diff --git "a/robin-7b/trainer_state.json" "b/robin-7b/trainer_state.json"
--- "a/robin-7b/trainer_state.json"
+++ "b/robin-7b/trainer_state.json"
@@ -1,1717 +1,13375 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 9.9867197875166,
-  "global_step": 5640,
+  "epoch": 5.0,
+  "global_step": 44500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.04,
-      "learning_rate": 9.411764705882353e-05,
-      "loss": 1.501,
+      "epoch": 0.0,
+      "learning_rate": 1.4981273408239701e-06,
+      "loss": 1.6359,
       "step": 20
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 0.00018823529411764707,
-      "loss": 1.4128,
+      "epoch": 0.0,
+      "learning_rate": 2.9962546816479402e-06,
+      "loss": 1.6528,
       "step": 40
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 0.0002823529411764706,
-      "loss": 1.325,
+      "epoch": 0.01,
+      "learning_rate": 4.49438202247191e-06,
+      "loss": 1.6445,
       "step": 60
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 0.00037647058823529414,
-      "loss": 1.2828,
+      "epoch": 0.01,
+      "learning_rate": 5.9925093632958805e-06,
+      "loss": 1.6479,
       "step": 80
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 0.00047058823529411766,
-      "loss": 1.2758,
+      "epoch": 0.01,
+      "learning_rate": 7.490636704119851e-06,
+      "loss": 1.608,
       "step": 100
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 0.0005647058823529412,
-      "loss": 1.2667,
+      "epoch": 0.01,
+      "learning_rate": 8.98876404494382e-06,
+      "loss": 1.6177,
       "step": 120
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 0.0006588235294117648,
-      "loss": 1.2504,
+      "epoch": 0.02,
+      "learning_rate": 1.0486891385767791e-05,
+      "loss": 1.5548,
       "step": 140
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 0.0007529411764705883,
-      "loss": 1.2394,
+      "epoch": 0.02,
+      "learning_rate": 1.1985018726591761e-05,
+      "loss": 1.527,
       "step": 160
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 0.0007999934028874321,
-      "loss": 1.228,
+      "epoch": 0.02,
+      "learning_rate": 1.348314606741573e-05,
+      "loss": 1.4841,
       "step": 180
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 0.0007999406272925394,
-      "loss": 1.2138,
+      "epoch": 0.02,
+      "learning_rate": 1.4981273408239702e-05,
+      "loss": 1.4651,
       "step": 200
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 0.0007998350830660272,
-      "loss": 1.2122,
+      "epoch": 0.02,
+      "learning_rate": 1.647940074906367e-05,
+      "loss": 1.4259,
       "step": 220
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 0.0007996767841335234,
-      "loss": 1.219,
+      "epoch": 0.03,
+      "learning_rate": 1.797752808988764e-05,
+      "loss": 1.4087,
       "step": 240
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 0.0007994657513811737,
-      "loss": 1.1998,
+      "epoch": 0.03,
+      "learning_rate": 1.9475655430711613e-05,
+      "loss": 1.4239,
       "step": 260
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 0.0007992020126528848,
-      "loss": 1.188,
+      "epoch": 0.03,
+      "learning_rate": 2.0973782771535582e-05,
+      "loss": 1.4119,
       "step": 280
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 0.0007988856027466511,
-      "loss": 1.1931,
+      "epoch": 0.03,
+      "learning_rate": 2.2471910112359552e-05,
+      "loss": 1.3844,
       "step": 300
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 0.000798516563409964,
-      "loss": 1.1753,
+      "epoch": 0.04,
+      "learning_rate": 2.3970037453183522e-05,
+      "loss": 1.3582,
       "step": 320
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 0.0007980949433343026,
-      "loss": 1.173,
+      "epoch": 0.04,
+      "learning_rate": 2.546816479400749e-05,
+      "loss": 1.3717,
       "step": 340
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 0.0007976207981487104,
-      "loss": 1.1637,
+      "epoch": 0.04,
+      "learning_rate": 2.696629213483146e-05,
+      "loss": 1.3852,
       "step": 360
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 0.0007970941904124546,
-      "loss": 1.1651,
+      "epoch": 0.04,
+      "learning_rate": 2.846441947565543e-05,
+      "loss": 1.3896,
       "step": 380
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 0.0007965151896067728,
-      "loss": 1.1559,
+      "epoch": 0.04,
+      "learning_rate": 2.9962546816479404e-05,
+      "loss": 1.3512,
       "step": 400
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 0.0007958838721257046,
-      "loss": 1.1658,
+      "epoch": 0.05,
+      "learning_rate": 3.1460674157303374e-05,
+      "loss": 1.3857,
       "step": 420
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 0.0007952003212660127,
-      "loss": 1.1496,
+      "epoch": 0.05,
+      "learning_rate": 3.295880149812734e-05,
+      "loss": 1.3396,
       "step": 440
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 0.0007944646272161933,
-      "loss": 1.1471,
+      "epoch": 0.05,
+      "learning_rate": 3.445692883895131e-05,
+      "loss": 1.3324,
       "step": 460
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 0.0007936768870445747,
-      "loss": 1.1311,
+      "epoch": 0.05,
+      "learning_rate": 3.595505617977528e-05,
+      "loss": 1.3522,
       "step": 480
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 0.0007928372046865116,
-      "loss": 1.1301,
+      "epoch": 0.06,
+      "learning_rate": 3.745318352059925e-05,
+      "loss": 1.3429,
       "step": 500
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 0.0007919456909306711,
-      "loss": 1.134,
+      "epoch": 0.06,
+      "learning_rate": 3.8951310861423226e-05,
+      "loss": 1.3368,
       "step": 520
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 0.0007910024634044154,
-      "loss": 1.1235,
+      "epoch": 0.06,
+      "learning_rate": 4.044943820224719e-05,
+      "loss": 1.3165,
       "step": 540
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 0.0007900076465582816,
-      "loss": 1.1239,
+      "epoch": 0.06,
+      "learning_rate": 4.1947565543071165e-05,
+      "loss": 1.3356,
       "step": 560
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 0.0007889613716495616,
-      "loss": 1.0878,
+      "epoch": 0.07,
+      "learning_rate": 4.344569288389513e-05,
+      "loss": 1.2966,
       "step": 580
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 0.0007878637767249839,
-      "loss": 1.0879,
+      "epoch": 0.07,
+      "learning_rate": 4.4943820224719104e-05,
+      "loss": 1.3255,
       "step": 600
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 0.0007867150066024996,
-      "loss": 1.0671,
+      "epoch": 0.07,
+      "learning_rate": 4.644194756554308e-05,
+      "loss": 1.2908,
       "step": 620
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 0.0007855152128521754,
-      "loss": 1.0689,
+      "epoch": 0.07,
+      "learning_rate": 4.7940074906367044e-05,
+      "loss": 1.2983,
       "step": 640
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 0.0007842645537761941,
-      "loss": 1.0794,
+      "epoch": 0.07,
+      "learning_rate": 4.943820224719101e-05,
+      "loss": 1.2979,
       "step": 660
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 0.0007829631943879694,
-      "loss": 1.0653,
+      "epoch": 0.08,
+      "learning_rate": 5.093632958801498e-05,
+      "loss": 1.3171,
       "step": 680
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 0.0007816113063903726,
-      "loss": 1.066,
+      "epoch": 0.08,
+      "learning_rate": 5.243445692883895e-05,
+      "loss": 1.3026,
       "step": 700
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 0.0007802090681530788,
-      "loss": 1.0675,
+      "epoch": 0.08,
+      "learning_rate": 5.393258426966292e-05,
+      "loss": 1.3105,
       "step": 720
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 0.0007787566646890325,
-      "loss": 1.0598,
+      "epoch": 0.08,
+      "learning_rate": 5.5430711610486895e-05,
+      "loss": 1.3146,
       "step": 740
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 0.0007772542876300359,
-      "loss": 1.0669,
+      "epoch": 0.09,
+      "learning_rate": 5.692883895131086e-05,
+      "loss": 1.2834,
       "step": 760
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 0.0007757021352014663,
-      "loss": 1.0558,
+      "epoch": 0.09,
+      "learning_rate": 5.8426966292134835e-05,
+      "loss": 1.2993,
       "step": 780
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 0.0007741004121961207,
-      "loss": 1.0578,
+      "epoch": 0.09,
+      "learning_rate": 5.992509363295881e-05,
+      "loss": 1.287,
       "step": 800
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 0.0007724493299471956,
-      "loss": 1.056,
+      "epoch": 0.09,
+      "learning_rate": 6.142322097378277e-05,
+      "loss": 1.293,
       "step": 820
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 0.0007707491063004035,
-      "loss": 1.0491,
+      "epoch": 0.09,
+      "learning_rate": 6.292134831460675e-05,
+      "loss": 1.2556,
       "step": 840
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 0.0007689999655852306,
-      "loss": 1.0497,
+      "epoch": 0.1,
+      "learning_rate": 6.441947565543071e-05,
+      "loss": 1.2979,
       "step": 860
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 0.0007672021385853376,
-      "loss": 1.0393,
+      "epoch": 0.1,
+      "learning_rate": 6.591760299625468e-05,
+      "loss": 1.3002,
       "step": 880
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 0.0007653558625081099,
-      "loss": 1.0379,
+      "epoch": 0.1,
+      "learning_rate": 6.741573033707866e-05,
+      "loss": 1.2408,
       "step": 900
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 0.0007634613809533613,
-      "loss": 1.049,
+      "epoch": 0.1,
+      "learning_rate": 6.891385767790263e-05,
+      "loss": 1.2753,
       "step": 920
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 0.0007615189438811918,
-      "loss": 1.0594,
+      "epoch": 0.11,
+      "learning_rate": 7.04119850187266e-05,
+      "loss": 1.2671,
       "step": 940
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 0.0007595288075790085,
-      "loss": 1.0375,
+      "epoch": 0.11,
+      "learning_rate": 7.191011235955056e-05,
+      "loss": 1.2925,
       "step": 960
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 0.0007574912346277103,
-      "loss": 1.0453,
+      "epoch": 0.11,
+      "learning_rate": 7.340823970037454e-05,
+      "loss": 1.2709,
       "step": 980
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 0.0007554064938670426,
-      "loss": 1.0348,
+      "epoch": 0.11,
+      "learning_rate": 7.49063670411985e-05,
+      "loss": 1.2614,
       "step": 1000
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 0.0007532748603601265,
-      "loss": 1.0441,
+      "epoch": 0.11,
+      "learning_rate": 7.640449438202247e-05,
+      "loss": 1.2979,
       "step": 1020
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 0.0007510966153571667,
-      "loss": 1.0237,
+      "epoch": 0.12,
+      "learning_rate": 7.790262172284645e-05,
+      "loss": 1.2534,
       "step": 1040
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 0.000748872046258343,
-      "loss": 1.0478,
+      "epoch": 0.12,
+      "learning_rate": 7.940074906367042e-05,
+      "loss": 1.2425,
       "step": 1060
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 0.0007466014465758899,
-      "loss": 1.0289,
+      "epoch": 0.12,
+      "learning_rate": 8.089887640449438e-05,
+      "loss": 1.2671,
       "step": 1080
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 0.0007442851158953712,
-      "loss": 1.026,
+      "epoch": 0.12,
+      "learning_rate": 8.239700374531836e-05,
+      "loss": 1.247,
       "step": 1100
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 0.0007419233598361512,
-      "loss": 1.0244,
+      "epoch": 0.13,
+      "learning_rate": 8.389513108614233e-05,
+      "loss": 1.2785,
       "step": 1120
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 0.0007395164900110721,
-      "loss": 0.9968,
+      "epoch": 0.13,
+      "learning_rate": 8.53932584269663e-05,
+      "loss": 1.2611,
       "step": 1140
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 0.0007370648239853385,
-      "loss": 0.9798,
+      "epoch": 0.13,
+      "learning_rate": 8.689138576779026e-05,
+      "loss": 1.238,
       "step": 1160
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 0.0007345686852346176,
-      "loss": 0.9529,
+      "epoch": 0.13,
+      "learning_rate": 8.838951310861424e-05,
+      "loss": 1.2352,
       "step": 1180
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 0.0007320284031023603,
-      "loss": 0.9666,
+      "epoch": 0.13,
+      "learning_rate": 8.988764044943821e-05,
+      "loss": 1.2389,
       "step": 1200
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 0.000729444312756346,
-      "loss": 0.9797,
+      "epoch": 0.14,
+      "learning_rate": 9.138576779026217e-05,
+      "loss": 1.2519,
       "step": 1220
     },
     {
-      "epoch": 2.2,
-      "learning_rate": 0.0007268167551444611,
-      "loss": 0.9674,
+      "epoch": 0.14,
+      "learning_rate": 9.288389513108615e-05,
+      "loss": 1.2466,
       "step": 1240
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 0.0007241460769497138,
-      "loss": 0.9666,
+      "epoch": 0.14,
+      "learning_rate": 9.438202247191012e-05,
+      "loss": 1.2414,
       "step": 1260
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 0.0007214326305444917,
-      "loss": 0.9785,
+      "epoch": 0.14,
+      "learning_rate": 9.588014981273409e-05,
+      "loss": 1.2565,
       "step": 1280
     },
     {
-      "epoch": 2.3,
-      "learning_rate": 0.0007186767739440701,
-      "loss": 0.9629,
+      "epoch": 0.15,
+      "learning_rate": 9.737827715355807e-05,
+      "loss": 1.2661,
       "step": 1300
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 0.0007158788707593748,
-      "loss": 0.973,
+      "epoch": 0.15,
+      "learning_rate": 9.887640449438202e-05,
+      "loss": 1.2708,
       "step": 1320
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 0.0007130392901490069,
-      "loss": 0.9649,
+      "epoch": 0.15,
+      "learning_rate": 9.999999668932716e-05,
+      "loss": 1.2292,
       "step": 1340
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 0.0007101584067705355,
-      "loss": 0.9766,
+      "epoch": 0.15,
+      "learning_rate": 9.999991723320065e-05,
+      "loss": 1.2506,
       "step": 1360
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 0.0007072366007310646,
-      "loss": 0.954,
+      "epoch": 0.16,
+      "learning_rate": 9.999973183573581e-05,
+      "loss": 1.2434,
       "step": 1380
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 0.0007042742575370822,
-      "loss": 0.9576,
+      "epoch": 0.16,
+      "learning_rate": 9.999944049732545e-05,
+      "loss": 1.2425,
       "step": 1400
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 0.0007012717680435956,
-      "loss": 0.9783,
+      "epoch": 0.16,
+      "learning_rate": 9.99990432185869e-05,
+      "loss": 1.2279,
       "step": 1420
     },
     {
-      "epoch": 2.55,
-      "learning_rate": 0.0006982295284025612,
-      "loss": 0.9553,
+      "epoch": 0.16,
+      "learning_rate": 9.999854000036192e-05,
+      "loss": 1.2039,
       "step": 1440
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 0.0006951479400106161,
-      "loss": 0.951,
+      "epoch": 0.16,
+      "learning_rate": 9.999793084371672e-05,
+      "loss": 1.2298,
       "step": 1460
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 0.000692027409456118,
-      "loss": 0.9647,
+      "epoch": 0.17,
+      "learning_rate": 9.999721574994201e-05,
+      "loss": 1.2467,
       "step": 1480
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 0.0006888683484654981,
-      "loss": 0.9656,
+      "epoch": 0.17,
+      "learning_rate": 9.999639472055294e-05,
+      "loss": 1.2039,
       "step": 1500
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 0.0006856711738489386,
-      "loss": 0.9507,
+      "epoch": 0.17,
+      "learning_rate": 9.999546775728917e-05,
+      "loss": 1.2668,
       "step": 1520
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 0.0006824363074453778,
-      "loss": 0.9496,
+      "epoch": 0.17,
+      "learning_rate": 9.999443486211473e-05,
+      "loss": 1.2045,
       "step": 1540
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 0.0006791641760668519,
-      "loss": 0.9571,
+      "epoch": 0.18,
+      "learning_rate": 9.99932960372182e-05,
+      "loss": 1.2156,
       "step": 1560
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 0.0006758552114421815,
-      "loss": 0.9494,
+      "epoch": 0.18,
+      "learning_rate": 9.99920512850125e-05,
+      "loss": 1.2157,
       "step": 1580
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 0.0006725098501600088,
-      "loss": 0.9556,
+      "epoch": 0.18,
+      "learning_rate": 9.999070060813509e-05,
+      "loss": 1.2027,
       "step": 1600
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 0.0006691285336111928,
-      "loss": 0.9411,
+      "epoch": 0.18,
+      "learning_rate": 9.99892440094478e-05,
+      "loss": 1.222,
       "step": 1620
     },
     {
-      "epoch": 2.9,
-      "learning_rate": 0.0006657117079305725,
-      "loss": 0.9501,
+      "epoch": 0.18,
+      "learning_rate": 9.998768149203695e-05,
+      "loss": 1.2139,
       "step": 1640
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 0.0006622598239381033,
-      "loss": 0.9598,
+      "epoch": 0.19,
+      "learning_rate": 9.998601305921322e-05,
+      "loss": 1.2042,
       "step": 1660
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 0.0006587733370793743,
-      "loss": 0.9599,
+      "epoch": 0.19,
+      "learning_rate": 9.998423871451174e-05,
+      "loss": 1.2379,
       "step": 1680
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 0.0006552527073655178,
-      "loss": 0.9306,
+      "epoch": 0.19,
+      "learning_rate": 9.998235846169204e-05,
+      "loss": 1.1764,
       "step": 1700
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 0.0006516983993125138,
-      "loss": 0.9013,
+      "epoch": 0.19,
+      "learning_rate": 9.998037230473809e-05,
+      "loss": 1.2254,
       "step": 1720
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 0.0006481108818799015,
-      "loss": 0.8798,
+      "epoch": 0.2,
+      "learning_rate": 9.997828024785817e-05,
+      "loss": 1.1891,
       "step": 1740
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 0.0006444906284089044,
-      "loss": 0.903,
+      "epoch": 0.2,
+      "learning_rate": 9.997608229548504e-05,
+      "loss": 1.1889,
       "step": 1760
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 0.000640838116559977,
-      "loss": 0.9017,
+      "epoch": 0.2,
+      "learning_rate": 9.997377845227576e-05,
+      "loss": 1.2035,
       "step": 1780
     },
     {
-      "epoch": 3.19,
-      "learning_rate": 0.0006371538282497815,
-      "loss": 0.9051,
+      "epoch": 0.2,
+      "learning_rate": 9.997136872311177e-05,
+      "loss": 1.2186,
       "step": 1800
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 0.0006334382495876036,
-      "loss": 0.8966,
+      "epoch": 0.2,
+      "learning_rate": 9.996885311309891e-05,
+      "loss": 1.1837,
       "step": 1820
     },
     {
-      "epoch": 3.26,
-      "learning_rate": 0.0006296918708112143,
-      "loss": 0.8863,
+      "epoch": 0.21,
+      "learning_rate": 9.996623162756733e-05,
+      "loss": 1.183,
       "step": 1840
     },
     {
-      "epoch": 3.29,
-      "learning_rate": 0.0006259151862221875,
-      "loss": 0.8926,
+      "epoch": 0.21,
+      "learning_rate": 9.996350427207148e-05,
+      "loss": 1.1707,
       "step": 1860
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 0.0006221086941206817,
-      "loss": 0.908,
+      "epoch": 0.21,
+      "learning_rate": 9.99606710523902e-05,
+      "loss": 1.2172,
       "step": 1880
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 0.0006182728967396925,
-      "loss": 0.9007,
+      "epoch": 0.21,
+      "learning_rate": 9.995773197452657e-05,
+      "loss": 1.1877,
       "step": 1900
     },
     {
-      "epoch": 3.4,
-      "learning_rate": 0.0006144083001787886,
-      "loss": 0.8872,
+      "epoch": 0.22,
+      "learning_rate": 9.995468704470802e-05,
+      "loss": 1.2208,
       "step": 1920
     },
     {
-      "epoch": 3.44,
-      "learning_rate": 0.0006105154143373362,
-      "loss": 0.8984,
+      "epoch": 0.22,
+      "learning_rate": 9.995153626938623e-05,
+      "loss": 1.1727,
       "step": 1940
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 0.0006065947528472215,
-      "loss": 0.9123,
+      "epoch": 0.22,
+      "learning_rate": 9.994827965523716e-05,
+      "loss": 1.2127,
       "step": 1960
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 0.0006026468330050827,
-      "loss": 0.8929,
+      "epoch": 0.22,
+      "learning_rate": 9.994491720916102e-05,
+      "loss": 1.1912,
       "step": 1980
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 0.0005986721757040564,
-      "loss": 0.9145,
+      "epoch": 0.22,
+      "learning_rate": 9.994144893828226e-05,
+      "loss": 1.1852,
       "step": 2000
     },
     {
-      "epoch": 3.58,
-      "learning_rate": 0.0005946713053650507,
-      "loss": 0.8867,
+      "epoch": 0.23,
+      "learning_rate": 9.993787484994957e-05,
+      "loss": 1.1689,
       "step": 2020
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 0.0005906447498675521,
-      "loss": 0.8914,
+      "epoch": 0.23,
+      "learning_rate": 9.993419495173582e-05,
+      "loss": 1.19,
       "step": 2040
     },
     {
-      "epoch": 3.65,
-      "learning_rate": 0.0005865930404799774,
-      "loss": 0.8946,
+      "epoch": 0.23,
+      "learning_rate": 9.99304092514381e-05,
+      "loss": 1.1833,
       "step": 2060
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 0.0005825167117895765,
-      "loss": 0.892,
+      "epoch": 0.23,
+      "learning_rate": 9.992651775707768e-05,
+      "loss": 1.1944,
       "step": 2080
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 0.0005784163016318987,
-      "loss": 0.8875,
+      "epoch": 0.24,
+      "learning_rate": 9.992252047689997e-05,
+      "loss": 1.1592,
       "step": 2100
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 0.0005742923510198303,
-      "loss": 0.888,
+      "epoch": 0.24,
+      "learning_rate": 9.991841741937448e-05,
+      "loss": 1.1744,
       "step": 2120
     },
     {
-      "epoch": 3.79,
-      "learning_rate": 0.0005701454040722124,
-      "loss": 0.9078,
+      "epoch": 0.24,
+      "learning_rate": 9.991420859319496e-05,
+      "loss": 1.1884,
       "step": 2140
     },
     {
-      "epoch": 3.82,
-      "learning_rate": 0.0005659760079420498,
-      "loss": 0.9027,
+      "epoch": 0.24,
+      "learning_rate": 9.990989400727916e-05,
+      "loss": 1.1372,
       "step": 2160
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 0.000561784712744318,
-      "loss": 0.8997,
+      "epoch": 0.24,
+      "learning_rate": 9.990547367076896e-05,
+      "loss": 1.1767,
       "step": 2180
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 0.0005575720714833808,
-      "loss": 0.9053,
+      "epoch": 0.25,
+      "learning_rate": 9.990094759303033e-05,
+      "loss": 1.1837,
       "step": 2200
