1592,
+      "step": 4520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.718371153248368e-05,
+      "loss": 0.1602,
+      "step": 4530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7177494560149207e-05,
+      "loss": 0.1447,
+      "step": 4540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7171277587814736e-05,
+      "loss": 0.1271,
+      "step": 4550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7165060615480262e-05,
+      "loss": 0.2224,
+      "step": 4560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7158843643145788e-05,
+      "loss": 0.1817,
+      "step": 4570
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7152626670811317e-05,
+      "loss": 0.1691,
+      "step": 4580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7146409698476843e-05,
+      "loss": 0.1586,
+      "step": 4590
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.714019272614237e-05,
+      "loss": 0.0611,
+      "step": 4600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7133975753807895e-05,
+      "loss": 0.1866,
+      "step": 4610
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7127758781473425e-05,
+      "loss": 0.1688,
+      "step": 4620
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.712154180913895e-05,
+      "loss": 0.129,
+      "step": 4630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7115324836804476e-05,
+      "loss": 0.182,
+      "step": 4640
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7109107864470006e-05,
+      "loss": 0.1664,
+      "step": 4650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7102890892135532e-05,
+      "loss": 0.3302,
+      "step": 4660
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7096673919801058e-05,
+      "loss": 0.1479,
+      "step": 4670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7090456947466584e-05,
+      "loss": 0.0942,
+      "step": 4680
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7084239975132113e-05,
+      "loss": 0.221,
+      "step": 4690
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.707802300279764e-05,
+      "loss": 0.1719,
+      "step": 4700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7071806030463165e-05,
+      "loss": 0.1876,
+      "step": 4710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7065589058128694e-05,
+      "loss": 0.118,
+      "step": 4720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.705937208579422e-05,
+      "loss": 0.1171,
+      "step": 4730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7053155113459746e-05,
+      "loss": 0.2566,
+      "step": 4740
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7046938141125272e-05,
+      "loss": 0.215,
+      "step": 4750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.70407211687908e-05,
+      "loss": 0.1127,
+      "step": 4760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7034504196456328e-05,
+      "loss": 0.1269,
+      "step": 4770
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7028287224121854e-05,
+      "loss": 0.0621,
+      "step": 4780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.7022070251787383e-05,
+      "loss": 0.2373,
+      "step": 4790
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.701585327945291e-05,
+      "loss": 0.1588,
+      "step": 4800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.7009636307118435e-05,
+      "loss": 0.2696,
+      "step": 4810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.700341933478396e-05,
+      "loss": 0.1642,
+      "step": 4820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6997202362449487e-05,
+      "loss": 0.1549,
+      "step": 4830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6990985390115016e-05,
+      "loss": 0.177,
+      "step": 4840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6984768417780542e-05,
+      "loss": 0.1161,
+      "step": 4850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6978551445446068e-05,
+      "loss": 0.1328,
+      "step": 4860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6972334473111594e-05,
+      "loss": 0.1233,
+      "step": 4870
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6966117500777124e-05,
+      "loss": 0.1404,
+      "step": 4880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.695990052844265e-05,
+      "loss": 0.0819,
+      "step": 4890
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6953683556108175e-05,
+      "loss": 0.1503,
+      "step": 4900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6947466583773705e-05,
+      "loss": 0.1954,
+      "step": 4910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.694124961143923e-05,
+      "loss": 0.2012,
+      "step": 4920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6935032639104757e-05,
+      "loss": 0.0777,
+      "step": 4930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6928815666770283e-05,
+      "loss": 0.083,
+      "step": 4940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.692259869443581e-05,
+      "loss": 0.1624,
+      "step": 4950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6916381722101338e-05,
+      "loss": 0.2015,
+      "step": 4960
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6910164749766864e-05,
+      "loss": 0.138,
+      "step": 4970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6903947777432393e-05,
+      "loss": 0.2621,
+      "step": 4980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.689773080509792e-05,
+      "loss": 0.1083,
+      "step": 4990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6891513832763445e-05,
+      "loss": 0.1087,
+      "step": 5000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.688529686042897e-05,
+      "loss": 0.1873,
+      "step": 5010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6879079888094497e-05,
+      "loss": 0.1259,
+      "step": 5020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6872862915760027e-05,
+      "loss": 0.0729,
+      "step": 5030
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6866645943425553e-05,
+      "loss": 0.1582,
+      "step": 5040
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6860428971091082e-05,
+      "loss": 0.1669,
+      "step": 5050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6854211998756608e-05,
+      "loss": 0.1894,
+      "step": 5060
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6847995026422134e-05,
+      "loss": 0.2505,
+      "step": 5070
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.684177805408766e-05,
+      "loss": 0.173,
+      "step": 5080
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6835561081753186e-05,
+      "loss": 0.1505,
+      "step": 5090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6829344109418715e-05,
+      "loss": 0.1305,
+      "step": 5100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.682312713708424e-05,
+      "loss": 0.1443,
+      "step": 5110
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.681691016474977e-05,
+      "loss": 0.1572,
+      "step": 5120
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.6810693192415297e-05,
+      "loss": 0.1542,
+      "step": 5130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6804476220080823e-05,
+      "loss": 0.1545,
+      "step": 5140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.679825924774635e-05,
+      "loss": 0.1387,
+      "step": 5150
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.6792042275411874e-05,
+      "loss": 0.1649,
+      "step": 5160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6785825303077404e-05,
+      "loss": 0.0822,
+      "step": 5170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.677960833074293e-05,
+      "loss": 0.2484,
+      "step": 5180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.6773391358408456e-05,
+      "loss": 0.0728,
+      "step": 5190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6767174386073982e-05,
+      "loss": 0.1138,
+      "step": 5200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6760957413739508e-05,
+      "loss": 0.2018,
+      "step": 5210
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.6754740441405037e-05,
+      "loss": 0.0981,
+      "step": 5220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6748523469070563e-05,
+      "loss": 0.1082,
+      "step": 5230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6742306496736092e-05,
+      "loss": 0.1607,
+      "step": 5240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.673608952440162e-05,
+      "loss": 0.1488,
+      "step": 5250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6729872552067144e-05,
+      "loss": 0.1016,
+      "step": 5260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.672365557973267e-05,
+      "loss": 0.1382,
+      "step": 5270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6717438607398196e-05,
+      "loss": 0.1637,
+      "step": 5280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6711221635063726e-05,
+      "loss": 0.1755,
+      "step": 5290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.670500466272925e-05,
+      "loss": 0.1522,
+      "step": 5300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.669878769039478e-05,
+      "loss": 0.1089,
+      "step": 5310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6692570718060307e-05,
+      "loss": 0.1372,
+      "step": 5320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6686353745725833e-05,
+      "loss": 0.2365,
+      "step": 5330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.668013677339136e-05,
+      "loss": 0.1474,
+      "step": 5340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.6673919801056885e-05,
+      "loss": 0.1116,
+      "step": 5350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.6667702828722414e-05,
+      "loss": 0.2007,
+      "step": 5360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.666148585638794e-05,
+      "loss": 0.1344,
+      "step": 5370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.665526888405347e-05,
+      "loss": 0.1078,
+      "step": 5380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6649051911718996e-05,
+      "loss": 0.0841,
+      "step": 5390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.664283493938452e-05,
+      "loss": 0.1305,
+      "step": 5400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6636617967050048e-05,
+      "loss": 0.1124,
+      "step": 5410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.6630400994715573e-05,
+      "loss": 0.2056,
+      "step": 5420
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6624184022381103e-05,
+      "loss": 0.1243,
+      "step": 5430
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.661796705004663e-05,
+      "loss": 0.2243,
+      "step": 5440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.6611750077712158e-05,
+      "loss": 0.1233,
+      "step": 5450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6605533105377684e-05,
+      "loss": 0.107,
+      "step": 5460
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6599316133043207e-05,
+      "loss": 0.1968,
+      "step": 5470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.6593099160708736e-05,
+      "loss": 0.1646,
+      "step": 5480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6586882188374262e-05,
+      "loss": 0.1242,
+      "step": 5490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.658066521603979e-05,
+      "loss": 0.2767,
+      "step": 5500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.6574448243705317e-05,
+      "loss": 0.1639,
+      "step": 5510
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6568231271370843e-05,
+      "loss": 0.121,
+      "step": 5520
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.656201429903637e-05,
+      "loss": 0.1553,
+      "step": 5530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.6555797326701895e-05,
+      "loss": 0.1349,
+      "step": 5540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6549580354367425e-05,
+      "loss": 0.197,
+      "step": 5550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.654336338203295e-05,
+      "loss": 0.1141,
+      "step": 5560
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.653714640969848e-05,
+      "loss": 0.1639,
+      "step": 5570
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.6530929437364006e-05,
+      "loss": 0.1332,
+      "step": 5580
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6524712465029532e-05,
+      "loss": 0.1509,
+      "step": 5590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6518495492695058e-05,
+      "loss": 0.0541,
+      "step": 5600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.6512278520360584e-05,
+      "loss": 0.1565,
+      "step": 5610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.6506061548026113e-05,
+      "loss": 0.1562,
+      "step": 5620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.649984457569164e-05,
+      "loss": 0.1232,
+      "step": 5630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.649362760335717e-05,
+      "loss": 0.2226,
+      "step": 5640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6487410631022695e-05,
+      "loss": 0.1225,
+      "step": 5650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.648119365868822e-05,
+      "loss": 0.0688,
+      "step": 5660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.6474976686353747e-05,
+      "loss": 0.1514,
+      "step": 5670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6468759714019272e-05,
+      "loss": 0.2463,
+      "step": 5680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6462542741684802e-05,
+      "loss": 0.1406,
+      "step": 5690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6456325769350328e-05,
+      "loss": 0.1332,
+      "step": 5700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6450108797015857e-05,
+      "loss": 0.1185,
+      "step": 5710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6443891824681383e-05,
+      "loss": 0.1782,
+      "step": 5720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.643767485234691e-05,
+      "loss": 0.1566,
+      "step": 5730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6431457880012435e-05,
+      "loss": 0.1028,
+      "step": 5740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.642524090767796e-05,
+      "loss": 0.1727,
+      "step": 5750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.641902393534349e-05,
+      "loss": 0.1729,
+      "step": 5760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6412806963009016e-05,
+      "loss": 0.1069,
+      "step": 5770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6406589990674542e-05,
+      "loss": 0.1962,
+      "step": 5780
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.640037301834007e-05,
+      "loss": 0.1599,
+      "step": 5790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6394156046005594e-05,
+      "loss": 0.1874,
+      "step": 5800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6387939073671124e-05,
+      "loss": 0.1428,
+      "step": 5810
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.638172210133665e-05,
+      "loss": 0.1933,
+      "step": 5820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.637550512900218e-05,