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 0.0005533386399800275,
-      "loss": 0.9054,
+      "epoch": 0.25,
+      "learning_rate": 9.989631578365322e-05,
+      "loss": 1.1564,
       "step": 2220
     },
     {
-      "epoch": 3.97,
-      "learning_rate": 0.0005490849767981348,
-      "loss": 0.8988,
+      "epoch": 0.25,
+      "learning_rate": 9.989157825245167e-05,
+      "loss": 1.1807,
       "step": 2240
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 0.0005448116431709716,
-      "loss": 0.8903,
+      "epoch": 0.25,
+      "learning_rate": 9.98867350094637e-05,
+      "loss": 1.1898,
       "step": 2260
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 0.0005405192029271477,
-      "loss": 0.8373,
+      "epoch": 0.26,
+      "learning_rate": 9.988178606495132e-05,
+      "loss": 1.2028,
       "step": 2280
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 0.0005362082224162223,
-      "loss": 0.8336,
+      "epoch": 0.26,
+      "learning_rate": 9.98767314294005e-05,
+      "loss": 1.1629,
       "step": 2300
     },
     {
-      "epoch": 4.11,
-      "learning_rate": 0.0005318792704339792,
-      "loss": 0.8483,
+      "epoch": 0.26,
+      "learning_rate": 9.987157111352117e-05,
+      "loss": 1.1963,
       "step": 2320
     },
     {
-      "epoch": 4.14,
-      "learning_rate": 0.0005275329181473787,
-      "loss": 0.8453,
+      "epoch": 0.26,
+      "learning_rate": 9.986630512824715e-05,
+      "loss": 1.1642,
       "step": 2340
     },
     {
-      "epoch": 4.18,
-      "learning_rate": 0.0005231697390191976,
-      "loss": 0.8351,
+      "epoch": 0.27,
+      "learning_rate": 9.986093348473617e-05,
+      "loss": 1.1624,
       "step": 2360
     },
     {
-      "epoch": 4.21,
-      "learning_rate": 0.000518790308732366,
-      "loss": 0.8329,
+      "epoch": 0.27,
+      "learning_rate": 9.985545619436984e-05,
+      "loss": 1.1595,
       "step": 2380
     },
     {
-      "epoch": 4.25,
-      "learning_rate": 0.0005143952051140103,
-      "loss": 0.8394,
+      "epoch": 0.27,
+      "learning_rate": 9.984987326875359e-05,
+      "loss": 1.1985,
       "step": 2400
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 0.000509985008059215,
-      "loss": 0.8526,
+      "epoch": 0.27,
+      "learning_rate": 9.984418471971671e-05,
+      "loss": 1.1912,
       "step": 2420
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 0.0005055602994545098,
-      "loss": 0.826,
+      "epoch": 0.27,
+      "learning_rate": 9.983839055931226e-05,
+      "loss": 1.2146,
       "step": 2440
     },
     {
-      "epoch": 4.36,
-      "learning_rate": 0.0005011216631010953,
-      "loss": 0.849,
+      "epoch": 0.28,
+      "learning_rate": 9.983249079981709e-05,
+      "loss": 1.183,
       "step": 2460
     },
     {
-      "epoch": 4.39,
-      "learning_rate": 0.0004966696846378156,
-      "loss": 0.8507,
+      "epoch": 0.28,
+      "learning_rate": 9.982648545373177e-05,
+      "loss": 1.1561,
       "step": 2480
     },
     {
-      "epoch": 4.43,
-      "learning_rate": 0.000492204951463888,
-      "loss": 0.8461,
+      "epoch": 0.28,
+      "learning_rate": 9.982037453378063e-05,
+      "loss": 1.1517,
       "step": 2500
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 0.00048772805266140154,
-      "loss": 0.8533,
+      "epoch": 0.28,
+      "learning_rate": 9.981415805291168e-05,
+      "loss": 1.1485,
       "step": 2520
     },
     {
-      "epoch": 4.5,
-      "learning_rate": 0.00048323957891759203,
-      "loss": 0.8384,
+      "epoch": 0.29,
+      "learning_rate": 9.980783602429656e-05,
+      "loss": 1.1721,
       "step": 2540
     },
     {
-      "epoch": 4.53,
-      "learning_rate": 0.00047874012244690696,
-      "loss": 0.842,
+      "epoch": 0.29,
+      "learning_rate": 9.98014084613306e-05,
+      "loss": 1.1825,
       "step": 2560
     },
     {
-      "epoch": 4.57,
-      "learning_rate": 0.000474230276912867,
-      "loss": 0.8608,
+      "epoch": 0.29,
+      "learning_rate": 9.979487537763269e-05,
+      "loss": 1.1338,
       "step": 2580
     },
     {
-      "epoch": 4.6,
-      "learning_rate": 0.00046971063734973833,
-      "loss": 0.8562,
+      "epoch": 0.29,
+      "learning_rate": 9.978823678704533e-05,
+      "loss": 1.193,
       "step": 2600
     },
     {
-      "epoch": 4.64,
-      "learning_rate": 0.0004651818000840229,
-      "loss": 0.8594,
+      "epoch": 0.29,
+      "learning_rate": 9.978149270363462e-05,
+      "loss": 1.1809,
       "step": 2620
     },
     {
-      "epoch": 4.67,
-      "learning_rate": 0.0004606443626557778,
-      "loss": 0.8608,
+      "epoch": 0.3,
+      "learning_rate": 9.977464314169005e-05,
+      "loss": 1.1425,
       "step": 2640
     },
     {
-      "epoch": 4.71,
-      "learning_rate": 0.0004560989237397758,
-      "loss": 0.8486,
+      "epoch": 0.3,
+      "learning_rate": 9.976768811572473e-05,
+      "loss": 1.1428,
       "step": 2660
     },
     {
-      "epoch": 4.75,
-      "learning_rate": 0.00045154608306651514,
-      "loss": 0.869,
+      "epoch": 0.3,
+      "learning_rate": 9.976062764047515e-05,
+      "loss": 1.1511,
       "step": 2680
     },
     {
-      "epoch": 4.78,
-      "learning_rate": 0.0004469864413430907,
-      "loss": 0.8482,
+      "epoch": 0.3,
+      "learning_rate": 9.975346173090128e-05,
+      "loss": 1.161,
       "step": 2700
     },
     {
-      "epoch": 4.82,
-      "learning_rate": 0.00044242060017393573,
-      "loss": 0.8583,
+      "epoch": 0.31,
+      "learning_rate": 9.974619040218644e-05,
+      "loss": 1.168,
       "step": 2720
     },
     {
-      "epoch": 4.85,
-      "learning_rate": 0.00043784916198144543,
-      "loss": 0.8582,
+      "epoch": 0.31,
+      "learning_rate": 9.973881366973738e-05,
+      "loss": 1.1474,
       "step": 2740
     },
     {
-      "epoch": 4.89,
-      "learning_rate": 0.00043327272992649317,
-      "loss": 0.8504,
+      "epoch": 0.31,
+      "learning_rate": 9.973133154918413e-05,
+      "loss": 1.1605,
       "step": 2760
     },
     {
-      "epoch": 4.92,
-      "learning_rate": 0.00042869190782884794,
-      "loss": 0.8592,
+      "epoch": 0.31,
+      "learning_rate": 9.972374405638e-05,
+      "loss": 1.1591,
       "step": 2780
     },
     {
-      "epoch": 4.96,
-      "learning_rate": 0.00042410730008750623,
-      "loss": 0.8545,
+      "epoch": 0.31,
+      "learning_rate": 9.971605120740166e-05,
+      "loss": 1.1494,
       "step": 2800
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 0.00041951951160094664,
-      "loss": 0.855,
+      "epoch": 0.32,
+      "learning_rate": 9.970825301854889e-05,
+      "loss": 1.1596,
       "step": 2820
     },
     {
-      "epoch": 5.03,
-      "learning_rate": 0.00041492914768731927,
-      "loss": 0.7869,
+      "epoch": 0.32,
+      "learning_rate": 9.970034950634478e-05,
+      "loss": 1.1543,
       "step": 2840
     },
     {
-      "epoch": 5.06,
-      "learning_rate": 0.0004103368140045789,
-      "loss": 0.8083,
+      "epoch": 0.32,
+      "learning_rate": 9.96923406875355e-05,
+      "loss": 1.128,
       "step": 2860
     },
     {
-      "epoch": 5.1,
-      "learning_rate": 0.00040574311647057366,
-      "loss": 0.8108,
+      "epoch": 0.32,
+      "learning_rate": 9.968422657909037e-05,
+      "loss": 1.1567,
       "step": 2880
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 0.00040114866118310045,
-      "loss": 0.7968,
+      "epoch": 0.33,
+      "learning_rate": 9.967600719820183e-05,
+      "loss": 1.1439,
       "step": 2900
     },
     {
-      "epoch": 5.17,
-      "learning_rate": 0.0003965540543399344,
-      "loss": 0.8016,
+      "epoch": 0.33,
+      "learning_rate": 9.966768256228536e-05,
+      "loss": 1.1156,
       "step": 2920
     },
     {
-      "epoch": 5.21,
-      "learning_rate": 0.00039195990215884756,
-      "loss": 0.7967,
+      "epoch": 0.33,
+      "learning_rate": 9.965925268897942e-05,
+      "loss": 1.1695,
       "step": 2940
     },
     {
-      "epoch": 5.24,
-      "learning_rate": 0.00038736681079762293,
-      "loss": 0.8096,
+      "epoch": 0.33,
+      "learning_rate": 9.96507175961455e-05,
+      "loss": 1.1286,
       "step": 2960
     },
     {
-      "epoch": 5.28,
-      "learning_rate": 0.0003827753862740779,
-      "loss": 0.8073,
+      "epoch": 0.33,
+      "learning_rate": 9.964207730186804e-05,
+      "loss": 1.1687,
       "step": 2980
     },
     {
-      "epoch": 5.31,
-      "learning_rate": 0.0003781862343861055,
-      "loss": 0.804,
+      "epoch": 0.34,
+      "learning_rate": 9.963333182445429e-05,
+      "loss": 1.1401,
       "step": 3000
     },
     {
-      "epoch": 5.35,
-      "learning_rate": 0.00037359996063174425,
-      "loss": 0.8119,
+      "epoch": 0.34,
+      "learning_rate": 9.962448118243451e-05,
+      "loss": 1.1419,
       "step": 3020
     },
     {
-      "epoch": 5.38,
-      "learning_rate": 0.0003690171701292887,
-      "loss": 0.7997,
+      "epoch": 0.34,
+      "learning_rate": 9.961552539456163e-05,
+      "loss": 1.1224,
       "step": 3040
     },
     {
-      "epoch": 5.42,
-      "learning_rate": 0.0003644384675374489,
-      "loss": 0.8202,
+      "epoch": 0.34,
+      "learning_rate": 9.96064644798115e-05,
+      "loss": 1.1506,
       "step": 3060
     },
     {
-      "epoch": 5.45,
-      "learning_rate": 0.0003598644569755713,
-      "loss": 0.815,
+      "epoch": 0.35,
+      "learning_rate": 9.959729845738264e-05,
+      "loss": 1.143,
       "step": 3080
     },
     {
-      "epoch": 5.49,
-      "learning_rate": 0.00035529574194393033,
-      "loss": 0.825,
+      "epoch": 0.35,
+      "learning_rate": 9.958802734669633e-05,
+      "loss": 1.1684,
       "step": 3100
     },
     {
-      "epoch": 5.52,
-      "learning_rate": 0.00035073292524410207,
-      "loss": 0.8171,
+      "epoch": 0.35,
+      "learning_rate": 9.957865116739641e-05,
+      "loss": 1.1226,
       "step": 3120
     },
     {
-      "epoch": 5.56,
-      "learning_rate": 0.00034617660889943,
-      "loss": 0.7921,
+      "epoch": 0.35,
+      "learning_rate": 9.956916993934947e-05,
+      "loss": 1.1404,
       "step": 3140
     },
     {
-      "epoch": 5.6,
-      "learning_rate": 0.00034162739407559285,
-      "loss": 0.8299,
+      "epoch": 0.36,
+      "learning_rate": 9.95595836826446e-05,
+      "loss": 1.1483,
       "step": 3160
     },
     {
-      "epoch": 5.63,
-      "learning_rate": 0.0003370858810012869,
-      "loss": 0.811,
+      "epoch": 0.36,
+      "learning_rate": 9.954989241759346e-05,
+      "loss": 1.137,
       "step": 3180
     },
     {
-      "epoch": 5.67,
-      "learning_rate": 0.00033255266888903006,
-      "loss": 0.8093,
+      "epoch": 0.36,
+      "learning_rate": 9.954009616473019e-05,
+      "loss": 1.1019,
       "step": 3200
     },
     {
-      "epoch": 5.7,
-      "learning_rate": 0.00032802835585610225,
-      "loss": 0.8106,
+      "epoch": 0.36,
+      "learning_rate": 9.95301949448114e-05,
+      "loss": 1.1063,
       "step": 3220
     },
     {
-      "epoch": 5.74,
-      "learning_rate": 0.00032351353884562783,
-      "loss": 0.8053,
+      "epoch": 0.36,
+      "learning_rate": 9.952018877881606e-05,
+      "loss": 1.1487,
       "step": 3240
     },
     {
-      "epoch": 5.77,
-      "learning_rate": 0.00031900881354781556,
-      "loss": 0.8161,
+      "epoch": 0.37,
+      "learning_rate": 9.951007768794558e-05,
+      "loss": 1.128,
       "step": 3260
     },
     {
-      "epoch": 5.81,
-      "learning_rate": 0.00031451477432136154,
-      "loss": 0.8186,
+      "epoch": 0.37,
+      "learning_rate": 9.949986169362362e-05,
+      "loss": 1.1343,
       "step": 3280
     },
     {
-      "epoch": 5.84,
-      "learning_rate": 0.0003100320141150293,
-      "loss": 0.8046,
+      "epoch": 0.37,
+      "learning_rate": 9.948954081749616e-05,
+      "loss": 1.1342,
       "step": 3300
     },
     {
-      "epoch": 5.88,
-      "learning_rate": 0.00030556112438941526,
-      "loss": 0.8236,
+      "epoch": 0.37,
+      "learning_rate": 9.947911508143135e-05,
+      "loss": 1.1387,
       "step": 3320
     },
     {
-      "epoch": 5.91,
-      "learning_rate": 0.00030110269503891084,
-      "loss": 0.8057,
+      "epoch": 0.38,
+      "learning_rate": 9.946858450751958e-05,
+      "loss": 1.1217,
       "step": 3340
     },
     {
-      "epoch": 5.95,
-      "learning_rate": 0.0002966573143138713,
-      "loss": 0.8109,
+      "epoch": 0.38,
+      "learning_rate": 9.945794911807334e-05,
+      "loss": 1.1276,
       "step": 3360
     },
     {
-      "epoch": 5.98,
-      "learning_rate": 0.00029222556874300036,
-      "loss": 0.8163,
+      "epoch": 0.38,
+      "learning_rate": 9.944720893562722e-05,
+      "loss": 1.1567,
       "step": 3380
     },
     {
-      "epoch": 6.02,
-      "learning_rate": 0.0002878080430559646,
-      "loss": 0.7901,
+      "epoch": 0.38,
+      "learning_rate": 9.943636398293785e-05,
+      "loss": 1.1307,
       "step": 3400
     },
     {
-      "epoch": 6.06,
-      "learning_rate": 0.0002834053201062417,
-      "loss": 0.7749,
+      "epoch": 0.38,
+      "learning_rate": 9.942541428298384e-05,
+      "loss": 1.1264,
       "step": 3420
     },
     {
-      "epoch": 6.09,
-      "learning_rate": 0.00027901798079421977,
-      "loss": 0.7775,
+      "epoch": 0.39,
+      "learning_rate": 9.941435985896573e-05,
+      "loss": 1.1148,
       "step": 3440
     },
     {
-      "epoch": 6.13,
-      "learning_rate": 0.0002746466039905513,
-      "loss": 0.7589,
+      "epoch": 0.39,
+      "learning_rate": 9.940320073430598e-05,
+      "loss": 1.1417,
       "step": 3460
     },
     {
-      "epoch": 6.16,
-      "learning_rate": 0.000270291766459777,
-      "loss": 0.7742,
+      "epoch": 0.39,
+      "learning_rate": 9.93919369326489e-05,
+      "loss": 1.1565,
       "step": 3480
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 0.00026595404278422684,
-      "loss": 0.7914,
+      "epoch": 0.39,
+      "learning_rate": 9.938056847786053e-05,
+      "loss": 1.1517,
       "step": 3500
     },
     {
-      "epoch": 6.23,
-      "learning_rate": 0.00026163400528820836,
-      "loss": 0.7909,
+      "epoch": 0.4,
+      "learning_rate": 9.936909539402874e-05,
+      "loss": 1.1505,
       "step": 3520
     },
     {
-      "epoch": 6.27,
-      "learning_rate": 0.0002573322239624947,
-      "loss": 0.7653,
+      "epoch": 0.4,
+      "learning_rate": 9.935751770546302e-05,
+      "loss": 1.1276,
       "step": 3540
     },
     {
-      "epoch": 6.3,
-      "learning_rate": 0.000253049266389118,
-      "loss": 0.7779,
+      "epoch": 0.4,
+      "learning_rate": 9.934583543669453e-05,
+      "loss": 1.1169,
       "step": 3560
     },
     {
-      "epoch": 6.34,
-      "learning_rate": 0.0002487856976664831,
-      "loss": 0.7771,
+      "epoch": 0.4,
+      "learning_rate": 9.933404861247603e-05,
+      "loss": 1.1185,
       "step": 3580
     },
     {
-      "epoch": 6.37,
-      "learning_rate": 0.00024454208033480683,
-      "loss": 0.771,
+      "epoch": 0.4,
+      "learning_rate": 9.93221572577818e-05,
+      "loss": 1.1566,
       "step": 3600
     },
     {
-      "epoch": 6.41,
-      "learning_rate": 0.00024031897430189695,
-      "loss": 0.7788,
+      "epoch": 0.41,
+      "learning_rate": 9.931016139780758e-05,
+      "loss": 1.1288,
       "step": 3620
     },
     {
-      "epoch": 6.45,
-      "learning_rate": 0.00023611693676927606,
-      "loss": 0.7798,
+      "epoch": 0.41,
+      "learning_rate": 9.929806105797058e-05,
+      "loss": 1.137,
       "step": 3640
     },
     {
-      "epoch": 6.48,
-      "learning_rate": 0.00023193652215866429,
-      "loss": 0.7779,
+      "epoch": 0.41,
+      "learning_rate": 9.928585626390935e-05,
+      "loss": 1.1266,
       "step": 3660
     },
     {
-      "epoch": 6.52,
-      "learning_rate": 0.00022777828203882875,
-      "loss": 0.7784,
+      "epoch": 0.41,
+      "learning_rate": 9.927354704148382e-05,
+      "loss": 1.1436,
       "step": 3680
     },
     {
-      "epoch": 6.55,
-      "learning_rate": 0.00022364276505280794,
-      "loss": 0.7946,
+      "epoch": 0.42,
+      "learning_rate": 9.926113341677507e-05,
+      "loss": 1.0938,
       "step": 3700
     },
     {
-      "epoch": 6.59,
-      "learning_rate": 0.0002195305168455239,
-      "loss": 0.774,
+      "epoch": 0.42,
+      "learning_rate": 9.924861541608553e-05,
+      "loss": 1.1159,
       "step": 3720
     },
     {
-      "epoch": 6.62,
-      "learning_rate": 0.00021544207999178917,
-      "loss": 0.7823,
+      "epoch": 0.42,
+      "learning_rate": 9.92359930659387e-05,
+      "loss": 1.0964,
       "step": 3740
     },
     {
-      "epoch": 6.66,
-      "learning_rate": 0.00021137799392471814,
-      "loss": 0.7779,
+      "epoch": 0.42,
+      "learning_rate": 9.922326639307917e-05,
+      "loss": 1.1329,
       "step": 3760
     },
     {
-      "epoch": 6.69,
-      "learning_rate": 0.00020733879486455433,
-      "loss": 0.7867,
+      "epoch": 0.42,
+      "learning_rate": 9.921043542447264e-05,
+      "loss": 1.1393,
       "step": 3780
     },
     {
-      "epoch": 6.73,
-      "learning_rate": 0.0002033250157479206,
-      "loss": 0.7738,
+      "epoch": 0.43,
+      "learning_rate": 9.919750018730571e-05,
+      "loss": 1.1251,
       "step": 3800
     },
     {
-      "epoch": 6.76,
-      "learning_rate": 0.0001993371861575028,
-      "loss": 0.7814,
+      "epoch": 0.43,
+      "learning_rate": 9.918446070898601e-05,
+      "loss": 1.1018,
       "step": 3820
     },
     {
-      "epoch": 6.8,
-      "learning_rate": 0.00019537583225217605,
-      "loss": 0.7695,
+      "epoch": 0.43,
+      "learning_rate": 9.917131701714192e-05,
+      "loss": 1.1376,
       "step": 3840
     },
     {
-      "epoch": 6.83,
-      "learning_rate": 0.00019144147669758322,
-      "loss": 0.7846,
+      "epoch": 0.43,
+      "learning_rate": 9.915806913962274e-05,
+      "loss": 1.1901,
       "step": 3860
     },
     {
-      "epoch": 6.87,
-      "learning_rate": 0.00018753463859717283,
-      "loss": 0.7779,
+      "epoch": 0.44,
+      "learning_rate": 9.914471710449845e-05,
+      "loss": 1.1236,
       "step": 3880
     },
     {
-      "epoch": 6.91,
-      "learning_rate": 0.0001836558334237088,
-      "loss": 0.7769,
+      "epoch": 0.44,
+      "learning_rate": 9.913126094005976e-05,
+      "loss": 1.1188,
       "step": 3900
     },
     {
-      "epoch": 6.94,
-      "learning_rate": 0.0001798055729512579,
-      "loss": 0.782,
+      "epoch": 0.44,
+      "learning_rate": 9.911770067481798e-05,
+      "loss": 1.1358,
       "step": 3920
     },
     {
-      "epoch": 6.98,
-      "learning_rate": 0.00017598436518766596,
-      "loss": 0.7683,
+      "epoch": 0.44,
+      "learning_rate": 9.910403633750502e-05,
+      "loss": 1.14,
       "step": 3940
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 0.0001721927143075305,
-      "loss": 0.7674,
+      "epoch": 0.44,
+      "learning_rate": 9.909026795707331e-05,
+      "loss": 1.1145,
       "step": 3960
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 0.00016843112058567935,
-      "loss": 0.7427,
+      "epoch": 0.45,
+      "learning_rate": 9.907639556269566e-05,
+      "loss": 1.1162,
       "step": 3980
     },
     {
-      "epoch": 7.08,
-      "learning_rate": 0.00016470008033116443,
-      "loss": 0.7627,
+      "epoch": 0.45,
+      "learning_rate": 9.906241918376537e-05,
+      "loss": 1.1131,
       "step": 4000
     },
     {
-      "epoch": 7.12,
-      "learning_rate": 0.00016100008582177705,
-      "loss": 0.7541,
+      "epoch": 0.45,
+      "learning_rate": 9.904833884989602e-05,
+      "loss": 1.1174,
       "step": 4020
     },
     {
-      "epoch": 7.15,
-      "learning_rate": 0.00015733162523909707,
-      "loss": 0.7654,
+      "epoch": 0.45,
+      "learning_rate": 9.90341545909214e-05,
+      "loss": 1.1196,
       "step": 4040
     },
     {
-      "epoch": 7.19,
-      "learning_rate": 0.0001536951826040813,
-      "loss": 0.7382,
+      "epoch": 0.46,
+      "learning_rate": 9.901986643689559e-05,
+      "loss": 1.1053,
       "step": 4060
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 0.0001500912377132013,
-      "loss": 0.7555,
+      "epoch": 0.46,
+      "learning_rate": 9.900547441809272e-05,
+      "loss": 1.1168,
       "step": 4080
     },
     {
-      "epoch": 7.26,
-      "learning_rate": 0.00014652026607513848,
-      "loss": 0.748,
+      "epoch": 0.46,
+      "learning_rate": 9.899097856500707e-05,
+      "loss": 1.101,
       "step": 4100
     },
     {
-      "epoch": 7.3,
-      "learning_rate": 0.00014298273884804478,
-      "loss": 0.753,
+      "epoch": 0.46,
+      "learning_rate": 9.897637890835289e-05,
+      "loss": 1.1039,
       "step": 4120
     },
     {
-      "epoch": 7.33,
-      "learning_rate": 0.00013947912277737808,
-      "loss": 0.7594,
+      "epoch": 0.47,
+      "learning_rate": 9.896167547906437e-05,
+      "loss": 1.1027,
       "step": 4140
     },
     {
-      "epoch": 7.37,
-      "learning_rate": 0.00013600988013431832,
-      "loss": 0.7387,
+      "epoch": 0.47,
+      "learning_rate": 9.894686830829558e-05,
+      "loss": 1.1028,
       "step": 4160
     },
     {
-      "epoch": 7.4,
-      "learning_rate": 0.00013257546865477572,
-      "loss": 0.7551,
+      "epoch": 0.47,
+      "learning_rate": 9.89319574274204e-05,
+      "loss": 1.0957,
       "step": 4180
     },
     {
-      "epoch": 7.44,
-      "learning_rate": 0.00012917634147899607,
-      "loss": 0.7609,
+      "epoch": 0.47,
+      "learning_rate": 9.891694286803246e-05,
+      "loss": 1.1397,
       "step": 4200
     },
     {
-      "epoch": 7.47,
-      "learning_rate": 0.00012581294709177327,
-      "loss": 0.762,
+      "epoch": 0.47,
+      "learning_rate": 9.890182466194505e-05,
+      "loss": 1.109,
       "step": 4220
     },
     {
-      "epoch": 7.51,
-      "learning_rate": 0.00012248572926327537,
-      "loss": 0.755,
+      "epoch": 0.48,
+      "learning_rate": 9.88866028411911e-05,
+      "loss": 1.1414,
       "step": 4240
     },
     {
-      "epoch": 7.54,
-      "learning_rate": 0.00011919512699049314,
-      "loss": 0.753,
+      "epoch": 0.48,
+      "learning_rate": 9.887127743802304e-05,
+      "loss": 1.136,
       "step": 4260
     },
     {
-      "epoch": 7.58,
-      "learning_rate": 0.00011594157443931872,
-      "loss": 0.7603,
+      "epoch": 0.48,
+      "learning_rate": 9.885584848491285e-05,
+      "loss": 1.1458,
       "step": 4280
     },
     {
-      "epoch": 7.61,
-      "learning_rate": 0.0001127255008872604,
-      "loss": 0.7565,
+      "epoch": 0.48,
+      "learning_rate": 9.884031601455179e-05,
+      "loss": 1.1379,
       "step": 4300
     },
     {
-      "epoch": 7.65,
-      "learning_rate": 0.00010954733066680401,
-      "loss": 0.7542,
+      "epoch": 0.49,
+      "learning_rate": 9.88246800598506e-05,
+      "loss": 1.1083,
       "step": 4320
     },
     {
-      "epoch": 7.68,
-      "learning_rate": 0.00010640748310942559,
-      "loss": 0.7674,
+      "epoch": 0.49,
+      "learning_rate": 9.880894065393915e-05,
+      "loss": 1.1063,
       "step": 4340
     },
     {
-      "epoch": 7.72,
-      "learning_rate": 0.00010330637249026445,
-      "loss": 0.7447,
+      "epoch": 0.49,
+      "learning_rate": 9.879309783016663e-05,
+      "loss": 1.101,
       "step": 4360
     },
     {
-      "epoch": 7.76,
-      "learning_rate": 0.00010024440797346324,
-      "loss": 0.7582,
+      "epoch": 0.49,
+      "learning_rate": 9.877715162210123e-05,
+      "loss": 1.1023,
       "step": 4380
     },
     {
-      "epoch": 7.79,
-      "learning_rate": 9.722199355818227e-05,
-      "loss": 0.7598,
+      "epoch": 0.49,
+      "learning_rate": 9.876110206353033e-05,
+      "loss": 1.1223,
       "step": 4400
     },
     {
-      "epoch": 7.83,
-      "learning_rate": 9.423952802529564e-05,
-      "loss": 0.7389,
+      "epoch": 0.5,
+      "learning_rate": 9.874494918846017e-05,
+      "loss": 1.1348,
       "step": 4420
     },
     {
-      "epoch": 7.86,
-      "learning_rate": 9.129740488477518e-05,
-      "loss": 0.759,
+      "epoch": 0.5,
+      "learning_rate": 9.872869303111595e-05,
+      "loss": 1.0998,
       "step": 4440
     },
     {
-      "epoch": 7.9,
-      "learning_rate": 8.83960123237706e-05,
-      "loss": 0.7442,
+      "epoch": 0.5,
+      "learning_rate": 9.871233362594175e-05,
+      "loss": 1.0983,
       "step": 4460
     },
     {
-      "epoch": 7.93,
-      "learning_rate": 8.553573315539188e-05,
-      "loss": 0.7627,
+      "epoch": 0.5,
+      "learning_rate": 9.869587100760034e-05,
+      "loss": 1.1114,
       "step": 4480
     },
     {
-      "epoch": 7.97,
-      "learning_rate": 8.271694476819956e-05,
-      "loss": 0.7531,
+      "epoch": 0.51,
+      "learning_rate": 9.86793052109732e-05,
+      "loss": 1.0765,
       "step": 4500
     },
     {
-      "epoch": 8.0,
-      "learning_rate": 7.994001907641262e-05,
-      "loss": 0.7567,
+      "epoch": 0.51,
+      "learning_rate": 9.866263627116049e-05,
+      "loss": 1.089,
       "step": 4520
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 7.720532247083743e-05,
-      "loss": 0.7274,
+      "epoch": 0.51,
+      "learning_rate": 9.864586422348081e-05,
+      "loss": 1.101,
       "step": 4540
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 7.451321577052533e-05,
-      "loss": 0.7347,
+      "epoch": 0.51,
+      "learning_rate": 9.862898910347132e-05,
+      "loss": 1.1247,
       "step": 4560
     },
     {
-      "epoch": 8.11,
-      "learning_rate": 7.18640541751661e-05,
-      "loss": 0.752,
+      "epoch": 0.51,
+      "learning_rate": 9.861201094688752e-05,
+      "loss": 1.113,
       "step": 4580
     },
     {
-      "epoch": 8.15,
-      "learning_rate": 6.925818721822239e-05,
-      "loss": 0.7293,
+      "epoch": 0.52,
+      "learning_rate": 9.859492978970325e-05,
+      "loss": 1.1354,
       "step": 4600
     },
     {
-      "epoch": 8.18,
-      "learning_rate": 6.669595872081211e-05,
-      "loss": 0.7213,
+      "epoch": 0.52,
+      "learning_rate": 9.857774566811058e-05,
+      "loss": 1.095,
       "step": 4620
     },
     {
-      "epoch": 8.22,
-      "learning_rate": 6.417770674634365e-05,
-      "loss": 0.7246,
+      "epoch": 0.52,
+      "learning_rate": 9.856045861851975e-05,
+      "loss": 1.1188,
       "step": 4640
     },
     {
-      "epoch": 8.25,
-      "learning_rate": 6.170376355591204e-05,
-      "loss": 0.7353,
+      "epoch": 0.52,
+      "learning_rate": 9.854306867755906e-05,
+      "loss": 1.0938,
       "step": 4660
     },
     {
-      "epoch": 8.29,
-      "learning_rate": 5.9274455564459896e-05,
-      "loss": 0.7396,
+      "epoch": 0.53,
+      "learning_rate": 9.852557588207487e-05,
+      "loss": 1.1113,
       "step": 4680
     },
     {
-      "epoch": 8.32,
-      "learning_rate": 5.689010329770965e-05,
-      "loss": 0.7401,
+      "epoch": 0.53,
+      "learning_rate": 9.850798026913145e-05,
+      "loss": 1.088,
       "step": 4700
     },
     {
-      "epoch": 8.36,
-      "learning_rate": 5.455102134987304e-05,
-      "loss": 0.7399,
+      "epoch": 0.53,
+      "learning_rate": 9.849028187601091e-05,
+      "loss": 1.1256,
       "step": 4720
     },
     {
-      "epoch": 8.39,
-      "learning_rate": 5.225751834214339e-05,
-      "loss": 0.7427,
+      "epoch": 0.53,
+      "learning_rate": 9.847248074021312e-05,
+      "loss": 1.0979,
       "step": 4740
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 5.000989688197555e-05,
-      "loss": 0.7289,
+      "epoch": 0.53,
+      "learning_rate": 9.845457689945567e-05,
+      "loss": 1.1227,
       "step": 4760
     },
     {
-      "epoch": 8.46,
-      "learning_rate": 4.780845352315968e-05,
-      "loss": 0.7398,
+      "epoch": 0.54,
+      "learning_rate": 9.84365703916738e-05,
+      "loss": 1.1265,
       "step": 4780
     },
     {
-      "epoch": 8.5,
-      "learning_rate": 4.565347872669339e-05,
-      "loss": 0.7332,
+      "epoch": 0.54,
+      "learning_rate": 9.841846125502021e-05,
+      "loss": 1.0997,
       "step": 4800
     },
     {
-      "epoch": 8.53,
-      "learning_rate": 4.3545256822458445e-05,
-      "loss": 0.7365,
+      "epoch": 0.54,
+      "learning_rate": 9.840024952786508e-05,
+      "loss": 1.0951,
       "step": 4820
     },
     {
-      "epoch": 8.57,
-      "learning_rate": 4.148406597170529e-05,
-      "loss": 0.7347,
+      "epoch": 0.54,
+      "learning_rate": 9.838193524879599e-05,
+      "loss": 1.1091,
       "step": 4840
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 3.947017813035254e-05,
-      "loss": 0.7515,
+      "epoch": 0.55,
+      "learning_rate": 9.836351845661777e-05,
+      "loss": 1.1148,
       "step": 4860
     },
     {
-      "epoch": 8.64,
-      "learning_rate": 3.7503859013104806e-05,
-      "loss": 0.7397,
+      "epoch": 0.55,
+      "learning_rate": 9.834499919035249e-05,
+      "loss": 1.1211,
       "step": 4880
     },
     {
-      "epoch": 8.68,
-      "learning_rate": 3.5585368058393834e-05,
-      "loss": 0.7597,
+      "epoch": 0.55,
+      "learning_rate": 9.832637748923934e-05,
+      "loss": 1.1033,
       "step": 4900
     },
     {
-      "epoch": 8.71,
-      "learning_rate": 3.3714958394147975e-05,
-      "loss": 0.7391,
+      "epoch": 0.55,
+      "learning_rate": 9.830765339273454e-05,
+      "loss": 1.0925,
       "step": 4920
     },
     {
-      "epoch": 8.75,
-      "learning_rate": 3.1892876804394144e-05,
-      "loss": 0.7359,
+      "epoch": 0.56,
+      "learning_rate": 9.828882694051124e-05,
+      "loss": 1.1061,
       "step": 4940
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 3.0119363696697078e-05,
-      "loss": 0.7412,
+      "epoch": 0.56,
+      "learning_rate": 9.826989817245953e-05,
+      "loss": 1.1083,
       "step": 4960
     },
     {
-      "epoch": 8.82,
-      "learning_rate": 2.839465307043927e-05,
-      "loss": 0.7485,
+      "epoch": 0.56,
+      "learning_rate": 9.825086712868625e-05,
+      "loss": 1.0864,
       "step": 4980
     },
     {
-      "epoch": 8.85,
-      "learning_rate": 2.6718972485947037e-05,
-      "loss": 0.7534,
+      "epoch": 0.56,
+      "learning_rate": 9.823173384951496e-05,
+      "loss": 1.1202,
       "step": 5000
     },
     {
-      "epoch": 8.89,
-      "learning_rate": 2.5092543034466264e-05,
-      "loss": 0.7405,
+      "epoch": 0.56,
+      "learning_rate": 9.821249837548582e-05,
+      "loss": 1.101,
       "step": 5020
     },
     {
-      "epoch": 8.92,
-      "learning_rate": 2.3515579308990597e-05,
-      "loss": 0.7415,
+      "epoch": 0.57,
+      "learning_rate": 9.819316074735554e-05,
+      "loss": 1.1191,
       "step": 5040
     },
     {
-      "epoch": 8.96,
-      "learning_rate": 2.1988289375948524e-05,
-      "loss": 0.7309,
+      "epoch": 0.57,
+      "learning_rate": 9.817372100609726e-05,
+      "loss": 1.0535,
       "step": 5060
     },
     {
-      "epoch": 9.0,
-      "learning_rate": 2.0510874747750575e-05,
-      "loss": 0.7418,
+      "epoch": 0.57,
+      "learning_rate": 9.81541791929005e-05,
+      "loss": 1.1082,
       "step": 5080
     },
     {
-      "epoch": 9.03,
-      "learning_rate": 1.9083530356201407e-05,
-      "loss": 0.7469,
+      "epoch": 0.57,
+      "learning_rate": 9.813453534917105e-05,
+      "loss": 1.1366,
       "step": 5100
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 1.7706444526780585e-05,
-      "loss": 0.7385,
+      "epoch": 0.58,
+      "learning_rate": 9.811478951653088e-05,
+      "loss": 1.072,
       "step": 5120
     },
     {
-      "epoch": 9.1,
-      "learning_rate": 1.637979895379429e-05,
-      "loss": 0.7333,
+      "epoch": 0.58,
+      "learning_rate": 9.809494173681804e-05,
+      "loss": 1.0954,
       "step": 5140
     },
     {
-      "epoch": 9.14,
-      "learning_rate": 1.5103768676402885e-05,
-      "loss": 0.7314,
+      "epoch": 0.58,
+      "learning_rate": 9.807499205208663e-05,
+      "loss": 1.1014,
       "step": 5160
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 1.38785220555254e-05,
-      "loss": 0.74,
+      "epoch": 0.58,
+      "learning_rate": 9.805494050460666e-05,
+      "loss": 1.0966,
       "step": 5180
     },
     {
-      "epoch": 9.21,
-      "learning_rate": 1.270422075162645e-05,
-      "loss": 0.7481,
+      "epoch": 0.58,
+      "learning_rate": 9.803478713686391e-05,
+      "loss": 1.1035,
       "step": 5200
     },
     {
-      "epoch": 9.24,
-      "learning_rate": 1.1581019703386143e-05,
-      "loss": 0.7222,
+      "epoch": 0.59,
+      "learning_rate": 9.801453199155996e-05,
+      "loss": 1.1061,
       "step": 5220
     },
     {
-      "epoch": 9.28,
-      "learning_rate": 1.0509067107257365e-05,
-      "loss": 0.7259,
+      "epoch": 0.59,
+      "learning_rate": 9.799417511161206e-05,
+      "loss": 1.0943,
       "step": 5240
     },
     {
-      "epoch": 9.31,
-      "learning_rate": 9.488504397912712e-06,
-      "loss": 0.7309,
+      "epoch": 0.59,
+      "learning_rate": 9.797371654015296e-05,
+      "loss": 1.0943,
       "step": 5260
     },
     {
-      "epoch": 9.35,
-      "learning_rate": 8.51946622958324e-06,
-      "loss": 0.7212,
+      "epoch": 0.59,
+      "learning_rate": 9.795315632053088e-05,
+      "loss": 1.0939,
       "step": 5280
     },
     {
-      "epoch": 9.38,
-      "learning_rate": 7.602080458292227e-06,
-      "loss": 0.7254,
+      "epoch": 0.6,
+      "learning_rate": 9.793249449630946e-05,
+      "loss": 1.083,
       "step": 5300
     },
     {
-      "epoch": 9.42,
-      "learning_rate": 6.7364681249854735e-06,
-      "loss": 0.728,
+      "epoch": 0.6,
+      "learning_rate": 9.791173111126759e-05,
+      "loss": 1.0958,
       "step": 5320
     },
     {
-      "epoch": 9.46,
-      "learning_rate": 5.922743439561229e-06,
-      "loss": 0.7341,
+      "epoch": 0.6,
+      "learning_rate": 9.789086620939936e-05,
+      "loss": 1.0671,
       "step": 5340
     },
     {
-      "epoch": 9.49,
-      "learning_rate": 5.161013765801137e-06,
-      "loss": 0.721,
+      "epoch": 0.6,
+      "learning_rate": 9.786989983491397e-05,
+      "loss": 1.0766,
       "step": 5360
     },
     {
-      "epoch": 9.53,
-      "learning_rate": 4.451379607204453e-06,
-      "loss": 0.725,
+      "epoch": 0.6,
+      "learning_rate": 9.784883203223558e-05,
+      "loss": 1.1059,
       "step": 5380
     },
     {
-      "epoch": 9.56,
-      "learning_rate": 3.7939345937275884e-06,
-      "loss": 0.7399,
+      "epoch": 0.61,
+      "learning_rate": 9.782766284600332e-05,
+      "loss": 1.1136,
       "step": 5400
     },
     {
-      "epoch": 9.6,
-      "learning_rate": 3.1887654694303883e-06,
-      "loss": 0.7301,
+      "epoch": 0.61,
+      "learning_rate": 9.780639232107108e-05,
+      "loss": 1.1049,
       "step": 5420
     },
     {
-      "epoch": 9.63,
-      "learning_rate": 2.635952081031201e-06,
-      "loss": 0.7222,
+      "epoch": 0.61,
+      "learning_rate": 9.778502050250749e-05,
+      "loss": 1.0934,
       "step": 5440
     },
     {
-      "epoch": 9.67,
-      "learning_rate": 2.1355673673715716e-06,
-      "loss": 0.7308,
+      "epoch": 0.61,
+      "learning_rate": 9.776354743559583e-05,
+      "loss": 1.0905,
       "step": 5460
     },
     {
-      "epoch": 9.7,
-      "learning_rate": 1.6876773497926046e-06,
-      "loss": 0.7417,
+      "epoch": 0.62,
+      "learning_rate": 9.774197316583387e-05,
+      "loss": 1.0722,
       "step": 5480
     },
     {
-      "epoch": 9.74,
-      "learning_rate": 1.292341123424423e-06,
-      "loss": 0.7348,
+      "epoch": 0.62,
+      "learning_rate": 9.77202977389338e-05,
+      "loss": 1.0761,
       "step": 5500
     },
     {
-      "epoch": 9.77,
-      "learning_rate": 9.496108493884936e-07,
-      "loss": 0.7278,
+      "epoch": 0.62,
+      "learning_rate": 9.769852120082222e-05,
+      "loss": 1.0705,
       "step": 5520
     },
     {
-      "epoch": 9.81,
-      "learning_rate": 6.595317479159313e-07,
-      "loss": 0.7257,
+      "epoch": 0.62,
+      "learning_rate": 9.767664359763991e-05,
+      "loss": 1.0709,
       "step": 5540
     },
     {
-      "epoch": 9.85,
-      "learning_rate": 4.2214209238085054e-07,
-      "loss": 0.7296,
+      "epoch": 0.62,
+      "learning_rate": 9.765466497574175e-05,
+      "loss": 1.0927,
       "step": 5560
     },
     {
-      "epoch": 9.88,
-      "learning_rate": 2.3747320425053786e-07,
-      "loss": 0.7301,
+      "epoch": 0.63,
+      "learning_rate": 9.763258538169675e-05,
+      "loss": 1.0942,
       "step": 5580
     },
     {
-      "epoch": 9.92,
-      "learning_rate": 1.0554944895293517e-07,
-      "loss": 0.739,
+      "epoch": 0.63,
+      "learning_rate": 9.761040486228783e-05,
+      "loss": 1.1113,
       "step": 5600
     },
     {
-      "epoch": 9.95,
-      "learning_rate": 2.638823266174484e-08,
-      "loss": 0.7284,
+      "epoch": 0.63,
+      "learning_rate": 9.758812346451171e-05,
+      "loss": 1.0902,
       "step": 5620
     },
     {
-      "epoch": 9.99,
-      "learning_rate": 0.0,
-      "loss": 0.7294,
+      "epoch": 0.63,
+      "learning_rate": 9.756574123557893e-05,
+      "loss": 1.1216,
       "step": 5640
     },
     {
-      "epoch": 9.99,
-      "step": 5640,
-      "total_flos": 1.4686171215861645e+19,
-      "train_loss": 0.8784972819876163,
-      "train_runtime": 13393.5014,
-      "train_samples_per_second": 53.949,
-      "train_steps_per_second": 0.421
+      "epoch": 0.64,
+      "learning_rate": 9.754325822291362e-05,
+      "loss": 1.1029,
+      "step": 5660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.752067447415342e-05,
+      "loss": 1.1237,
+      "step": 5680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.749799003714954e-05,
+      "loss": 1.0988,
+      "step": 5700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.747520495996641e-05,
+      "loss": 1.087,
+      "step": 5720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.745231929088174e-05,
+      "loss": 1.0668,
+      "step": 5740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.74293330783864e-05,
+      "loss": 1.0756,
+      "step": 5760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.740624637118425e-05,
+      "loss": 1.1091,
+      "step": 5780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.73830592181921e-05,
+      "loss": 1.0985,
+      "step": 5800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.735977166853962e-05,
+      "loss": 1.0667,
+      "step": 5820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.733638377156915e-05,
+      "loss": 1.0753,
+      "step": 5840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.731289557683567e-05,
+      "loss": 1.0869,
+      "step": 5860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.72893071341067e-05,
+      "loss": 1.0944,
+      "step": 5880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.726561849336216e-05,
+      "loss": 1.0817,
+      "step": 5900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.724182970479422e-05,
+      "loss": 1.1044,
+      "step": 5920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.72179408188073e-05,
+      "loss": 1.1296,
+      "step": 5940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.71939518860179e-05,
+      "loss": 1.0678,
+      "step": 5960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.71698629572545e-05,
+      "loss": 1.0837,
+      "step": 5980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.714567408355744e-05,
+      "loss": 1.0911,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.712138531617883e-05,
+      "loss": 1.0979,
+      "step": 6020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.709699670658248e-05,
+      "loss": 1.0742,
+      "step": 6040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.707250830644367e-05,
+      "loss": 1.0689,
+      "step": 6060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.704792016764922e-05,
+      "loss": 1.1154,
+      "step": 6080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.702323234229717e-05,
+      "loss": 1.089,
+      "step": 6100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.699844488269687e-05,
+      "loss": 1.0787,
+      "step": 6120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.69735578413687e-05,
+      "loss": 1.0688,
+      "step": 6140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.69485712710441e-05,
+      "loss": 1.0665,
+      "step": 6160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.692348522466537e-05,
+      "loss": 1.0686,
+      "step": 6180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.689829975538559e-05,
+      "loss": 1.0872,
+      "step": 6200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.687301491656849e-05,
+      "loss": 1.0818,
+      "step": 6220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.684763076178836e-05,
+      "loss": 1.0844,
+      "step": 6240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.682214734482989e-05,
+      "loss": 1.0977,
+      "step": 6260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.679656471968814e-05,
+      "loss": 1.072,
+      "step": 6280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.677088294056833e-05,
+      "loss": 1.0869,
+      "step": 6300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.674510206188584e-05,
+      "loss": 1.0839,
+      "step": 6320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.671922213826589e-05,
+      "loss": 1.077,
+      "step": 6340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.669324322454373e-05,
+      "loss": 1.0974,
+      "step": 6360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.666716537576422e-05,
+      "loss": 1.1057,
+      "step": 6380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.664098864718191e-05,
+      "loss": 1.0543,
+      "step": 6400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.661471309426085e-05,
+      "loss": 1.0699,
+      "step": 6420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.658833877267448e-05,
+      "loss": 1.091,
+      "step": 6440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.65618657383055e-05,
+      "loss": 1.0425,
+      "step": 6460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.653529404724578e-05,
+      "loss": 1.0519,
+      "step": 6480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.650862375579622e-05,
+      "loss": 1.1099,
+      "step": 6500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.648185492046663e-05,
+      "loss": 1.1131,
+      "step": 6520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.645498759797566e-05,
+      "loss": 1.1014,
+      "step": 6540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.642802184525058e-05,
+      "loss": 1.0645,
+      "step": 6560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.640095771942725e-05,
+      "loss": 1.0989,
+      "step": 6580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.637379527784997e-05,
+      "loss": 1.1347,
+      "step": 6600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.634653457807135e-05,
+      "loss": 1.1031,
+      "step": 6620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.631917567785213e-05,
+      "loss": 1.1107,
+      "step": 6640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.629171863516126e-05,
+      "loss": 1.0845,
+      "step": 6660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.626416350817549e-05,
+      "loss": 1.0804,
+      "step": 6680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.623651035527947e-05,
+      "loss": 1.081,
+      "step": 6700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.620875923506556e-05,
+      "loss": 1.1133,
+      "step": 6720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.618091020633365e-05,
+      "loss": 1.0605,
+      "step": 6740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.615296332809112e-05,
+      "loss": 1.034,
+      "step": 6760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.612491865955265e-05,
+      "loss": 1.0704,
+      "step": 6780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.609677626014015e-05,
+      "loss": 1.0791,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.606853618948256e-05,
+      "loss": 1.0931,
+      "step": 6820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.604019850741582e-05,
+      "loss": 1.0579,
+      "step": 6840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.601176327398267e-05,
+      "loss": 1.0252,
+      "step": 6860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.598323054943252e-05,
+      "loss": 1.0754,
+      "step": 6880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.59546003942214e-05,
+      "loss": 1.0665,
+      "step": 6900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.592587286901172e-05,
+      "loss": 1.1004,
+      "step": 6920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.589704803467225e-05,
+      "loss": 1.1064,
+      "step": 6940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.586812595227792e-05,
+      "loss": 1.0677,
+      "step": 6960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.583910668310971e-05,
+      "loss": 1.0727,
+      "step": 6980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.580999028865452e-05,
+      "loss": 1.0799,
+      "step": 7000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.578077683060507e-05,
+      "loss": 1.0665,
+      "step": 7020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.57514663708597e-05,
+      "loss": 1.0729,
+      "step": 7040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.572205897152229e-05,
+      "loss": 1.0472,
+      "step": 7060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.569255469490214e-05,
+      "loss": 1.0633,
+      "step": 7080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.566295360351383e-05,
+      "loss": 1.0578,
+      "step": 7100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.563325576007701e-05,
+      "loss": 1.0679,
+      "step": 7120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.560346122751638e-05,
+      "loss": 1.0878,
+      "step": 7140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.557357006896152e-05,
+      "loss": 1.0521,
+      "step": 7160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.554358234774669e-05,
+      "loss": 1.0543,
+      "step": 7180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.55134981274108e-05,
+      "loss": 1.0781,
+      "step": 7200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.548331747169719e-05,
+      "loss": 1.0668,
+      "step": 7220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.545304044455357e-05,
+      "loss": 1.0908,
+      "step": 7240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.542266711013182e-05,
+      "loss": 1.0535,
+      "step": 7260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.539219753278785e-05,
+      "loss": 1.0587,
+      "step": 7280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.536163177708155e-05,
+      "loss": 1.0406,