+      "loss": 0.1226,
+      "step": 5830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6369288156667705e-05,
+      "loss": 0.1701,
+      "step": 5840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.636307118433323e-05,
+      "loss": 0.0851,
+      "step": 5850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6356854211998757e-05,
+      "loss": 0.1121,
+      "step": 5860
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6350637239664283e-05,
+      "loss": 0.1369,
+      "step": 5870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6344420267329812e-05,
+      "loss": 0.2528,
+      "step": 5880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6338203294995338e-05,
+      "loss": 0.1135,
+      "step": 5890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6331986322660868e-05,
+      "loss": 0.1414,
+      "step": 5900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6325769350326394e-05,
+      "loss": 0.2017,
+      "step": 5910
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.631955237799192e-05,
+      "loss": 0.1068,
+      "step": 5920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6313335405657446e-05,
+      "loss": 0.1441,
+      "step": 5930
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.630711843332297e-05,
+      "loss": 0.0821,
+      "step": 5940
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.63009014609885e-05,
+      "loss": 0.1278,
+      "step": 5950
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.6294684488654027e-05,
+      "loss": 0.2407,
+      "step": 5960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6288467516319553e-05,
+      "loss": 0.1555,
+      "step": 5970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6282250543985082e-05,
+      "loss": 0.2165,
+      "step": 5980
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.6276033571650608e-05,
+      "loss": 0.1543,
+      "step": 5990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6269816599316134e-05,
+      "loss": 0.1399,
+      "step": 6000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.626359962698166e-05,
+      "loss": 0.1243,
+      "step": 6010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.625738265464719e-05,
+      "loss": 0.1104,
+      "step": 6020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.6251165682312715e-05,
+      "loss": 0.1832,
+      "step": 6030
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.624494870997824e-05,
+      "loss": 0.1198,
+      "step": 6040
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.623873173764377e-05,
+      "loss": 0.1688,
+      "step": 6050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.6232514765309297e-05,
+      "loss": 0.1383,
+      "step": 6060
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6226297792974823e-05,
+      "loss": 0.1133,
+      "step": 6070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.622008082064035e-05,
+      "loss": 0.0912,
+      "step": 6080
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6213863848305878e-05,
+      "loss": 0.1409,
+      "step": 6090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6207646875971404e-05,
+      "loss": 0.2431,
+      "step": 6100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.620142990363693e-05,
+      "loss": 0.1538,
+      "step": 6110
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.6195212931302456e-05,
+      "loss": 0.0826,
+      "step": 6120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6188995958967982e-05,
+      "loss": 0.1321,
+      "step": 6130
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.618277898663351e-05,
+      "loss": 0.1502,
+      "step": 6140
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6176562014299037e-05,
+      "loss": 0.1742,
+      "step": 6150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.6170345041964563e-05,
+      "loss": 0.1627,
+      "step": 6160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6164128069630093e-05,
+      "loss": 0.1329,
+      "step": 6170
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.615791109729562e-05,
+      "loss": 0.1655,
+      "step": 6180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6151694124961145e-05,
+      "loss": 0.2386,
+      "step": 6190
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.614547715262667e-05,
+      "loss": 0.1213,
+      "step": 6200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.61392601802922e-05,
+      "loss": 0.1011,
+      "step": 6210
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6133043207957726e-05,
+      "loss": 0.1658,
+      "step": 6220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6126826235623252e-05,
+      "loss": 0.2215,
+      "step": 6230
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.612060926328878e-05,
+      "loss": 0.1032,
+      "step": 6240
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.6114392290954307e-05,
+      "loss": 0.1599,
+      "step": 6250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.6108175318619833e-05,
+      "loss": 0.1868,
+      "step": 6260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.610195834628536e-05,
+      "loss": 0.1248,
+      "step": 6270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.609574137395089e-05,
+      "loss": 0.1081,
+      "step": 6280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6089524401616414e-05,
+      "loss": 0.1373,
+      "step": 6290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.608330742928194e-05,
+      "loss": 0.199,
+      "step": 6300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.607709045694747e-05,
+      "loss": 0.1365,
+      "step": 6310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6070873484612996e-05,
+      "loss": 0.1114,
+      "step": 6320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6064656512278522e-05,
+      "loss": 0.0893,
+      "step": 6330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6058439539944048e-05,
+      "loss": 0.0914,
+      "step": 6340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6052222567609574e-05,
+      "loss": 0.0797,
+      "step": 6350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6046005595275103e-05,
+      "loss": 0.1205,
+      "step": 6360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.603978862294063e-05,
+      "loss": 0.1733,
+      "step": 6370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.603357165060616e-05,
+      "loss": 0.0943,
+      "step": 6380
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6027354678271684e-05,
+      "loss": 0.2009,
+      "step": 6390
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.602113770593721e-05,
+      "loss": 0.1884,
+      "step": 6400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6014920733602736e-05,
+      "loss": 0.2447,
+      "step": 6410
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6008703761268262e-05,
+      "loss": 0.1264,
+      "step": 6420
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.600248678893379e-05,
+      "loss": 0.2177,
+      "step": 6430
+    },
+    {
+      "epoch": 2.0,
+      "eval_FN": 180,
+      "eval_FP": 215,
+      "eval_TN": 5623,
+      "eval_TP": 415,
+      "eval_accuracy": 0.9385978548111301,
+      "eval_f1": 0.6775510204081632,
+      "eval_loss": 0.1442219316959381,
+      "eval_precision": 0.6587301587301587,
+      "eval_recall": 0.6974789915966386,
+      "eval_runtime": 44.2603,
+      "eval_samples_per_second": 145.345,
+      "eval_steps_per_second": 9.105,
+      "step": 6434
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5996269816599318e-05,
+      "loss": 0.0834,
+      "step": 6440