+      "step": 7300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.533096990777657e-05,
+      "loss": 1.0645,
+      "step": 7320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.530021198984019e-05,
+      "loss": 1.0425,
+      "step": 7340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.526935808844324e-05,
+      "loss": 1.0564,
+      "step": 7360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.523840826895988e-05,
+      "loss": 1.0811,
+      "step": 7380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.520736259696753e-05,
+      "loss": 1.0727,
+      "step": 7400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.51762211382467e-05,
+      "loss": 1.0687,
+      "step": 7420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.514498395878086e-05,
+      "loss": 1.0902,
+      "step": 7440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.51136511247563e-05,
+      "loss": 1.0613,
+      "step": 7460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.508222270256195e-05,
+      "loss": 1.0809,
+      "step": 7480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.505069875878934e-05,
+      "loss": 1.0682,
+      "step": 7500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.501907936023231e-05,
+      "loss": 1.07,
+      "step": 7520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.498736457388703e-05,
+      "loss": 1.0797,
+      "step": 7540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.495555446695175e-05,
+      "loss": 1.0468,
+      "step": 7560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.492364910682668e-05,
+      "loss": 1.0903,
+      "step": 7580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.489164856111387e-05,
+      "loss": 1.0929,
+      "step": 7600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.485955289761703e-05,
+      "loss": 1.0669,
+      "step": 7620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.482736218434143e-05,
+      "loss": 1.0938,
+      "step": 7640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.479507648949372e-05,
+      "loss": 1.0582,
+      "step": 7660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.476269588148186e-05,
+      "loss": 1.0793,
+      "step": 7680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.473022042891477e-05,
+      "loss": 1.0696,
+      "step": 7700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.469765020060251e-05,
+      "loss": 1.0599,
+      "step": 7720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.46649852655558e-05,
+      "loss": 1.0667,
+      "step": 7740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.463222569298615e-05,
+      "loss": 1.0126,
+      "step": 7760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.459937155230549e-05,
+      "loss": 1.0694,
+      "step": 7780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.456642291312618e-05,
+      "loss": 1.0566,
+      "step": 7800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.45333798452608e-05,
+      "loss": 1.0664,
+      "step": 7820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.450024241872201e-05,
+      "loss": 1.0408,
+      "step": 7840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.446701070372237e-05,
+      "loss": 1.0734,
+      "step": 7860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.443368477067427e-05,
+      "loss": 1.0604,
+      "step": 7880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.440026469018968e-05,
+      "loss": 1.0746,
+      "step": 7900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.43667505330801e-05,
+      "loss": 1.0358,
+      "step": 7920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.433314237035631e-05,
+      "loss": 1.0631,
+      "step": 7940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.429944027322834e-05,
+      "loss": 1.0296,
+      "step": 7960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.426564431310521e-05,
+      "loss": 1.0412,
+      "step": 7980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.42317545615948e-05,
+      "loss": 1.0554,
+      "step": 8000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.419777109050376e-05,
+      "loss": 1.0978,
+      "step": 8020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.416369397183728e-05,
+      "loss": 1.0215,
+      "step": 8040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.4129523277799e-05,
+      "loss": 1.0424,
+      "step": 8060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.409525908079082e-05,
+      "loss": 1.0453,
+      "step": 8080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.406090145341277e-05,
+      "loss": 1.0588,
+      "step": 8100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.402645046846281e-05,
+      "loss": 1.0439,
+      "step": 8120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.399190619893676e-05,
+      "loss": 1.0501,
+      "step": 8140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.395726871802804e-05,
+      "loss": 1.0565,
+      "step": 8160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.392253809912758e-05,
+      "loss": 1.0456,
+      "step": 8180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.388771441582369e-05,
+      "loss": 1.0547,
+      "step": 8200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.385279774190184e-05,
+      "loss": 1.0344,
+      "step": 8220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.381778815134455e-05,
+      "loss": 1.0286,
+      "step": 8240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.378268571833116e-05,
+      "loss": 1.1033,
+      "step": 8260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.374749051723781e-05,
+      "loss": 1.0667,
+      "step": 8280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.371220262263713e-05,
+      "loss": 1.0416,
+      "step": 8300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.36768221092982e-05,
+      "loss": 1.0674,
+      "step": 8320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.364134905218632e-05,
+      "loss": 1.0305,
+      "step": 8340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.360578352646285e-05,
+      "loss": 1.0604,
+      "step": 8360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.357012560748513e-05,
+      "loss": 1.061,
+      "step": 8380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.353437537080625e-05,
+      "loss": 1.0678,
+      "step": 8400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.349853289217485e-05,
+      "loss": 1.0767,
+      "step": 8420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.34625982475351e-05,
+      "loss": 1.0562,
+      "step": 8440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.342657151302637e-05,
+      "loss": 1.0301,
+      "step": 8460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.339045276498325e-05,
+      "loss": 1.063,
+      "step": 8480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.33542420799352e-05,
+      "loss": 1.0157,
+      "step": 8500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.331793953460653e-05,
+      "loss": 1.0564,
+      "step": 8520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.328154520591614e-05,
+      "loss": 1.0817,
+      "step": 8540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.324505917097749e-05,
+      "loss": 1.0453,
+      "step": 8560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.320848150709826e-05,
+      "loss": 1.0442,
+      "step": 8580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.317181229178031e-05,
+      "loss": 1.0379,
+      "step": 8600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.313505160271952e-05,
+      "loss": 1.0686,
+      "step": 8620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.30981995178055e-05,
+      "loss": 1.0417,
+      "step": 8640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.306125611512159e-05,
+      "loss": 1.0569,
+      "step": 8660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.302422147294458e-05,
+      "loss": 1.0646,
+      "step": 8680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.298709566974462e-05,
+      "loss": 1.0439,
+      "step": 8700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.294987878418495e-05,
+      "loss": 1.0529,
+      "step": 8720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.291257089512185e-05,
+      "loss": 1.0369,
+      "step": 8740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.287517208160439e-05,
+      "loss": 1.0509,
+      "step": 8760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.283768242287433e-05,
+      "loss": 1.0825,
+      "step": 8780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.280010199836588e-05,
+      "loss": 1.0583,
+      "step": 8800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.276243088770559e-05,
+      "loss": 1.0528,
+      "step": 8820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.272466917071216e-05,
+      "loss": 1.0307,
+      "step": 8840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.268681692739623e-05,
+      "loss": 1.0538,
+      "step": 8860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.264887423796029e-05,
+      "loss": 1.0459,
+      "step": 8880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.261084118279847e-05,
+      "loss": 1.052,
+      "step": 8900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.257271784249635e-05,
+      "loss": 0.9985,
+      "step": 8920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.253450429783081e-05,
+      "loss": 1.0312,
+      "step": 8940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.249620062976988e-05,
+      "loss": 1.0433,
+      "step": 8960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.245780691947252e-05,
+      "loss": 1.0061,
+      "step": 8980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.24193232482885e-05,
+      "loss": 1.0237,
+      "step": 9000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.238074969775818e-05,
+      "loss": 1.0311,
+      "step": 9020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.234208634961236e-05,
+      "loss": 1.0467,
+      "step": 9040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.230333328577212e-05,
+      "loss": 1.0217,
+      "step": 9060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.226449058834863e-05,
+      "loss": 1.024,
+      "step": 9080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.222555833964296e-05,
+      "loss": 1.0373,
+      "step": 9100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.218653662214593e-05,
+      "loss": 1.0248,
+      "step": 9120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.214742551853798e-05,
+      "loss": 1.0597,
+      "step": 9140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.210822511168884e-05,
+      "loss": 1.0138,
+      "step": 9160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.206893548465758e-05,
+      "loss": 1.0406,
+      "step": 9180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.20295567206922e-05,
+      "loss": 1.0399,
+      "step": 9200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.199008890322963e-05,
+      "loss": 1.0282,
+      "step": 9220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.19505321158955e-05,
+      "loss": 1.0156,
+      "step": 9240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.191088644250389e-05,
+      "loss": 1.0146,
+      "step": 9260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.187115196705731e-05,
+      "loss": 0.9898,
+      "step": 9280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.183132877374631e-05,
+      "loss": 1.0027,
+      "step": 9300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.17914169469495e-05,
+      "loss": 1.0047,
+      "step": 9320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.17514165712333e-05,
+      "loss": 1.0628,
+      "step": 9340
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.171132773135165e-05,
+      "loss": 1.041,
+      "step": 9360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.167115051224606e-05,
+      "loss": 1.0387,
+      "step": 9380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.16308849990452e-05,
+      "loss": 1.0521,
+      "step": 9400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.159053127706487e-05,
+      "loss": 0.9995,
+      "step": 9420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.155008943180776e-05,
+      "loss": 1.0068,
+      "step": 9440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.150955954896327e-05,
+      "loss": 1.0396,
+      "step": 9460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.146894171440735e-05,
+      "loss": 0.9964,
+      "step": 9480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.14282360142023e-05,
+      "loss": 0.9995,
+      "step": 9500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.138744253459658e-05,
+      "loss": 1.0396,
+      "step": 9520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.134656136202466e-05,
+      "loss": 1.0167,
+      "step": 9540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.130559258310679e-05,
+      "loss": 1.0319,
+      "step": 9560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.126453628464888e-05,
+      "loss": 1.0222,
+      "step": 9580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.122339255364224e-05,
+      "loss": 0.9881,
+      "step": 9600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.118216147726347e-05,
+      "loss": 1.0193,
+      "step": 9620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.11408431428742e-05,
+      "loss": 1.0327,
+      "step": 9640
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.109943763802097e-05,
+      "loss": 1.0149,
+      "step": 9660
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.105794505043505e-05,
+      "loss": 0.9916,
+      "step": 9680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.101636546803218e-05,
+      "loss": 1.0443,
+      "step": 9700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.09746989789124e-05,
+      "loss": 1.0485,
+      "step": 9720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.093294567135998e-05,
+      "loss": 1.0536,
+      "step": 9740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.089110563384304e-05,
+      "loss": 1.0167,
+      "step": 9760
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.084917895501357e-05,
+      "loss": 1.0192,
+      "step": 9780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.080716572370704e-05,
+      "loss": 1.0163,
+      "step": 9800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.07650660289424e-05,
+      "loss": 1.0041,
+      "step": 9820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.072287995992172e-05,
+      "loss": 1.0364,
+      "step": 9840
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.068060760603014e-05,
+      "loss": 1.0567,
+      "step": 9860
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.063824905683562e-05,
+      "loss": 1.0234,
+      "step": 9880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.059580440208869e-05,
+      "loss": 1.0011,
+      "step": 9900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.05532737317224e-05,
+      "loss": 1.0375,
+      "step": 9920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.051065713585203e-05,
+      "loss": 1.0142,
+      "step": 9940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.04679547047749e-05,
+      "loss": 1.0143,
+      "step": 9960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.04251665289702e-05,
+      "loss": 1.0232,
+      "step": 9980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.038229269909883e-05,
+      "loss": 1.0121,
+      "step": 10000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.033933330600316e-05,
+      "loss": 1.0262,
+      "step": 10020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.029628844070686e-05,
+      "loss": 1.0398,
+      "step": 10040
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.025315819441467e-05,
+      "loss": 1.0101,
+      "step": 10060
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.020994265851226e-05,
+      "loss": 1.0367,
+      "step": 10080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.016664192456604e-05,
+      "loss": 0.9912,
+      "step": 10100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.012325608432291e-05,
+      "loss": 1.0516,
+      "step": 10120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.00797852297101e-05,
+      "loss": 1.0297,
+      "step": 10140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.003622945283496e-05,
+      "loss": 1.0401,
+      "step": 10160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.99925888459848e-05,
+      "loss": 1.0365,
+      "step": 10180
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.994886350162666e-05,
+      "loss": 1.0227,
+      "step": 10200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.990505351240714e-05,
+      "loss": 1.0176,
+      "step": 10220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.986115897115213e-05,
+      "loss": 1.0419,
+      "step": 10240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.981717997086674e-05,
+      "loss": 1.0206,
+      "step": 10260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.977311660473499e-05,
+      "loss": 0.9932,
+      "step": 10280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.972896896611971e-05,
+      "loss": 1.0083,
+      "step": 10300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.968473714856222e-05,
+      "loss": 1.0006,
+      "step": 10320
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.964042124578224e-05,
+      "loss": 1.0136,
+      "step": 10340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.959602135167766e-05,
+      "loss": 1.0254,
+      "step": 10360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.955153756032428e-05,
+      "loss": 0.9969,
+      "step": 10380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.950696996597576e-05,
+      "loss": 0.9852,
+      "step": 10400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.946231866306325e-05,
+      "loss": 0.9993,
+      "step": 10420
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.941758374619525e-05,
+      "loss": 1.019,
+      "step": 10440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.93727653101575e-05,
+      "loss": 1.0028,
+      "step": 10460
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.932786344991264e-05,
+      "loss": 1.028,
+      "step": 10480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.928287826060009e-05,
+      "loss": 1.0152,
+      "step": 10500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.923780983753583e-05,
+      "loss": 1.0209,
+      "step": 10520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.919265827621218e-05,
+      "loss": 1.0409,
+      "step": 10540
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.914742367229768e-05,
+      "loss": 1.0155,
+      "step": 10560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.910210612163673e-05,
+      "loss": 0.994,
+      "step": 10580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.905670572024958e-05,
+      "loss": 1.0145,
+      "step": 10600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.901122256433195e-05,
+      "loss": 1.0211,
+      "step": 10620
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.89656567502549e-05,
+      "loss": 0.9977,
+      "step": 10640
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.89200083745647e-05,
+      "loss": 0.9889,
+      "step": 10660
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.887427753398248e-05,
+      "loss": 1.0104,
+      "step": 10680
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.882846432540413e-05,
+      "loss": 1.0535,
+      "step": 10700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.87825688459001e-05,
+      "loss": 1.0043,
+      "step": 10720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.873659119271507e-05,
+      "loss": 1.0144,
+      "step": 10740
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.869053146326793e-05,
+      "loss": 1.0118,
+      "step": 10760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.864438975515141e-05,
+      "loss": 1.0366,
+      "step": 10780
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.859816616613194e-05,
+      "loss": 1.0125,
+      "step": 10800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.855186079414949e-05,
+      "loss": 0.9938,
+      "step": 10820
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.850547373731727e-05,
+      "loss": 1.0046,
+      "step": 10840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.845900509392158e-05,
+      "loss": 1.0199,
+      "step": 10860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.841245496242157e-05,
+      "loss": 1.0019,
+      "step": 10880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.836582344144911e-05,
+      "loss": 0.9985,
+      "step": 10900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.831911062980845e-05,
+      "loss": 1.0138,
+      "step": 10920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.827231662647611e-05,
+      "loss": 1.0053,
+      "step": 10940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.822544153060064e-05,
+      "loss": 1.012,
+      "step": 10960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.817848544150243e-05,
+      "loss": 1.0178,
+      "step": 10980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.813144845867345e-05,
+      "loss": 1.0143,
+      "step": 11000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.808433068177708e-05,
+      "loss": 1.0087,
+      "step": 11020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.80371322106479e-05,
+      "loss": 1.0242,
+      "step": 11040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.798985314529146e-05,
+      "loss": 1.0364,
+      "step": 11060
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.794249358588407e-05,
+      "loss": 0.9967,
+      "step": 11080
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.789505363277259e-05,
+      "loss": 1.0144,
+      "step": 11100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.784753338647424e-05,
+      "loss": 1.016,
+      "step": 11120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.779993294767635e-05,
+      "loss": 1.0061,
+      "step": 11140
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.77522524172362e-05,
+      "loss": 1.0166,
+      "step": 11160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.770449189618069e-05,
+      "loss": 1.0228,
+      "step": 11180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.76566514857063e-05,
+      "loss": 1.0037,
+      "step": 11200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.76087312871787e-05,
+      "loss": 1.033,
+      "step": 11220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.75607314021327e-05,
+      "loss": 1.0347,
+      "step": 11240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.751265193227189e-05,
+      "loss": 0.9958,
+      "step": 11260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.746449297946853e-05,
+      "loss": 1.0154,
+      "step": 11280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.741625464576322e-05,
+      "loss": 1.0308,
+      "step": 11300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.736793703336482e-05,
+      "loss": 1.0114,
+      "step": 11320
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.731954024465017e-05,
+      "loss": 1.0231,