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5990052844264844e-05,
+      "loss": 0.1551,
+      "step": 6450
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.598383587193037e-05,
+      "loss": 0.1112,
+      "step": 6460
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.59776188995959e-05,
+      "loss": 0.1085,
+      "step": 6470
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.5971401927261425e-05,
+      "loss": 0.1218,
+      "step": 6480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.596518495492695e-05,
+      "loss": 0.1167,
+      "step": 6490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.595896798259248e-05,
+      "loss": 0.1141,
+      "step": 6500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.5952751010258006e-05,
+      "loss": 0.1566,
+      "step": 6510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5946534037923532e-05,
+      "loss": 0.091,
+      "step": 6520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5940317065589058e-05,
+      "loss": 0.1367,
+      "step": 6530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.5934100093254584e-05,
+      "loss": 0.1538,
+      "step": 6540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5927883120920113e-05,
+      "loss": 0.2202,
+      "step": 6550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.592166614858564e-05,
+      "loss": 0.1087,
+      "step": 6560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.591544917625117e-05,
+      "loss": 0.079,
+      "step": 6570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5909232203916695e-05,
+      "loss": 0.1876,
+      "step": 6580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.590301523158222e-05,
+      "loss": 0.1809,
+      "step": 6590
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5896798259247747e-05,
+      "loss": 0.072,
+      "step": 6600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5890581286913273e-05,
+      "loss": 0.1088,
+      "step": 6610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5884364314578802e-05,
+      "loss": 0.1324,
+      "step": 6620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5878147342244328e-05,
+      "loss": 0.2361,
+      "step": 6630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5871930369909857e-05,
+      "loss": 0.1797,
+      "step": 6640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5865713397575383e-05,
+      "loss": 0.1244,
+      "step": 6650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.585949642524091e-05,
+      "loss": 0.1151,
+      "step": 6660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5853279452906435e-05,
+      "loss": 0.0487,
+      "step": 6670
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.584706248057196e-05,
+      "loss": 0.0961,
+      "step": 6680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.584084550823749e-05,
+      "loss": 0.1198,
+      "step": 6690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5834628535903017e-05,
+      "loss": 0.0608,
+      "step": 6700
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5828411563568546e-05,
+      "loss": 0.1113,
+      "step": 6710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.582219459123407e-05,
+      "loss": 0.1172,
+      "step": 6720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5815977618899594e-05,
+      "loss": 0.1196,
+      "step": 6730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5809760646565124e-05,
+      "loss": 0.1029,
+      "step": 6740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.580354367423065e-05,
+      "loss": 0.137,
+      "step": 6750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.579732670189618e-05,
+      "loss": 0.137,
+      "step": 6760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5791109729561705e-05,
+      "loss": 0.1026,
+      "step": 6770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.578489275722723e-05,
+      "loss": 0.1019,
+      "step": 6780
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5778675784892757e-05,
+      "loss": 0.2507,
+      "step": 6790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5772458812558283e-05,
+      "loss": 0.1154,
+      "step": 6800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5766241840223812e-05,
+      "loss": 0.0863,
+      "step": 6810
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.576002486788934e-05,
+      "loss": 0.1455,
+      "step": 6820
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5753807895554868e-05,
+      "loss": 0.0983,
+      "step": 6830
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5747590923220394e-05,
+      "loss": 0.0725,
+      "step": 6840
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.574137395088592e-05,
+      "loss": 0.0593,
+      "step": 6850
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.5735156978551446e-05,
+      "loss": 0.178,
+      "step": 6860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.572894000621697e-05,
+      "loss": 0.1573,
+      "step": 6870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.57227230338825e-05,
+      "loss": 0.1572,
+      "step": 6880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5716506061548027e-05,
+      "loss": 0.1082,
+      "step": 6890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.5710289089213556e-05,
+      "loss": 0.1543,
+      "step": 6900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5704072116879082e-05,
+      "loss": 0.1381,
+      "step": 6910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5697855144544608e-05,
+      "loss": 0.1034,
+      "step": 6920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5691638172210134e-05,
+      "loss": 0.1135,
+      "step": 6930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.568542119987566e-05,
+      "loss": 0.0822,
+      "step": 6940
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.567920422754119e-05,
+      "loss": 0.0963,
+      "step": 6950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5672987255206716e-05,
+      "loss": 0.0925,
+      "step": 6960
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5666770282872245e-05,
+      "loss": 0.1573,
+      "step": 6970
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.566055331053777e-05,
+      "loss": 0.0988,
+      "step": 6980
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5654336338203297e-05,
+      "loss": 0.1147,
+      "step": 6990
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5648119365868823e-05,
+      "loss": 0.0866,
+      "step": 7000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.564190239353435e-05,
+      "loss": 0.1177,
+      "step": 7010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.5635685421199878e-05,
+      "loss": 0.0592,
+      "step": 7020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5629468448865404e-05,
+      "loss": 0.1371,
+      "step": 7030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.562325147653093e-05,
+      "loss": 0.2505,
+      "step": 7040
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5617034504196456e-05,
+      "loss": 0.048,
+      "step": 7050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.5610817531861982e-05,
+      "loss": 0.1072,
+      "step": 7060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.560460055952751e-05,
+      "loss": 0.1376,
+      "step": 7070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5598383587193037e-05,
+      "loss": 0.0766,
+      "step": 7080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.5592166614858567e-05,
+      "loss": 0.0947,
+      "step": 7090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5585949642524093e-05,
+      "loss": 0.1359,
+      "step": 7100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.557973267018962e-05,
+      "loss": 0.1368,
+      "step": 7110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.5573515697855145e-05,
+      "loss": 0.0927,
+      "step": 7120
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.556729872552067e-05,
+      "loss": 0.0625,
+      "step": 7130
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.55610817531862e-05,
+      "loss": 0.1248,
+      "step": 7140
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.5554864780851726e-05,
+      "loss": 0.0578,
+      "step": 7150
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5548647808517255e-05,
+      "loss": 0.1635,
+      "step": 7160
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.554243083618278e-05,
+      "loss": 0.1489,
+      "step": 7170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.5536213863848307e-05,
+      "loss": 0.109,
+      "step": 7180
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5529996891513833e-05,
+      "loss": 0.0638,
+      "step": 7190
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.552377991917936e-05,
+      "loss": 0.1613,
+      "step": 7200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.551756294684489e-05,
+      "loss": 0.1554,
+      "step": 7210
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.5511345974510415e-05,
+      "loss": 0.1052,
+      "step": 7220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5505129002175944e-05,
+      "loss": 0.1478,
+      "step": 7230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.549891202984147e-05,
+      "loss": 0.1352,
+      "step": 7240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.5492695057506996e-05,
+      "loss": 0.0863,
+      "step": 7250
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5486478085172522e-05,
+      "loss": 0.0673,
+      "step": 7260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5480261112838048e-05,
+      "loss": 0.0467,
+      "step": 7270
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.5474044140503577e-05,
+      "loss": 0.1552,
+      "step": 7280
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.5467827168169103e-05,
+      "loss": 0.087,
+      "step": 7290
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.546161019583463e-05,
+      "loss": 0.1214,
+      "step": 7300
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.545539322350016e-05,
+      "loss": 0.13,
+      "step": 7310
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5449176251165684e-05,
+      "loss": 0.1702,
+      "step": 7320
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.544295927883121e-05,
+      "loss": 0.1657,
+      "step": 7330
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5436742306496736e-05,
+      "loss": 0.0894,
+      "step": 7340
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.5430525334162266e-05,
+      "loss": 0.1037,
+      "step": 7350
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5424308361827792e-05,
+      "loss": 0.1483,
+      "step": 7360
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5418091389493318e-05,
+      "loss": 0.0934,
+      "step": 7370
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5411874417158844e-05,
+      "loss": 0.1178,
+      "step": 7380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.540565744482437e-05,
+      "loss": 0.1009,
+      "step": 7390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.53994404724899e-05,
+      "loss": 0.0862,
+      "step": 7400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5393223500155425e-05,
+      "loss": 0.2053,
+      "step": 7410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5387006527820954e-05,
+      "loss": 0.1238,
+      "step": 7420
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.538078955548648e-05,
+      "loss": 0.1629,
+      "step": 7430
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5374572583152006e-05,
+      "loss": 0.0602,
+      "step": 7440
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5368355610817532e-05,
+      "loss": 0.1006,
+      "step": 7450
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5362138638483058e-05,
+      "loss": 0.1712,
+      "step": 7460
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5355921666148588e-05,
+      "loss": 0.2494,
+      "step": 7470
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5349704693814114e-05,
+      "loss": 0.2267,
+      "step": 7480
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.534348772147964e-05,
+      "loss": 0.0833,
+      "step": 7490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.533727074914517e-05,
+      "loss": 0.1265,
+      "step": 7500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.5331053776810695e-05,
+      "loss": 0.0803,
+      "step": 7510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.532483680447622e-05,
+      "loss": 0.131,
+      "step": 7520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5318619832141747e-05,
+      "loss": 0.2193,
+      "step": 7530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.5312402859807276e-05,
+      "loss": 0.1408,
+      "step": 7540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5306185887472802e-05,
+      "loss": 0.1236,
+      "step": 7550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5299968915138328e-05,
+      "loss": 0.0865,
+      "step": 7560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.5293751942803858e-05,
+      "loss": 0.0754,
+      "step": 7570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5287534970469383e-05,
+      "loss": 0.1592,
+      "step": 7580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.528131799813491e-05,
+      "loss": 0.1693,
+      "step": 7590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.5275101025800435e-05,
+      "loss": 0.1168,
+      "step": 7600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5268884053465965e-05,
+      "loss": 0.0974,
+      "step": 7610
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.526266708113149e-05,
+      "loss": 0.0989,
+      "step": 7620
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5256450108797017e-05,
+      "loss": 0.0577,
+      "step": 7630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.5250233136462544e-05,
+      "loss": 0.1346,
+      "step": 7640
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.524401616412807e-05,
+      "loss": 0.2002,
+      "step": 7650
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5237799191793598e-05,
+      "loss": 0.121,
+      "step": 7660
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.5231582219459124e-05,
+      "loss": 0.1193,
+      "step": 7670
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.522536524712465e-05,
+      "loss": 0.1186,
+      "step": 7680
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.521914827479018e-05,
+      "loss": 0.1177,
+      "step": 7690
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.5212931302455705e-05,
+      "loss": 0.1918,
+      "step": 7700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5206714330121233e-05,
+      "loss": 0.0608,
+      "step": 7710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5200497357786759e-05,
+      "loss": 0.1264,
+      "step": 7720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.5194280385452287e-05,
+      "loss": 0.1598,
+      "step": 7730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5188063413117813e-05,
+      "loss": 0.1663,
+      "step": 7740
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5181846440783339e-05,