+      "step": 11340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.727106438216384e-05,
+      "loss": 1.01,
+      "step": 11360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.722250954861795e-05,
+      "loss": 0.983,
+      "step": 11380
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.717387584689195e-05,
+      "loss": 0.9938,
+      "step": 11400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.712516338003241e-05,
+      "loss": 0.9918,
+      "step": 11420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.707637225125276e-05,
+      "loss": 1.0061,
+      "step": 11440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.702750256393316e-05,
+      "loss": 1.0493,
+      "step": 11460
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.697855442162012e-05,
+      "loss": 0.9973,
+      "step": 11480
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.692952792802651e-05,
+      "loss": 1.0343,
+      "step": 11500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.688042318703111e-05,
+      "loss": 1.0333,
+      "step": 11520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.683124030267855e-05,
+      "loss": 1.0276,
+      "step": 11540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.678197937917901e-05,
+      "loss": 1.0016,
+      "step": 11560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.673264052090801e-05,
+      "loss": 0.9907,
+      "step": 11580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.668322383240626e-05,
+      "loss": 1.0153,
+      "step": 11600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.663372941837929e-05,
+      "loss": 1.0084,
+      "step": 11620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.658415738369737e-05,
+      "loss": 1.0255,
+      "step": 11640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.653450783339523e-05,
+      "loss": 0.9996,
+      "step": 11660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.648478087267187e-05,
+      "loss": 0.9922,
+      "step": 11680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.643497660689024e-05,
+      "loss": 1.021,
+      "step": 11700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.638509514157715e-05,
+      "loss": 1.0102,
+      "step": 11720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.633513658242295e-05,
+      "loss": 1.0281,
+      "step": 11740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.628510103528134e-05,
+      "loss": 1.0185,
+      "step": 11760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.623498860616918e-05,
+      "loss": 0.9624,
+      "step": 11780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.618479940126617e-05,
+      "loss": 1.0486,
+      "step": 11800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.613453352691473e-05,
+      "loss": 1.0069,
+      "step": 11820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.608419108961971e-05,
+      "loss": 1.0449,
+      "step": 11840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.603377219604823e-05,
+      "loss": 0.9983,
+      "step": 11860
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.59832769530293e-05,
+      "loss": 1.0164,
+      "step": 11880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.59327054675538e-05,
+      "loss": 1.0222,
+      "step": 11900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.588205784677415e-05,
+      "loss": 1.0372,
+      "step": 11920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.583133419800404e-05,
+      "loss": 1.0078,
+      "step": 11940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.578053462871827e-05,
+      "loss": 1.0216,
+      "step": 11960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.57296592465525e-05,
+      "loss": 1.0257,
+      "step": 11980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.567870815930305e-05,
+      "loss": 0.9813,
+      "step": 12000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.562768147492662e-05,
+      "loss": 0.9851,
+      "step": 12020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.557657930154007e-05,
+      "loss": 1.0091,
+      "step": 12040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.552540174742025e-05,
+      "loss": 0.9911,
+      "step": 12060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.547414892100373e-05,
+      "loss": 1.0127,
+      "step": 12080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.542282093088651e-05,
+      "loss": 0.9931,
+      "step": 12100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.537141788582393e-05,
+      "loss": 0.9987,
+      "step": 12120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.53199398947303e-05,
+      "loss": 1.0192,
+      "step": 12140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.526838706667873e-05,
+      "loss": 0.984,
+      "step": 12160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.521675951090094e-05,
+      "loss": 0.9841,
+      "step": 12180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.516505733678695e-05,
+      "loss": 1.0117,
+      "step": 12200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.511328065388488e-05,
+      "loss": 0.9758,
+      "step": 12220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.506142957190073e-05,
+      "loss": 1.0073,
+      "step": 12240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.500950420069817e-05,
+      "loss": 1.0178,
+      "step": 12260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.495750465029821e-05,
+      "loss": 1.0222,
+      "step": 12280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.490543103087912e-05,
+      "loss": 0.9812,
+      "step": 12300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.485328345277603e-05,
+      "loss": 1.0102,
+      "step": 12320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.48010620264808e-05,
+      "loss": 1.009,
+      "step": 12340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.47487668626418e-05,
+      "loss": 1.0279,
+      "step": 12360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.469639807206357e-05,
+      "loss": 1.019,
+      "step": 12380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.46439557657067e-05,
+      "loss": 1.0064,
+      "step": 12400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.459144005468756e-05,
+      "loss": 1.0037,
+      "step": 12420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.453885105027802e-05,
+      "loss": 0.9955,
+      "step": 12440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.448618886390522e-05,
+      "loss": 0.9949,
+      "step": 12460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.443345360715143e-05,
+      "loss": 0.9902,
+      "step": 12480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.43806453917537e-05,
+      "loss": 1.026,
+      "step": 12500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.432776432960366e-05,
+      "loss": 0.983,
+      "step": 12520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.427481053274734e-05,
+      "loss": 1.0039,
+      "step": 12540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.422178411338481e-05,
+      "loss": 0.99,
+      "step": 12560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.416868518387009e-05,
+      "loss": 1.0346,
+      "step": 12580
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.411551385671077e-05,
+      "loss": 1.0002,
+      "step": 12600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.406227024456788e-05,
+      "loss": 0.9916,
+      "step": 12620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.400895446025558e-05,
+      "loss": 1.0303,
+      "step": 12640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.3955566616741e-05,
+      "loss": 0.982,
+      "step": 12660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.39021068271439e-05,
+      "loss": 0.996,
+      "step": 12680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.38485752047365e-05,
+      "loss": 1.0142,
+      "step": 12700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.379497186294322e-05,
+      "loss": 1.032,
+      "step": 12720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.374129691534046e-05,
+      "loss": 1.0146,
+      "step": 12740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.36875504756563e-05,
+      "loss": 1.0217,
+      "step": 12760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.363373265777034e-05,
+      "loss": 0.9988,
+      "step": 12780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.357984357571337e-05,
+      "loss": 1.0119,
+      "step": 12800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.352588334366728e-05,
+      "loss": 0.9801,
+      "step": 12820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.347185207596457e-05,
+      "loss": 0.9835,
+      "step": 12840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.341774988708837e-05,
+      "loss": 0.9979,
+      "step": 12860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.336357689167203e-05,
+      "loss": 1.0108,
+      "step": 12880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.33093332044989e-05,
+      "loss": 0.9943,
+      "step": 12900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.325501894050218e-05,
+      "loss": 0.9962,
+      "step": 12920
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.320063421476454e-05,
+      "loss": 1.0326,
+      "step": 12940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.314617914251805e-05,
+      "loss": 0.9947,
+      "step": 12960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.30916538391437e-05,
+      "loss": 1.0253,
+      "step": 12980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.30370584201714e-05,
+      "loss": 0.9905,
+      "step": 13000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.298239300127954e-05,
+      "loss": 1.0028,
+      "step": 13020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.292765769829487e-05,
+      "loss": 1.0467,
+      "step": 13040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.287285262719224e-05,
+      "loss": 1.0042,
+      "step": 13060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.281797790409425e-05,
+      "loss": 1.0102,
+      "step": 13080
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.276303364527116e-05,
+      "loss": 1.0038,
+      "step": 13100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.270801996714051e-05,
+      "loss": 1.0028,
+      "step": 13120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.265293698626694e-05,
+      "loss": 0.9969,
+      "step": 13140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.259778481936197e-05,
+      "loss": 0.9955,
+      "step": 13160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.254256358328365e-05,
+      "loss": 1.0106,
+      "step": 13180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.248727339503641e-05,
+      "loss": 1.0129,
+      "step": 13200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.243191437177077e-05,
+      "loss": 0.981,
+      "step": 13220
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.237648663078314e-05,
+      "loss": 1.0168,
+      "step": 13240
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.232099028951548e-05,
+      "loss": 0.9942,
+      "step": 13260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.22654254655551e-05,
+      "loss": 1.0028,
+      "step": 13280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.22097922766344e-05,
+      "loss": 1.0011,
+      "step": 13300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.215409084063075e-05,
+      "loss": 1.0141,
+      "step": 13320
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.209832127556598e-05,
+      "loss": 0.9799,
+      "step": 13340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.204248369960634e-05,
+      "loss": 0.9972,
+      "step": 13360
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.198657823106219e-05,
+      "loss": 0.9737,
+      "step": 13380
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.193060498838774e-05,
+      "loss": 1.0,
+      "step": 13400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.187456409018074e-05,
+      "loss": 1.0246,
+      "step": 13420
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.18184556551824e-05,
+      "loss": 1.0043,
+      "step": 13440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.176227980227694e-05,
+      "loss": 0.9859,
+      "step": 13460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.170603665049146e-05,
+      "loss": 0.9851,
+      "step": 13480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.164972631899566e-05,
+      "loss": 1.0025,
+      "step": 13500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.159334892710156e-05,
+      "loss": 0.9988,
+      "step": 13520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.15369045942633e-05,
+      "loss": 1.0531,
+      "step": 13540
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.148039344007685e-05,
+      "loss": 0.983,
+      "step": 13560
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.142381558427974e-05,
+      "loss": 1.0149,
+      "step": 13580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.136717114675083e-05,
+      "loss": 1.0205,
+      "step": 13600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.131046024751009e-05,
+      "loss": 1.015,
+      "step": 13620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.12536830067183e-05,
+      "loss": 1.0131,
+      "step": 13640
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.119683954467677e-05,
+      "loss": 1.026,
+      "step": 13660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.113992998182715e-05,
+      "loss": 0.9792,
+      "step": 13680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.108295443875116e-05,
+      "loss": 1.0232,
+      "step": 13700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.102591303617031e-05,
+      "loss": 1.0185,
+      "step": 13720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.096880589494563e-05,
+      "loss": 0.9907,
+      "step": 13740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.091163313607749e-05,
+      "loss": 1.0145,
+      "step": 13760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.085439488070521e-05,
+      "loss": 0.9818,
+      "step": 13780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.079709125010699e-05,
+      "loss": 0.9885,
+      "step": 13800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.073972236569947e-05,
+      "loss": 1.0026,
+      "step": 13820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.06822883490376e-05,
+      "loss": 0.9577,
+      "step": 13840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.06247893218143e-05,
+      "loss": 1.0351,
+      "step": 13860
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.056722540586024e-05,
+      "loss": 0.9972,
+      "step": 13880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.050959672314359e-05,
+      "loss": 1.0099,
+      "step": 13900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.045190339576978e-05,
+      "loss": 0.9896,
+      "step": 13920
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.039414554598113e-05,
+      "loss": 1.0083,
+      "step": 13940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.033632329615676e-05,
+      "loss": 0.9907,
+      "step": 13960
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.027843676881218e-05,
+      "loss": 1.0125,
+      "step": 13980
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.022048608659913e-05,
+      "loss": 1.0153,
+      "step": 14000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.016247137230525e-05,
+      "loss": 0.9991,
+      "step": 14020
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.010439274885391e-05,
+      "loss": 0.9889,
+      "step": 14040
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.004625033930382e-05,
+      "loss": 1.0172,
+      "step": 14060
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.998804426684889e-05,
+      "loss": 0.9886,
+      "step": 14080
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.992977465481793e-05,
+      "loss": 1.0062,
+      "step": 14100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.987144162667431e-05,
+      "loss": 0.9952,
+      "step": 14120
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.981304530601586e-05,
+      "loss": 1.0364,
+      "step": 14140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.975458581657446e-05,
+      "loss": 1.021,
+      "step": 14160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.969606328221583e-05,
+      "loss": 1.0118,
+      "step": 14180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.96374778269393e-05,
+      "loss": 1.0187,
+      "step": 14200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.95788295748775e-05,
+      "loss": 1.0267,
+      "step": 14220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.952011865029614e-05,
+      "loss": 1.0121,
+      "step": 14240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.946134517759368e-05,
+      "loss": 0.971,
+      "step": 14260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.940250928130116e-05,
+      "loss": 1.0182,
+      "step": 14280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.934361108608183e-05,
+      "loss": 0.9876,
+      "step": 14300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.9284650716731e-05,
+      "loss": 0.988,
+      "step": 14320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.922562829817564e-05,
+      "loss": 1.0128,
+      "step": 14340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.916654395547427e-05,
+      "loss": 1.029,
+      "step": 14360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.91073978138166e-05,
+      "loss": 0.9795,
+      "step": 14380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.904818999852323e-05,
+      "loss": 0.9923,
+      "step": 14400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.898892063504548e-05,
+      "loss": 1.0132,
+      "step": 14420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.89295898489651e-05,
+      "loss": 0.999,
+      "step": 14440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.887019776599391e-05,
+      "loss": 0.9659,
+      "step": 14460
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.88107445119737e-05,
+      "loss": 0.9859,
+      "step": 14480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.875123021287579e-05,
+      "loss": 0.9863,
+      "step": 14500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.869165499480089e-05,
+      "loss": 1.0014,
+      "step": 14520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.863201898397878e-05,
+      "loss": 0.9914,
+      "step": 14540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.857232230676802e-05,
+      "loss": 0.9893,
+      "step": 14560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.851256508965577e-05,
+      "loss": 0.9932,
+      "step": 14580
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.845274745925744e-05,
+      "loss": 1.0187,
+      "step": 14600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.83928695423164e-05,
+      "loss": 1.0001,
+      "step": 14620
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.83329314657038e-05,
+      "loss": 0.991,
+      "step": 14640
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.827293335641825e-05,
+      "loss": 1.0057,
+      "step": 14660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.82128753415856e-05,
+      "loss": 0.9592,
+      "step": 14680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.815275754845854e-05,
+      "loss": 1.0121,
+      "step": 14700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.809258010441649e-05,
+      "loss": 0.9967,
+      "step": 14720
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.803234313696524e-05,
+      "loss": 0.9933,
+      "step": 14740
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.79720467737367e-05,
+      "loss": 0.9841,
+      "step": 14760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.791169114248864e-05,
+      "loss": 1.0147,
+      "step": 14780
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.785127637110438e-05,
+      "loss": 1.0291,
+      "step": 14800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.779080258759259e-05,
+      "loss": 0.9922,
+      "step": 14820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.773026992008692e-05,
+      "loss": 1.0002,
+      "step": 14840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.766967849684584e-05,
+      "loss": 1.0118,
+      "step": 14860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.760902844625228e-05,
+      "loss": 1.0237,
+      "step": 14880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.754831989681345e-05,
+      "loss": 1.0038,
+      "step": 14900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.74875529771604e-05,
+      "loss": 1.0016,
+      "step": 14920
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.742672781604794e-05,
+      "loss": 0.9977,
+      "step": 14940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.736584454235427e-05,
+      "loss": 0.9864,
+      "step": 14960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.730490328508072e-05,
+      "loss": 0.9771,
+      "step": 14980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.724390417335144e-05,
+      "loss": 0.983,
+      "step": 15000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.718284733641323e-05,
+      "loss": 0.968,
+      "step": 15020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.712173290363514e-05,
+      "loss": 0.9642,
+      "step": 15040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.706056100450831e-05,
+      "loss": 1.0196,
+      "step": 15060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.699933176864558e-05,
+      "loss": 0.9708,
+      "step": 15080
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.693804532578131e-05,
+      "loss": 0.9916,
+      "step": 15100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.687670180577109e-05,
+      "loss": 1.0076,
+      "step": 15120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.681530133859142e-05,
+      "loss": 0.9733,
+      "step": 15140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.675384405433947e-05,
+      "loss": 0.9965,
+      "step": 15160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.66923300832328e-05,
+      "loss": 0.9941,
+      "step": 15180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.663075955560906e-05,
+      "loss": 0.9851,
+      "step": 15200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.656913260192574e-05,
+      "loss": 0.9879,
+      "step": 15220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.650744935275992e-05,
+      "loss": 1.0127,
+      "step": 15240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.644570993880791e-05,
+      "loss": 0.9718,
+      "step": 15260
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.63839144908851e-05,
+      "loss": 0.9495,
+      "step": 15280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.632206313992548e-05,
+      "loss": 0.9745,
+      "step": 15300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.626015601698163e-05,
+      "loss": 0.9862,
+      "step": 15320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.619819325322422e-05,
+      "loss": 0.9923,
+      "step": 15340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.613617497994178e-05,