+      "loss": 0.0774,
+      "step": 7750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5175629468448866e-05,
+      "loss": 0.1418,
+      "step": 7760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5169412496114392e-05,
+      "loss": 0.1425,
+      "step": 7770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5163195523779922e-05,
+      "loss": 0.0723,
+      "step": 7780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5156978551445448e-05,
+      "loss": 0.0909,
+      "step": 7790
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5150761579110975e-05,
+      "loss": 0.1163,
+      "step": 7800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5144544606776501e-05,
+      "loss": 0.1204,
+      "step": 7810
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5138327634442027e-05,
+      "loss": 0.0835,
+      "step": 7820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5132110662107555e-05,
+      "loss": 0.188,
+      "step": 7830
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.512589368977308e-05,
+      "loss": 0.0687,
+      "step": 7840
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.511967671743861e-05,
+      "loss": 0.1783,
+      "step": 7850
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.5113459745104136e-05,
+      "loss": 0.1173,
+      "step": 7860
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5107242772769662e-05,
+      "loss": 0.0876,
+      "step": 7870
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.510102580043519e-05,
+      "loss": 0.1425,
+      "step": 7880
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5094808828100716e-05,
+      "loss": 0.2578,
+      "step": 7890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5088591855766243e-05,
+      "loss": 0.0885,
+      "step": 7900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.508237488343177e-05,
+      "loss": 0.1448,
+      "step": 7910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5076157911097297e-05,
+      "loss": 0.1591,
+      "step": 7920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5069940938762823e-05,
+      "loss": 0.0906,
+      "step": 7930
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5063723966428349e-05,
+      "loss": 0.1595,
+      "step": 7940
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5057506994093878e-05,
+      "loss": 0.1158,
+      "step": 7950
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5051290021759404e-05,
+      "loss": 0.1886,
+      "step": 7960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5045073049424932e-05,
+      "loss": 0.0923,
+      "step": 7970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5038856077090458e-05,
+      "loss": 0.1137,
+      "step": 7980
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5032639104755986e-05,
+      "loss": 0.1795,
+      "step": 7990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5026422132421512e-05,
+      "loss": 0.1336,
+      "step": 8000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5020205160087038e-05,
+      "loss": 0.1185,
+      "step": 8010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5013988187752567e-05,
+      "loss": 0.0517,
+      "step": 8020
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5007771215418093e-05,
+      "loss": 0.2702,
+      "step": 8030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.500155424308362e-05,
+      "loss": 0.0832,
+      "step": 8040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4995337270749147e-05,
+      "loss": 0.1874,
+      "step": 8050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4989120298414673e-05,
+      "loss": 0.1549,
+      "step": 8060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.49829033260802e-05,
+      "loss": 0.0999,
+      "step": 8070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4976686353745726e-05,
+      "loss": 0.1083,
+      "step": 8080
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4970469381411254e-05,
+      "loss": 0.0432,
+      "step": 8090
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.496425240907678e-05,
+      "loss": 0.0799,
+      "step": 8100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.495803543674231e-05,
+      "loss": 0.1371,
+      "step": 8110
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4951818464407835e-05,
+      "loss": 0.0976,
+      "step": 8120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4945601492073361e-05,
+      "loss": 0.0924,
+      "step": 8130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4939384519738889e-05,
+      "loss": 0.1938,
+      "step": 8140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4933167547404415e-05,
+      "loss": 0.1181,
+      "step": 8150
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4926950575069942e-05,
+      "loss": 0.1318,
+      "step": 8160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4920733602735468e-05,
+      "loss": 0.1228,
+      "step": 8170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4914516630400998e-05,
+      "loss": 0.1695,
+      "step": 8180
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4908299658066524e-05,
+      "loss": 0.0969,
+      "step": 8190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.490208268573205e-05,
+      "loss": 0.1613,
+      "step": 8200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4895865713397577e-05,
+      "loss": 0.1951,
+      "step": 8210
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4889648741063103e-05,
+      "loss": 0.093,
+      "step": 8220
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4883431768728631e-05,
+      "loss": 0.1596,
+      "step": 8230
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4877214796394157e-05,
+      "loss": 0.1504,
+      "step": 8240
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4870997824059683e-05,
+      "loss": 0.1215,
+      "step": 8250
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.486478085172521e-05,
+      "loss": 0.2947,
+      "step": 8260
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4858563879390737e-05,
+      "loss": 0.0705,
+      "step": 8270
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4852346907056266e-05,
+      "loss": 0.0785,
+      "step": 8280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4846129934721792e-05,
+      "loss": 0.1156,
+      "step": 8290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.483991296238732e-05,
+      "loss": 0.1164,
+      "step": 8300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4833695990052846e-05,
+      "loss": 0.1608,
+      "step": 8310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4827479017718372e-05,
+      "loss": 0.1059,
+      "step": 8320
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.48212620453839e-05,
+      "loss": 0.1681,
+      "step": 8330
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.4815045073049425e-05,
+      "loss": 0.1004,
+      "step": 8340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4808828100714955e-05,
+      "loss": 0.0897,
+      "step": 8350
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4802611128380479e-05,
+      "loss": 0.1266,
+      "step": 8360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4796394156046008e-05,
+      "loss": 0.1657,
+      "step": 8370
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4790177183711534e-05,
+      "loss": 0.1736,
+      "step": 8380
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.478396021137706e-05,
+      "loss": 0.0936,
+      "step": 8390
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4777743239042588e-05,
+      "loss": 0.1358,