+      "loss": 0.9779,
+      "step": 15360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.607410132854059e-05,
+      "loss": 0.9875,
+      "step": 15380
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.60119724305441e-05,
+      "loss": 1.011,
+      "step": 15400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.594978841759297e-05,
+      "loss": 0.9933,
+      "step": 15420
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.588754942144452e-05,
+      "loss": 0.9842,
+      "step": 15440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.582525557397264e-05,
+      "loss": 0.9784,
+      "step": 15460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.576290700716742e-05,
+      "loss": 0.9794,
+      "step": 15480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.570050385313487e-05,
+      "loss": 1.0136,
+      "step": 15500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.563804624409672e-05,
+      "loss": 1.0115,
+      "step": 15520
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.557553431239002e-05,
+      "loss": 0.9926,
+      "step": 15540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.551296819046693e-05,
+      "loss": 0.9946,
+      "step": 15560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.545034801089448e-05,
+      "loss": 0.9707,
+      "step": 15580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.538767390635416e-05,
+      "loss": 0.9644,
+      "step": 15600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.53249460096418e-05,
+      "loss": 0.9909,
+      "step": 15620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.526216445366713e-05,
+      "loss": 0.994,
+      "step": 15640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.519932937145364e-05,
+      "loss": 0.9701,
+      "step": 15660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.513644089613818e-05,
+      "loss": 0.9868,
+      "step": 15680
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.507349916097077e-05,
+      "loss": 1.018,
+      "step": 15700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.501050429931429e-05,
+      "loss": 0.9759,
+      "step": 15720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.49474564446441e-05,
+      "loss": 1.0035,
+      "step": 15740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.488435573054795e-05,
+      "loss": 0.9836,
+      "step": 15760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.482120229072552e-05,
+      "loss": 0.9725,
+      "step": 15780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.475799625898825e-05,
+      "loss": 0.9832,
+      "step": 15800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.469473776925897e-05,
+      "loss": 0.9895,
+      "step": 15820
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463142695557171e-05,
+      "loss": 0.979,
+      "step": 15840
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.456806395207132e-05,
+      "loss": 0.987,
+      "step": 15860
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.450464889301326e-05,
+      "loss": 0.9911,
+      "step": 15880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.444118191276326e-05,
+      "loss": 0.9616,
+      "step": 15900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.43776631457971e-05,
+      "loss": 0.9772,
+      "step": 15920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.431409272670027e-05,
+      "loss": 0.9851,
+      "step": 15940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.425047079016765e-05,
+      "loss": 0.9971,
+      "step": 15960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.418679747100339e-05,
+      "loss": 0.9858,
+      "step": 15980
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.412307290412041e-05,
+      "loss": 0.9759,
+      "step": 16000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.405929722454026e-05,
+      "loss": 1.0255,
+      "step": 16020
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.399547056739278e-05,
+      "loss": 0.9645,
+      "step": 16040
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.39315930679158e-05,
+      "loss": 0.9821,
+      "step": 16060
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.386766486145496e-05,
+      "loss": 0.9783,
+      "step": 16080
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.380368608346322e-05,
+      "loss": 0.9899,
+      "step": 16100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.373965686950078e-05,
+      "loss": 0.9705,
+      "step": 16120
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.367557735523467e-05,
+      "loss": 0.9869,
+      "step": 16140
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.361144767643849e-05,
+      "loss": 0.983,
+      "step": 16160
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.354726796899219e-05,
+      "loss": 1.0142,
+      "step": 16180
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.348303836888163e-05,
+      "loss": 0.9991,
+      "step": 16200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.341875901219845e-05,
+      "loss": 0.9927,
+      "step": 16220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.33544300351397e-05,
+      "loss": 1.0073,
+      "step": 16240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.329005157400754e-05,
+      "loss": 0.998,
+      "step": 16260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.322562376520904e-05,
+      "loss": 0.9921,
+      "step": 16280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.316114674525578e-05,
+      "loss": 0.9735,
+      "step": 16300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.30966206507636e-05,
+      "loss": 0.9668,
+      "step": 16320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.303204561845236e-05,
+      "loss": 0.9921,
+      "step": 16340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.29674217851456e-05,
+      "loss": 0.9809,
+      "step": 16360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.290274928777024e-05,
+      "loss": 0.9915,
+      "step": 16380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.283802826335635e-05,
+      "loss": 0.9805,
+      "step": 16400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.277325884903674e-05,
+      "loss": 0.9791,
+      "step": 16420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.270844118204688e-05,
+      "loss": 0.9702,
+      "step": 16440
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.264357539972434e-05,
+      "loss": 0.9717,
+      "step": 16460
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.257866163950873e-05,
+      "loss": 0.9982,
+      "step": 16480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.251370003894133e-05,
+      "loss": 0.9703,
+      "step": 16500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.244869073566466e-05,
+      "loss": 0.9482,
+      "step": 16520
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.238363386742249e-05,
+      "loss": 1.0209,
+      "step": 16540
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.23185295720592e-05,
+      "loss": 0.9672,
+      "step": 16560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.225337798751981e-05,
+      "loss": 1.0046,
+      "step": 16580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.218817925184944e-05,
+      "loss": 1.0102,
+      "step": 16600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.212293350319313e-05,
+      "loss": 0.9578,
+      "step": 16620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.205764087979557e-05,
+      "loss": 0.9835,
+      "step": 16640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.199230152000074e-05,
+      "loss": 1.0221,
+      "step": 16660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.192691556225167e-05,
+      "loss": 0.9698,
+      "step": 16680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.186148314509008e-05,
+      "loss": 0.9918,
+      "step": 16700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.179600440715615e-05,
+      "loss": 0.9828,
+      "step": 16720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.173047948718826e-05,
+      "loss": 1.0046,
+      "step": 16740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.166490852402254e-05,
+      "loss": 0.9624,
+      "step": 16760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.159929165659277e-05,
+      "loss": 0.9675,
+      "step": 16780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.153362902392994e-05,
+      "loss": 0.9811,
+      "step": 16800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.146792076516202e-05,
+      "loss": 0.9725,
+      "step": 16820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.140216701951366e-05,
+      "loss": 0.9751,
+      "step": 16840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.13363679263059e-05,
+      "loss": 0.9734,
+      "step": 16860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.127052362495583e-05,
+      "loss": 0.9879,
+      "step": 16880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.120463425497637e-05,
+      "loss": 0.9824,
+      "step": 16900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.11386999559759e-05,
+      "loss": 0.9833,
+      "step": 16920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.1072720867658e-05,
+      "loss": 0.9703,
+      "step": 16940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.100669712982119e-05,
+      "loss": 0.9864,
+      "step": 16960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.094062888235852e-05,
+      "loss": 0.9809,
+      "step": 16980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.087451626525745e-05,
+      "loss": 0.9887,
+      "step": 17000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.080835941859932e-05,
+      "loss": 0.9817,
+      "step": 17020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.074215848255933e-05,
+      "loss": 0.9697,
+      "step": 17040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.067591359740599e-05,
+      "loss": 0.9717,
+      "step": 17060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.060962490350098e-05,
+      "loss": 0.9435,
+      "step": 17080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.05432925412988e-05,
+      "loss": 0.9578,
+      "step": 17100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.047691665134643e-05,
+      "loss": 0.9845,
+      "step": 17120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.041049737428316e-05,
+      "loss": 0.9647,
+      "step": 17140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.034403485084014e-05,
+      "loss": 0.9781,
+      "step": 17160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.027752922184017e-05,
+      "loss": 0.9348,
+      "step": 17180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.021098062819743e-05,
+      "loss": 0.9801,
+      "step": 17200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.014438921091703e-05,
+      "loss": 0.9987,
+      "step": 17220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.007775511109495e-05,
+      "loss": 0.9797,
+      "step": 17240
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.001107846991751e-05,
+      "loss": 0.9882,
+      "step": 17260
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.994435942866117e-05,
+      "loss": 0.9706,
+      "step": 17280
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.98775981286923e-05,
+      "loss": 0.9733,
+      "step": 17300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.981079471146672e-05,
+      "loss": 0.9788,
+      "step": 17320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.974394931852956e-05,
+      "loss": 0.9797,
+      "step": 17340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.967706209151488e-05,
+      "loss": 0.9357,
+      "step": 17360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.96101331721453e-05,
+      "loss": 0.9818,
+      "step": 17380
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.954316270223189e-05,
+      "loss": 0.9728,
+      "step": 17400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.94761508236737e-05,
+      "loss": 1.0087,
+      "step": 17420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.940909767845753e-05,
+      "loss": 0.9928,
+      "step": 17440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.934200340865761e-05,
+      "loss": 0.9464,
+      "step": 17460
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.927486815643528e-05,
+      "loss": 0.9691,
+      "step": 17480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.920769206403881e-05,
+      "loss": 0.9875,
+      "step": 17500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.914047527380288e-05,
+      "loss": 0.9622,
+      "step": 17520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.907321792814848e-05,
+      "loss": 0.9906,
+      "step": 17540
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.900592016958252e-05,
+      "loss": 0.9778,
+      "step": 17560
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.89385821406975e-05,
+      "loss": 0.9606,
+      "step": 17580
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.887120398417132e-05,
+      "loss": 0.9551,
+      "step": 17600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.880378584276682e-05,
+      "loss": 0.9555,
+      "step": 17620
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.87363278593316e-05,
+      "loss": 0.9735,
+      "step": 17640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.866883017679772e-05,
+      "loss": 0.9791,
+      "step": 17660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.860129293818124e-05,
+      "loss": 0.9975,
+      "step": 17680
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.853371628658217e-05,
+      "loss": 0.9816,
+      "step": 17700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.846610036518396e-05,
+      "loss": 0.985,
+      "step": 17720
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.839844531725321e-05,
+      "loss": 0.9667,
+      "step": 17740
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.833075128613955e-05,
+      "loss": 0.95,
+      "step": 17760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.826301841527512e-05,
+      "loss": 0.9984,
+      "step": 17780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.819524684817438e-05,
+      "loss": 0.9795,
+      "step": 17800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.812743672843378e-05,
+      "loss": 0.932,
+      "step": 17820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.805958819973144e-05,
+      "loss": 0.9332,
+      "step": 17840
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.799170140582689e-05,
+      "loss": 0.9567,
+      "step": 17860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.792377649056071e-05,
+      "loss": 0.9452,
+      "step": 17880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.785581359785428e-05,
+      "loss": 0.9466,
+      "step": 17900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.778781287170946e-05,
+      "loss": 0.9355,
+      "step": 17920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.771977445620818e-05,
+      "loss": 0.9449,
+      "step": 17940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.765169849551235e-05,
+      "loss": 0.9313,
+      "step": 17960
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.758358513386335e-05,
+      "loss": 0.9742,
+      "step": 17980
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.751543451558186e-05,
+      "loss": 0.919,
+      "step": 18000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.744724678506746e-05,
+      "loss": 0.955,
+      "step": 18020
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.737902208679837e-05,
+      "loss": 0.961,
+      "step": 18040
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.731076056533114e-05,
+      "loss": 0.9699,
+      "step": 18060
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.724246236530036e-05,
+      "loss": 0.9497,
+      "step": 18080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.717412763141832e-05,
+      "loss": 0.9476,
+      "step": 18100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.710575650847474e-05,
+      "loss": 0.9641,
+      "step": 18120
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.70373491413364e-05,
+      "loss": 0.9534,
+      "step": 18140
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.69689056749469e-05,
+      "loss": 0.9416,
+      "step": 18160
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.690042625432635e-05,
+      "loss": 0.9359,
+      "step": 18180
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6831911024571e-05,
+      "loss": 0.9538,
+      "step": 18200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.676336013085302e-05,
+      "loss": 0.9375,
+      "step": 18220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.669477371842008e-05,
+      "loss": 0.9479,
+      "step": 18240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.662615193259519e-05,
+      "loss": 0.9275,
+      "step": 18260
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.655749491877623e-05,
+      "loss": 0.9417,
+      "step": 18280
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.648880282243579e-05,
+      "loss": 0.9558,
+      "step": 18300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.642007578912074e-05,
+      "loss": 0.9539,
+      "step": 18320
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.635131396445199e-05,
+      "loss": 0.955,
+      "step": 18340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.628251749412421e-05,
+      "loss": 0.9321,
+      "step": 18360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.621368652390542e-05,
+      "loss": 0.9354,
+      "step": 18380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.614482119963677e-05,
+      "loss": 0.932,
+      "step": 18400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.607592166723219e-05,
+      "loss": 0.9318,
+      "step": 18420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.600698807267811e-05,
+      "loss": 0.9294,
+      "step": 18440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.59380205620331e-05,
+      "loss": 0.946,
+      "step": 18460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.586901928142761e-05,
+      "loss": 0.9338,
+      "step": 18480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.579998437706367e-05,
+      "loss": 0.9615,
+      "step": 18500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.573091599521448e-05,
+      "loss": 0.9363,
+      "step": 18520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.566181428222424e-05,
+      "loss": 0.9891,
+      "step": 18540
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.559267938450778e-05,
+      "loss": 0.9679,
+      "step": 18560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.552351144855015e-05,
+      "loss": 0.9197,
+      "step": 18580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.545431062090653e-05,
+      "loss": 0.9464,
+      "step": 18600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.538507704820169e-05,
+      "loss": 0.9829,
+      "step": 18620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.531581087712984e-05,
+      "loss": 0.9383,
+      "step": 18640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.524651225445423e-05,
+      "loss": 0.941,
+      "step": 18660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.517718132700689e-05,
+      "loss": 0.9647,
+      "step": 18680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.510781824168828e-05,
+      "loss": 0.9517,
+      "step": 18700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.5038423145467e-05,
+      "loss": 0.9526,
+      "step": 18720
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.496899618537947e-05,
+      "loss": 0.943,
+      "step": 18740
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.489953750852966e-05,
+      "loss": 0.9427,
+      "step": 18760
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.483004726208873e-05,
+      "loss": 0.9405,
+      "step": 18780
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.476052559329467e-05,
+      "loss": 0.9578,
+      "step": 18800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.469097264945214e-05,
+      "loss": 0.967,
+      "step": 18820
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.4621388577932e-05,
+      "loss": 0.958,
+      "step": 18840
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.45517735261711e-05,
+      "loss": 0.9582,
+      "step": 18860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.448212764167191e-05,
+      "loss": 0.9493,
+      "step": 18880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.441245107200223e-05,
+      "loss": 0.9368,
+      "step": 18900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.43427439647949e-05,
+      "loss": 0.9792,
+      "step": 18920
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.427300646774744e-05,
+      "loss": 0.9427,
+      "step": 18940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.420323872862179e-05,
+      "loss": 0.9504,
+      "step": 18960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.413344089524393e-05,
+      "loss": 0.9439,
+      "step": 18980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.406361311550361e-05,
+      "loss": 0.92,
+      "step": 19000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.399375553735407e-05,
+      "loss": 0.9736,
+      "step": 19020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.392386830881164e-05,
+      "loss": 0.9712,
+      "step": 19040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.385395157795552e-05,
+      "loss": 0.9777,
+      "step": 19060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.378400549292739e-05,
+      "loss": 0.9232,
+      "step": 19080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.371403020193109e-05,
+      "loss": 0.9597,
+      "step": 19100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.364402585323245e-05,
+      "loss": 0.9131,
+      "step": 19120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.357399259515877e-05,
+      "loss": 0.9555,
+      "step": 19140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.350393057609865e-05,
+      "loss": 0.9488,
+      "step": 19160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.343383994450158e-05,
+      "loss": 0.9597,
+      "step": 19180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.336372084887775e-05,
+      "loss": 0.9153,
+      "step": 19200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.329357343779763e-05,
+      "loss": 0.9319,
+      "step": 19220
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.322339785989163e-05,
+      "loss": 0.9723,
+      "step": 19240
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.315319426384993e-05,
+      "loss": 0.9426,
+      "step": 19260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.308296279842205e-05,
+      "loss": 0.9569,
+      "step": 19280
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.301270361241649e-05,
+      "loss": 0.9376,
+      "step": 19300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.294241685470057e-05,
+      "loss": 0.953,
+      "step": 19320
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.287210267420001e-05,
+      "loss": 0.9552,
+      "step": 19340
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.280176121989861e-05,
+      "loss": 0.941,
+      "step": 19360
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.273139264083798e-05,
+      "loss": 0.9632,
+      "step": 19380
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.266099708611719e-05,
+      "loss": 0.9531,
+      "step": 19400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.259057470489246e-05,
+      "loss": 0.9426,
+      "step": 19420
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.252012564637689e-05,
+      "loss": 0.9947,
+      "step": 19440
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.244965005984008e-05,
+      "loss": 0.9713,
+      "step": 19460
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.23791480946078e-05,
+      "loss": 0.9208,
+      "step": 19480
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.23086199000618e-05,
+      "loss": 0.9401,
+      "step": 19500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.223806562563929e-05,
+      "loss": 0.9537,
+      "step": 19520
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.216748542083286e-05,
+      "loss": 0.9889,
+      "step": 19540
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.209687943518996e-05,
+      "loss": 0.9211,
+      "step": 19560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.202624781831268e-05,
+      "loss": 0.9332,
+      "step": 19580
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.195559071985745e-05,
+      "loss": 0.9656,
+      "step": 19600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.188490828953465e-05,
+      "loss": 0.9292,
+      "step": 19620
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.181420067710838e-05,
+      "loss": 0.9479,
+      "step": 19640
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.174346803239604e-05,
+      "loss": 0.9307,
+      "step": 19660
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.167271050526812e-05,
+      "loss": 0.9564,
+      "step": 19680
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.160192824564778e-05,
+      "loss": 0.9316,
+      "step": 19700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.153112140351066e-05,
+      "loss": 0.9171,
+      "step": 19720
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.14602901288844e-05,
+      "loss": 0.9369,
+      "step": 19740
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.138943457184847e-05,
+      "loss": 0.9351,
+      "step": 19760
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.131855488253379e-05,
+      "loss": 0.9421,
+      "step": 19780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.124765121112233e-05,
+      "loss": 0.9732,
+      "step": 19800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.1176723707847e-05,
+      "loss": 0.933,
+      "step": 19820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.110577252299108e-05,
+      "loss": 0.9656,
+      "step": 19840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.103479780688816e-05,
+      "loss": 0.9369,
+      "step": 19860
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.096379970992157e-05,
+      "loss": 0.9743,
+      "step": 19880
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.089277838252422e-05,
+      "loss": 0.9686,
+      "step": 19900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.0821733975178276e-05,
+      "loss": 0.927,
+      "step": 19920
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.0750666638414765e-05,
+      "loss": 0.9462,
+      "step": 19940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.067957652281332e-05,
+      "loss": 0.9591,
+      "step": 19960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.060846377900182e-05,
+      "loss": 0.9595,
+      "step": 19980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.0537328557656105e-05,
+      "loss": 0.9518,
+      "step": 20000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.046617100949965e-05,
+      "loss": 0.9485,
+      "step": 20020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.0394991285303196e-05,
+      "loss": 0.961,
+      "step": 20040
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.03237895358845e-05,
+      "loss": 0.9536,
+      "step": 20060
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.025256591210799e-05,
+      "loss": 0.9449,
+      "step": 20080
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.0181320564884444e-05,
+      "loss": 0.947,
+      "step": 20100
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.011005364517068e-05,
+      "loss": 0.9491,
+      "step": 20120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.003876530396916e-05,
+      "loss": 0.9332,
+      "step": 20140
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.99674556923278e-05,
+      "loss": 0.9612,
+      "step": 20160
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.989612496133956e-05,
+      "loss": 0.9536,
+      "step": 20180
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.9824773262142165e-05,
+      "loss": 0.9351,
+      "step": 20200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.975340074591774e-05,
+      "loss": 0.955,
+      "step": 20220
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.968200756389255e-05,
+      "loss": 0.9472,
+      "step": 20240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.9610593867336614e-05,
+      "loss": 0.9185,
+      "step": 20260
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.9539159807563437e-05,
+      "loss": 0.9523,
+      "step": 20280
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.9467705535929686e-05,
+      "loss": 0.9177,
+      "step": 20300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.939623120383481e-05,
+      "loss": 0.9517,
+      "step": 20320
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.9324736962720805e-05,
+      "loss": 0.9449,
+      "step": 20340
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.925322296407181e-05,
+      "loss": 0.9656,
+      "step": 20360
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.918168935941388e-05,
+      "loss": 0.9617,
+      "step": 20380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.911013630031457e-05,
+      "loss": 0.9479,
+      "step": 20400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.903856393838265e-05,
+      "loss": 0.9431,
+      "step": 20420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.896697242526785e-05,
+      "loss": 0.9583,
+      "step": 20440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.8895361912660374e-05,
+      "loss": 0.9749,
+      "step": 20460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.882373255229081e-05,
+      "loss": 0.9783,
+      "step": 20480
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.875208449592957e-05,
+      "loss": 0.9316,
+      "step": 20500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.868041789538675e-05,
+      "loss": 0.9483,
+      "step": 20520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.8608732902511695e-05,
+      "loss": 0.9384,
+      "step": 20540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.853702966919275e-05,
+      "loss": 0.9093,
+      "step": 20560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.8465308347356895e-05,
+      "loss": 0.9331,
+      "step": 20580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.8393569088969425e-05,
+      "loss": 0.9641,
+      "step": 20600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.8321812046033666e-05,
+      "loss": 0.9628,
+      "step": 20620
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.825003737059062e-05,
+      "loss": 0.9644,
+      "step": 20640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.81782452147186e-05,
+      "loss": 0.9429,
+      "step": 20660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.810643573053306e-05,
+      "loss": 0.9444,
+      "step": 20680
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.803460907018607e-05,
+      "loss": 0.9412,
+      "step": 20700
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.796276538586615e-05,
+      "loss": 0.9411,
+      "step": 20720
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7890904829797856e-05,
+      "loss": 0.9342,
+      "step": 20740
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.781902755424151e-05,
+      "loss": 0.9188,
+      "step": 20760
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7747133711492895e-05,
+      "loss": 0.9423,
+      "step": 20780
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.767522345388282e-05,
+      "loss": 0.9363,
+      "step": 20800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.760329693377693e-05,
+      "loss": 0.9369,
+      "step": 20820
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7531354303575324e-05,
+      "loss": 0.9655,
+      "step": 20840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7459395715712205e-05,
+      "loss": 0.9417,
+      "step": 20860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.738742132265562e-05,
+      "loss": 0.9504,
+      "step": 20880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.731543127690709e-05,
+      "loss": 0.9594,
+      "step": 20900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.724342573100131e-05,
+      "loss": 0.9268,
+      "step": 20920
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7171404837505796e-05,
+      "loss": 0.9299,
+      "step": 20940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.709936874902061e-05,
+      "loss": 0.9372,
+      "step": 20960
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.702731761817799e-05,
+      "loss": 0.9087,
+      "step": 20980
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.695525159764206e-05,
+      "loss": 0.9338,
+      "step": 21000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.688317084010847e-05,
+      "loss": 0.9435,
+      "step": 21020
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.681107549830414e-05,
+      "loss": 0.9552,
+      "step": 21040
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.673896572498683e-05,
+      "loss": 0.9095,
+      "step": 21060
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.6666841672944925e-05,
+      "loss": 0.9398,
+      "step": 21080
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.659470349499707e-05,
+      "loss": 0.949,
+      "step": 21100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.652255134399178e-05,
+      "loss": 0.9129,
+      "step": 21120
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.645038537280726e-05,
+      "loss": 0.9701,
+      "step": 21140
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.6378205734350916e-05,
+      "loss": 0.9328,
+      "step": 21160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.630601258155917e-05,
+      "loss": 0.9371,
+      "step": 21180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.623380606739708e-05,
+      "loss": 0.9384,
+      "step": 21200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.616158634485793e-05,
+      "loss": 0.9557,
+      "step": 21220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.608935356696313e-05,
+      "loss": 0.9288,
+      "step": 21240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.6017107886761634e-05,
+      "loss": 0.9439,
+      "step": 21260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.5944849457329786e-05,
+      "loss": 0.9502,
+      "step": 21280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.5872578431770936e-05,
+      "loss": 0.9466,
+      "step": 21300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.5800294963215116e-05,
+      "loss": 0.9338,
+      "step": 21320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.5727999204818736e-05,
+      "loss": 0.9466,
+      "step": 21340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.565569130976422e-05,
+      "loss": 0.9392,
+      "step": 21360
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.5583371431259745e-05,
+      "loss": 0.9885,
+      "step": 21380
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.551103972253884e-05,
+      "loss": 0.9347,
+      "step": 21400
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.543869633686013e-05,
+      "loss": 0.934,
+      "step": 21420
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.536634142750699e-05,
+      "loss": 0.9347,
+      "step": 21440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.529397514778716e-05,
+      "loss": 0.9379,
+      "step": 21460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.522159765103251e-05,
+      "loss": 0.9509,
+      "step": 21480
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.5149209090598686e-05,
+      "loss": 0.9212,
+      "step": 21500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.5076809619864754e-05,
+      "loss": 0.9608,
+      "step": 21520
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.5004399392232906e-05,
+      "loss": 0.9088,
+      "step": 21540
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.493197856112812e-05,
+      "loss": 0.9442,
+      "step": 21560
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.485954727999785e-05,
+      "loss": 0.9389,
+      "step": 21580
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.478710570231168e-05,
+      "loss": 0.9191,
+      "step": 21600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.4714653981561015e-05,
+      "loss": 0.9181,
+      "step": 21620
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.464219227125877e-05,
+      "loss": 0.9255,
+      "step": 21640
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.4569720724939025e-05,
+      "loss": 0.9249,
+      "step": 21660
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.449723949615664e-05,
+      "loss": 0.9217,
+      "step": 21680
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.442474873848706e-05,
+      "loss": 0.9316,
+      "step": 21700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.43522486055259e-05,
+      "loss": 0.9689,
+      "step": 21720
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.427973925088865e-05,
+      "loss": 0.931,
+      "step": 21740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.42072208282103e-05,
+      "loss": 0.9415,
+      "step": 21760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.4134693491145085e-05,
+      "loss": 0.9149,
+      "step": 21780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.4062157393366134e-05,
+      "loss": 0.9355,
+      "step": 21800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.398961268856512e-05,
+      "loss": 0.9458,
+      "step": 21820
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.391705953045195e-05,
+      "loss": 0.9411,
+      "step": 21840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.3844498072754476e-05,
+      "loss": 0.9501,
+      "step": 21860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.377192846921808e-05,
+      "loss": 0.9204,
+      "step": 21880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.369935087360547e-05,
+      "loss": 0.9033,
+      "step": 21900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.362676543969622e-05,
+      "loss": 0.9327,
+      "step": 21920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.3554172321286576e-05,
+      "loss": 0.9579,
+      "step": 21940
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.348157167218901e-05,
+      "loss": 0.9313,
+      "step": 21960
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.340896364623198e-05,
+      "loss": 0.913,
+      "step": 21980
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.333634839725958e-05,
+      "loss": 0.9628,
+      "step": 22000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.3263726079131194e-05,
+      "loss": 0.9603,
+      "step": 22020
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.319109684572118e-05,
+      "loss": 0.9116,
+      "step": 22040
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.311846085091856e-05,
+      "loss": 0.9344,
+      "step": 22060
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3045818248626676e-05,
+      "loss": 0.9263,
+      "step": 22080
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.29731691927629e-05,
+      "loss": 0.9622,
+      "step": 22100
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.29005138372582e-05,
+      "loss": 0.9481,
+      "step": 22120
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.282785233605698e-05,
+      "loss": 0.9256,
+      "step": 22140
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.2755184843116635e-05,
+      "loss": 0.9808,
+      "step": 22160
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.268251151240722e-05,
+      "loss": 0.968,
+      "step": 22180
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.2609832497911215e-05,
+      "loss": 0.95,
+      "step": 22200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.253714795362309e-05,
+      "loss": 0.9662,
+      "step": 22220
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.246445803354907e-05,
+      "loss": 0.9352,
+      "step": 22240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.2391762891706764e-05,
+      "loss": 0.9437,
+      "step": 22260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.231906268212483e-05,
+      "loss": 0.9409,
+      "step": 22280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.224635755884268e-05,
+      "loss": 0.9487,
+      "step": 22300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.217364767591014e-05,
+      "loss": 0.9401,
+      "step": 22320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.210093318738709e-05,
+      "loss": 0.952,
+      "step": 22340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.20282142473432e-05,
+      "loss": 0.9752,
+      "step": 22360
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.195549100985756e-05,
+      "loss": 0.9655,
+      "step": 22380
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.188276362901836e-05,
+      "loss": 0.9752,
+      "step": 22400
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.1810032258922605e-05,
+      "loss": 0.9632,
+      "step": 22420
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.173729705367568e-05,
+      "loss": 0.9166,
+      "step": 22440
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.166455816739118e-05,
+      "loss": 0.9433,
+      "step": 22460
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.159181575419043e-05,
+      "loss": 0.9459,
+      "step": 22480
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.151906996820227e-05,
+      "loss": 0.9316,
+      "step": 22500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.144632096356269e-05,
+      "loss": 0.945,
+      "step": 22520
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.137356889441444e-05,
+      "loss": 0.9192,
+      "step": 22540
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.1300813914906853e-05,
+      "loss": 0.9338,
+      "step": 22560
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.122805617919536e-05,
+      "loss": 0.9607,
+      "step": 22580
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.115529584144125e-05,
+      "loss": 0.9123,
+      "step": 22600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.108253305581134e-05,
+      "loss": 0.9547,
+      "step": 22620
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.100976797647761e-05,
+      "loss": 0.9258,
+      "step": 22640
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0937000757616934e-05,
+      "loss": 0.9028,
+      "step": 22660
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.086423155341068e-05,
+      "loss": 0.9422,
+      "step": 22680
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.079146051804444e-05,
+      "loss": 0.9331,
+      "step": 22700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.071868780570772e-05,
+      "loss": 0.9227,
+      "step": 22720
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.0645913570593484e-05,
+      "loss": 0.9301,
+      "step": 22740
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.057313796689804e-05,
+      "loss": 0.9385,
+      "step": 22760
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.050036114882052e-05,
+      "loss": 0.9192,
+      "step": 22780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.042758327056265e-05,
+      "loss": 0.9302,
+      "step": 22800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.03548044863284e-05,
+      "loss": 0.9305,
+      "step": 22820
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.028202495032366e-05,
+      "loss": 0.9267,
+      "step": 22840
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.020924481675593e-05,
+      "loss": 0.9608,
+      "step": 22860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.013646423983392e-05,
+      "loss": 0.9333,
+      "step": 22880
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.006368337376737e-05,
+      "loss": 0.9643,
+      "step": 22900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.999090237276657e-05,
+      "loss": 0.9209,
+      "step": 22920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.991812139104207e-05,
+      "loss": 0.9363,
+      "step": 22940
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.984534058280445e-05,
+      "loss": 0.9439,
+      "step": 22960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.97725601022639e-05,
+      "loss": 0.9514,
+      "step": 22980
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.969978010362989e-05,
+      "loss": 0.9453,
+      "step": 23000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.9627000741110865e-05,
+      "loss": 0.9394,
+      "step": 23020
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.955422216891397e-05,
+      "loss": 0.9316,
+      "step": 23040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.9481444541244665e-05,
+      "loss": 0.9088,
+      "step": 23060
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.9408668012306344e-05,
+      "loss": 0.9212,
+      "step": 23080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.933589273630013e-05,
+      "loss": 0.9114,
+      "step": 23100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.9263118867424515e-05,
+      "loss": 0.9269,
+      "step": 23120
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.919034655987493e-05,
+      "loss": 0.9383,
+      "step": 23140
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.911757596784357e-05,
+      "loss": 0.9495,
+      "step": 23160
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.904480724551897e-05,
+      "loss": 0.9556,
+      "step": 23180
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.89720405470857e-05,
+      "loss": 0.9101,
+      "step": 23200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.8899276026724034e-05,
+      "loss": 0.9385,
+      "step": 23220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.882651383860963e-05,
+      "loss": 0.9146,
+      "step": 23240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.875375413691327e-05,
+      "loss": 0.8875,
+      "step": 23260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.868099707580035e-05,
+      "loss": 0.9435,
+      "step": 23280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8608242809430744e-05,
+      "loss": 0.9215,
+      "step": 23300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8535491491958415e-05,
+      "loss": 0.9206,
+      "step": 23320
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.846274327753107e-05,
+      "loss": 0.9159,
+      "step": 23340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.8389998320289785e-05,
+      "loss": 0.9272,