+      "step": 8400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4771526266708114e-05,
+      "loss": 0.1408,
+      "step": 8410
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4765309294373641e-05,
+      "loss": 0.0124,
+      "step": 8420
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4759092322039167e-05,
+      "loss": 0.1789,
+      "step": 8430
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4752875349704693e-05,
+      "loss": 0.0409,
+      "step": 8440
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4746658377370223e-05,
+      "loss": 0.1761,
+      "step": 8450
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4740441405035749e-05,
+      "loss": 0.1538,
+      "step": 8460
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.4734224432701276e-05,
+      "loss": 0.1543,
+      "step": 8470
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4728007460366802e-05,
+      "loss": 0.1233,
+      "step": 8480
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.472179048803233e-05,
+      "loss": 0.1629,
+      "step": 8490
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.4715573515697856e-05,
+      "loss": 0.1622,
+      "step": 8500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4709356543363382e-05,
+      "loss": 0.0964,
+      "step": 8510
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.470313957102891e-05,
+      "loss": 0.1029,
+      "step": 8520
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4696922598694436e-05,
+      "loss": 0.1491,
+      "step": 8530
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.4690705626359965e-05,
+      "loss": 0.0323,
+      "step": 8540
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4684488654025491e-05,
+      "loss": 0.0994,
+      "step": 8550
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4678271681691019e-05,
+      "loss": 0.2367,
+      "step": 8560
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4672054709356545e-05,
+      "loss": 0.0729,
+      "step": 8570
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.466583773702207e-05,
+      "loss": 0.1231,
+      "step": 8580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4659620764687598e-05,
+      "loss": 0.2463,
+      "step": 8590
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4653403792353124e-05,
+      "loss": 0.1234,
+      "step": 8600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4647186820018654e-05,
+      "loss": 0.2137,
+      "step": 8610
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.464096984768418e-05,
+      "loss": 0.1017,
+      "step": 8620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.4634752875349705e-05,
+      "loss": 0.1139,
+      "step": 8630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4628535903015233e-05,
+      "loss": 0.2016,
+      "step": 8640
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4622318930680759e-05,
+      "loss": 0.0894,
+      "step": 8650
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.4616101958346287e-05,
+      "loss": 0.1332,
+      "step": 8660
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4609884986011813e-05,
+      "loss": 0.1425,
+      "step": 8670
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.460366801367734e-05,
+      "loss": 0.0611,
+      "step": 8680
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4597451041342866e-05,
+      "loss": 0.1086,
+      "step": 8690
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.4591234069008392e-05,
+      "loss": 0.1983,
+      "step": 8700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4585017096673922e-05,
+      "loss": 0.0662,
+      "step": 8710
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4578800124339448e-05,
+      "loss": 0.0869,
+      "step": 8720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.4572583152004975e-05,
+      "loss": 0.1326,
+      "step": 8730
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4566366179670501e-05,
+      "loss": 0.0881,
+      "step": 8740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4560149207336029e-05,
+      "loss": 0.1335,
+      "step": 8750
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.4553932235001555e-05,
+      "loss": 0.087,
+      "step": 8760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4547715262667081e-05,
+      "loss": 0.1915,
+      "step": 8770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.454149829033261e-05,
+      "loss": 0.1296,
+      "step": 8780
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.4535281317998136e-05,
+      "loss": 0.1222,
+      "step": 8790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4529064345663664e-05,
+      "loss": 0.0501,
+      "step": 8800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.452284737332919e-05,
+      "loss": 0.1298,
+      "step": 8810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4516630400994716e-05,
+      "loss": 0.1141,
+      "step": 8820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.4510413428660244e-05,
+      "loss": 0.2094,
+      "step": 8830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.450419645632577e-05,
+      "loss": 0.0449,
+      "step": 8840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4497979483991297e-05,
+      "loss": 0.0897,
+      "step": 8850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4491762511656823e-05,
+      "loss": 0.1066,
+      "step": 8860
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4485545539322353e-05,
+      "loss": 0.1737,
+      "step": 8870
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4479328566987879e-05,
+      "loss": 0.1213,
+      "step": 8880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.4473111594653404e-05,
+      "loss": 0.0663,
+      "step": 8890
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4466894622318932e-05,
+      "loss": 0.0481,
+      "step": 8900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4460677649984458e-05,
+      "loss": 0.1341,
+      "step": 8910
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4454460677649986e-05,
+      "loss": 0.2781,
+      "step": 8920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4448243705315512e-05,
+      "loss": 0.071,
+      "step": 8930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4442026732981041e-05,
+      "loss": 0.093,
+      "step": 8940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4435809760646567e-05,
+      "loss": 0.1591,
+      "step": 8950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4429592788312093e-05,
+      "loss": 0.1921,
+      "step": 8960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.442337581597762e-05,
+      "loss": 0.1215,
+      "step": 8970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4417158843643147e-05,
+      "loss": 0.1263,
+      "step": 8980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4410941871308674e-05,
+      "loss": 0.11,
+      "step": 8990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.44047248989742e-05,
+      "loss": 0.1593,
+      "step": 9000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4398507926639728e-05,
+      "loss": 0.0777,
+      "step": 9010
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4392290954305254e-05,
+      "loss": 0.1573,
+      "step": 9020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.438607398197078e-05,
+      "loss": 0.1599,
+      "step": 9030
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.437985700963631e-05,
+      "loss": 0.0813,
+      "step": 9040
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4373640037301835e-05,
+      "loss": 0.1656,
+      "step": 9050
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4367423064967363e-05,
+      "loss": 0.0634,
+      "step": 9060
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4361206092632889e-05,
+      "loss": 0.1483,
+      "step": 9070
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4354989120298415e-05,
+      "loss": 0.1012,
+      "step": 9080
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4348772147963943e-05,
+      "loss": 0.0794,
+      "step": 9090
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4342555175629469e-05,
+      "loss": 0.1159,
+      "step": 9100
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4336338203294998e-05,
+      "loss": 0.205,
+      "step": 9110
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4330121230960524e-05,
+      "loss": 0.1442,
+      "step": 9120
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4323904258626052e-05,
+      "loss": 0.1038,
+      "step": 9130
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4317687286291578e-05,
+      "loss": 0.1202,
+      "step": 9140
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4311470313957103e-05,
+      "loss": 0.069,
+      "step": 9150
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4305253341622631e-05,
+      "loss": 0.0529,
+      "step": 9160
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4299036369288157e-05,
+      "loss": 0.2056,
+      "step": 9170
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4292819396953685e-05,
+      "loss": 0.1181,
+      "step": 9180
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.428660242461921e-05,
+      "loss": 0.1022,
+      "step": 9190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.428038545228474e-05,
+      "loss": 0.1316,
+      "step": 9200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4274168479950266e-05,
+      "loss": 0.09,
+      "step": 9210
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4267951507615792e-05,
+      "loss": 0.1311,
+      "step": 9220
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.426173453528132e-05,
+      "loss": 0.1883,
+      "step": 9230
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4255517562946846e-05,
+      "loss": 0.0925,
+      "step": 9240
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4249300590612373e-05,
+      "loss": 0.115,
+      "step": 9250
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.42430836182779e-05,
+      "loss": 0.1349,
+      "step": 9260
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4236866645943425e-05,
+      "loss": 0.1311,
+      "step": 9270
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4230649673608955e-05,
+      "loss": 0.0384,
+      "step": 9280
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.422443270127448e-05,
+      "loss": 0.0586,
+      "step": 9290
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4218215728940008e-05,
+      "loss": 0.1187,
+      "step": 9300
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4211998756605534e-05,
+      "loss": 0.165,
+      "step": 9310
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4205781784271062e-05,
+      "loss": 0.1296,
+      "step": 9320
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4199564811936588e-05,
+      "loss": 0.1562,
+      "step": 9330
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4193347839602114e-05,
+      "loss": 0.0444,
+      "step": 9340
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4187130867267642e-05,
+      "loss": 0.0732,
+      "step": 9350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4180913894933168e-05,
+      "loss": 0.0838,
+      "step": 9360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4174696922598697e-05,
+      "loss": 0.1256,
+      "step": 9370
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4168479950264223e-05,
+      "loss": 0.0454,
+      "step": 9380
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.416226297792975e-05,
+      "loss": 0.181,
+      "step": 9390
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4156046005595277e-05,
+      "loss": 0.0668,
+      "step": 9400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4149829033260803e-05,
+      "loss": 0.1217,
+      "step": 9410
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.414361206092633e-05,
+      "loss": 0.115,
+      "step": 9420
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4137395088591856e-05,
+      "loss": 0.1578,
+      "step": 9430
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4131178116257386e-05,
+      "loss": 0.1059,
+      "step": 9440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.412496114392291e-05,
+      "loss": 0.1351,
+      "step": 9450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4118744171588436e-05,
+      "loss": 0.1246,
+      "step": 9460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4112527199253965e-05,
+      "loss": 0.1782,
+      "step": 9470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4106310226919491e-05,
+      "loss": 0.08,
+      "step": 9480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4100093254585019e-05,
+      "loss": 0.1594,
+      "step": 9490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4093876282250545e-05,
+      "loss": 0.1556,
+      "step": 9500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4087659309916072e-05,
+      "loss": 0.1313,
+      "step": 9510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4081442337581598e-05,
+      "loss": 0.0783,
+      "step": 9520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4075225365247124e-05,
+      "loss": 0.1787,
+      "step": 9530
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4069008392912654e-05,
+      "loss": 0.1155,
+      "step": 9540
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.406279142057818e-05,
+      "loss": 0.1876,
+      "step": 9550
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4056574448243707e-05,
+      "loss": 0.0751,
+      "step": 9560
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4050357475909233e-05,
+      "loss": 0.0745,
+      "step": 9570
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4044140503574761e-05,
+      "loss": 0.181,
+      "step": 9580
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4037923531240287e-05,
+      "loss": 0.1124,
+      "step": 9590
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4031706558905813e-05,
+      "loss": 0.1081,
+      "step": 9600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.402548958657134e-05,
+      "loss": 0.1191,
+      "step": 9610
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4019272614236867e-05,
+      "loss": 0.1654,
+      "step": 9620
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4013055641902396e-05,
+      "loss": 0.134,
+      "step": 9630
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4006838669567922e-05,
+      "loss": 0.1469,
+      "step": 9640
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4000621697233448e-05,
+      "loss": 0.1217,
+      "step": 9650
+    },
+    {
+      "epoch": 3.0,
+      "eval_FN": 193,
+      "eval_FP": 172,
+      "eval_TN": 5666,
+      "eval_TP": 402,
+      "eval_accuracy": 0.9432613088761076,
+      "eval_f1": 0.6877673224978614,
+      "eval_loss": 0.16720782220363617,
+      "eval_precision": 0.7003484320557491,
+      "eval_recall": 0.6756302521008404,
+      "eval_runtime": 44.2749,
+      "eval_samples_per_second": 145.297,
+      "eval_steps_per_second": 9.102,
+      "step": 9651
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 32170,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 4.062382072547328e+16,
+  "trial_name": null,
+  "trial_params": null
+}