+      "step": 23360
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8317256774368815e-05,
+      "loss": 0.9164,
+      "step": 23380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.824451879389513e-05,
+      "loss": 0.9521,
+      "step": 23400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8171784532988165e-05,
+      "loss": 0.9555,
+      "step": 23420
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.809905414575947e-05,
+      "loss": 0.9317,
+      "step": 23440
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.802632778631241e-05,
+      "loss": 0.9336,
+      "step": 23460
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.795360560874181e-05,
+      "loss": 0.9299,
+      "step": 23480
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.7880887767133565e-05,
+      "loss": 0.9365,
+      "step": 23500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.7808174415564484e-05,
+      "loss": 0.9178,
+      "step": 23520
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.773546570810182e-05,
+      "loss": 0.931,
+      "step": 23540
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.766276179880296e-05,
+      "loss": 0.9326,
+      "step": 23560
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.759006284171515e-05,
+      "loss": 0.9195,
+      "step": 23580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.7517368990875146e-05,
+      "loss": 0.9536,
+      "step": 23600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.744468040030891e-05,
+      "loss": 0.9326,
+      "step": 23620
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.737199722403117e-05,
+      "loss": 0.9336,
+      "step": 23640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.729931961604529e-05,
+      "loss": 0.9292,
+      "step": 23660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.722664773034278e-05,
+      "loss": 0.9481,
+      "step": 23680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.7153981720902997e-05,
+      "loss": 0.9314,
+      "step": 23700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7081321741692904e-05,
+      "loss": 0.9054,
+      "step": 23720
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7008667946666674e-05,
+      "loss": 0.946,
+      "step": 23740
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.693602048976537e-05,
+      "loss": 0.9389,
+      "step": 23760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.686337952491659e-05,
+      "loss": 0.9484,
+      "step": 23780
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.679074520603423e-05,
+      "loss": 0.9318,
+      "step": 23800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.671811768701811e-05,
+      "loss": 0.9421,
+      "step": 23820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6645497121753564e-05,
+      "loss": 0.9526,
+      "step": 23840
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.657288366411127e-05,
+      "loss": 0.9352,
+      "step": 23860
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.650027746794686e-05,
+      "loss": 0.9146,
+      "step": 23880
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.642767868710045e-05,
+      "loss": 0.9241,
+      "step": 23900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.635508747539661e-05,
+      "loss": 0.9173,
+      "step": 23920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6282503986643775e-05,
+      "loss": 0.9367,
+      "step": 23940
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6209928374634036e-05,
+      "loss": 0.939,
+      "step": 23960
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6137360793142794e-05,
+      "loss": 0.9138,
+      "step": 23980
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.606480139592843e-05,
+      "loss": 0.9526,
+      "step": 24000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.599225033673203e-05,
+      "loss": 0.9391,
+      "step": 24020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.591970776927692e-05,
+      "loss": 0.9484,
+      "step": 24040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.584717384726853e-05,
+      "loss": 0.9413,
+      "step": 24060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.577464872439391e-05,
+      "loss": 0.9497,
+      "step": 24080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.57021325543215e-05,
+      "loss": 0.9306,
+      "step": 24100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.562962549070074e-05,
+      "loss": 0.9218,
+      "step": 24120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.555712768716179e-05,
+      "loss": 0.9342,
+      "step": 24140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548463929731522e-05,
+      "loss": 0.9354,
+      "step": 24160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5412160474751595e-05,
+      "loss": 0.9286,
+      "step": 24180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5339691373041236e-05,
+      "loss": 0.9458,
+      "step": 24200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.526723214573389e-05,
+      "loss": 0.956,
+      "step": 24220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.519478294635837e-05,
+      "loss": 0.9322,
+      "step": 24240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.51223439284222e-05,
+      "loss": 0.9161,
+      "step": 24260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.504991524541138e-05,
+      "loss": 0.9273,
+      "step": 24280
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.497749705079001e-05,
+      "loss": 0.9667,
+      "step": 24300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.490508949799993e-05,
+      "loss": 0.9419,
+      "step": 24320
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.483269274046046e-05,
+      "loss": 0.9533,
+      "step": 24340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.4760306931568044e-05,
+      "loss": 0.9396,
+      "step": 24360
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.468793222469596e-05,
+      "loss": 0.917,
+      "step": 24380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.461556877319385e-05,
+      "loss": 0.9475,
+      "step": 24400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.454321673038766e-05,
+      "loss": 0.9314,
+      "step": 24420
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.447087624957906e-05,
+      "loss": 0.9221,
+      "step": 24440
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.4398547484045245e-05,
+      "loss": 0.9088,
+      "step": 24460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.4326230587038594e-05,
+      "loss": 0.9398,
+      "step": 24480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.425392571178635e-05,
+      "loss": 0.9181,
+      "step": 24500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.418163301149027e-05,
+      "loss": 0.9506,
+      "step": 24520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.41093526393263e-05,
+      "loss": 0.9425,
+      "step": 24540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.4037084748444284e-05,
+      "loss": 0.931,
+      "step": 24560
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.3964829491967655e-05,
+      "loss": 0.9209,
+      "step": 24580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.389258702299298e-05,
+      "loss": 0.9217,
+      "step": 24600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.3820357494589816e-05,
+      "loss": 0.9193,
+      "step": 24620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.3748141059800276e-05,
+      "loss": 0.9129,
+      "step": 24640
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.367593787163875e-05,
+      "loss": 0.9262,
+      "step": 24660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.3603748083091495e-05,
+      "loss": 0.9432,
+      "step": 24680
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.353157184711645e-05,
+      "loss": 0.942,
+      "step": 24700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.34594093166428e-05,
+      "loss": 0.94,
+      "step": 24720
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.33872606445707e-05,
+      "loss": 0.9462,
+      "step": 24740
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.331512598377092e-05,
+      "loss": 0.9453,
+      "step": 24760
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.3243005487084595e-05,
+      "loss": 0.9504,
+      "step": 24780
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.3170899307322826e-05,
+      "loss": 0.9104,
+      "step": 24800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.309880759726633e-05,
+      "loss": 0.9403,
+      "step": 24820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.302673050966523e-05,
+      "loss": 0.9555,
+      "step": 24840
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.295466819723864e-05,
+      "loss": 0.9586,
+      "step": 24860
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.288262081267435e-05,
+      "loss": 0.9811,
+      "step": 24880
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.281058850862856e-05,
+      "loss": 0.929,
+      "step": 24900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.27385714377255e-05,
+      "loss": 0.9444,
+      "step": 24920
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.266656975255709e-05,
+      "loss": 0.9514,
+      "step": 24940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.259458360568271e-05,
+      "loss": 0.9328,
+      "step": 24960
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.252261314962878e-05,
+      "loss": 0.9218,
+      "step": 24980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.245065853688848e-05,
+      "loss": 0.9279,
+      "step": 25000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.237871991992142e-05,
+      "loss": 0.9317,
+      "step": 25020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2306797451153314e-05,
+      "loss": 0.9481,
+      "step": 25040
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.223489128297568e-05,
+      "loss": 0.9363,
+      "step": 25060
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.216300156774548e-05,
+      "loss": 0.9445,
+      "step": 25080
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.209112845778481e-05,
+      "loss": 0.917,
+      "step": 25100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.201927210538058e-05,
+      "loss": 0.9566,
+      "step": 25120
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.194743266278426e-05,
+      "loss": 0.9185,
+      "step": 25140
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1875610282211364e-05,
+      "loss": 0.9006,
+      "step": 25160
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1803805115841366e-05,
+      "loss": 0.92,
+      "step": 25180
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.173201731581724e-05,
+      "loss": 0.9291,
+      "step": 25200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.166024703424511e-05,
+      "loss": 0.9299,
+      "step": 25220
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.1588494423194046e-05,
+      "loss": 0.9074,
+      "step": 25240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.151675963469565e-05,
+      "loss": 0.9473,
+      "step": 25260
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.1445042820743764e-05,
+      "loss": 0.9326,
+      "step": 25280
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.137334413329414e-05,
+      "loss": 0.9276,
+      "step": 25300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.130166372426412e-05,
+      "loss": 0.9333,
+      "step": 25320
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.123000174553235e-05,
+      "loss": 0.9269,
+      "step": 25340
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.1158358348938374e-05,
+      "loss": 0.9264,
+      "step": 25360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.1086733686282395e-05,
+      "loss": 0.9503,
+      "step": 25380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.1015127909324936e-05,
+      "loss": 0.9071,
+      "step": 25400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.094354116978647e-05,
+      "loss": 0.9654,
+      "step": 25420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.087197361934714e-05,
+      "loss": 0.949,
+      "step": 25440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0800425409646456e-05,
+      "loss": 0.9102,
+      "step": 25460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0728896692282926e-05,
+      "loss": 0.943,
+      "step": 25480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.065738761881375e-05,
+      "loss": 0.9392,
+      "step": 25500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.0585898340754506e-05,
+      "loss": 0.9151,
+      "step": 25520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.051442900957888e-05,
+      "loss": 0.937,
+      "step": 25540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.0442979776718237e-05,
+      "loss": 0.9283,
+      "step": 25560
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.037155079356137e-05,
+      "loss": 0.92,
+      "step": 25580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.030014221145417e-05,
+      "loss": 0.9313,
+      "step": 25600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.022875418169931e-05,
+      "loss": 0.9367,
+      "step": 25620
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.0157386855555906e-05,
+      "loss": 0.9642,
+      "step": 25640
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.00860403842392e-05,
+      "loss": 0.9027,
+      "step": 25660
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.001471491892026e-05,
+      "loss": 0.9215,
+      "step": 25680
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.9943410610725665e-05,
+      "loss": 0.9546,
+      "step": 25700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.9872127610737095e-05,
+      "loss": 0.909,
+      "step": 25720
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.9800866069991173e-05,
+      "loss": 0.9495,
+      "step": 25740
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.9729626139478995e-05,
+      "loss": 0.9311,
+      "step": 25760
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.965840797014586e-05,
+      "loss": 0.9387,
+      "step": 25780
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9587211712891005e-05,
+      "loss": 0.919,
+      "step": 25800
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9516037518567204e-05,
+      "loss": 0.938,
+      "step": 25820
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9444885537980526e-05,
+      "loss": 0.929,
+      "step": 25840
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.9373755921889886e-05,
+      "loss": 0.9502,
+      "step": 25860
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.93026488210069e-05,
+      "loss": 0.9294,
+      "step": 25880
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.9231564385995476e-05,
+      "loss": 0.9643,
+      "step": 25900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.91605027674714e-05,
+      "loss": 0.9125,
+      "step": 25920
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.908946411600222e-05,
+      "loss": 0.9356,
+      "step": 25940
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9018448582106795e-05,
+      "loss": 0.9298,
+      "step": 25960
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.894745631625495e-05,
+      "loss": 0.9532,
+      "step": 25980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.887648746886727e-05,
+      "loss": 0.9326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8805542190314705e-05,
+      "loss": 0.9003,
+      "step": 26020
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.873462063091825e-05,
+      "loss": 0.9361,
+      "step": 26040
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.866372294094864e-05,
+      "loss": 0.9186,
+      "step": 26060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.859284927062604e-05,
+      "loss": 0.9442,
+      "step": 26080
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.8521999770119786e-05,
+      "loss": 0.929,
+      "step": 26100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.845117458954787e-05,
+      "loss": 0.9241,
+      "step": 26120
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.838037387897688e-05,
+      "loss": 0.9246,
+      "step": 26140
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8309597788421474e-05,
+      "loss": 0.9379,
+      "step": 26160
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.823884646784421e-05,
+      "loss": 0.9221,
+      "step": 26180
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8168120067155096e-05,
+      "loss": 0.9069,
+      "step": 26200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.809741873621138e-05,
+      "loss": 0.9144,
+      "step": 26220
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.802674262481719e-05,
+      "loss": 0.9446,
+      "step": 26240
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.79560918827232e-05,
+      "loss": 0.972,
+      "step": 26260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7885466659626334e-05,
+      "loss": 0.9395,
+      "step": 26280
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.781486710516948e-05,
+      "loss": 0.9471,
+      "step": 26300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.77442933689411e-05,
+      "loss": 0.9502,
+      "step": 26320
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.767374560047495e-05,
+      "loss": 0.9296,
+      "step": 26340
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.760322394924979e-05,
+      "loss": 0.9643,
+      "step": 26360
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.753272856468903e-05,
+      "loss": 0.929,
+      "step": 26380
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.746225959616042e-05,
+      "loss": 0.9233,
+      "step": 26400
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7391817192975745e-05,
+      "loss": 0.9169,
+      "step": 26420
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.732140150439048e-05,
+      "loss": 0.891,
+      "step": 26440
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.725101267960359e-05,
+      "loss": 0.9535,
+      "step": 26460
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.718065086775695e-05,
+      "loss": 0.9363,
+      "step": 26480
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7110316217935357e-05,
+      "loss": 0.964,
+      "step": 26500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7040008879166e-05,
+      "loss": 0.9386,
+      "step": 26520
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.696972900041816e-05,
+      "loss": 0.9268,
+      "step": 26540
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6899476730603e-05,
+      "loss": 0.9139,
+      "step": 26560
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.682925221857315e-05,
+      "loss": 0.9274,
+      "step": 26580
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.675905561312244e-05,
+      "loss": 0.9195,
+      "step": 26600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.668888706298554e-05,
+      "loss": 0.9289,
+      "step": 26620
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.66187467168377e-05,
+      "loss": 0.9429,
+      "step": 26640
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.654863472329445e-05,
+      "loss": 0.9156,
+      "step": 26660
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.647855123091115e-05,
+      "loss": 0.9355,
+      "step": 26680
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.640849638818286e-05,
+      "loss": 0.9313,
+      "step": 26700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.633847034354389e-05,
+      "loss": 0.86,
+      "step": 26720
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.626847324536755e-05,
+      "loss": 0.9101,
+      "step": 26740
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.6198505241965806e-05,
+      "loss": 0.9232,
+      "step": 26760
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.6128566481588977e-05,
+      "loss": 0.9325,
+      "step": 26780
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.605865711242544e-05,
+      "loss": 0.9364,
+      "step": 26800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.598877728260127e-05,
+      "loss": 0.8978,
+      "step": 26820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.591892714017995e-05,
+      "loss": 0.892,
+      "step": 26840
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5849106833162124e-05,
+      "loss": 0.8974,
+      "step": 26860
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.577931650948512e-05,
+      "loss": 0.8923,
+      "step": 26880
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5709556317022823e-05,
+      "loss": 0.9018,
+      "step": 26900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.563982640358523e-05,
+      "loss": 0.9101,
+      "step": 26920
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.55701269169182e-05,
+      "loss": 0.8988,
+      "step": 26940
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.550045800470311e-05,
+      "loss": 0.8896,
+      "step": 26960
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.5430819814556544e-05,
+      "loss": 0.9021,
+      "step": 26980
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.536121249403004e-05,
+      "loss": 0.8667,
+      "step": 27000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.5291636190609665e-05,
+      "loss": 0.8931,
+      "step": 27020
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.52220910517158e-05,
+      "loss": 0.8927,
+      "step": 27040
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.515257722470281e-05,
+      "loss": 0.9163,
+      "step": 27060
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.50830948568587e-05,
+      "loss": 0.9169,
+      "step": 27080
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.50136440954048e-05,
+      "loss": 0.9031,
+      "step": 27100
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.494422508749547e-05,
+      "loss": 0.9213,
+      "step": 27120
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.487483798021785e-05,
+      "loss": 0.9082,
+      "step": 27140
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.480548292059139e-05,
+      "loss": 0.911,
+      "step": 27160
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.473616005556773e-05,
+      "loss": 0.9075,
+      "step": 27180
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4666869532030224e-05,
+      "loss": 0.9072,
+      "step": 27200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.459761149679378e-05,
+      "loss": 0.9181,
+      "step": 27220
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4528386096604366e-05,
+      "loss": 0.908,
+      "step": 27240
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.445919347813888e-05,
+      "loss": 0.8838,
+      "step": 27260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.439003378800475e-05,
+      "loss": 0.8977,
+      "step": 27280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4320907172739594e-05,
+      "loss": 0.9024,
+      "step": 27300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.425181377881099e-05,
+      "loss": 0.9174,
+      "step": 27320
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4182753752616094e-05,
+      "loss": 0.916,
+      "step": 27340
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.411372724048144e-05,
+      "loss": 0.9103,
+      "step": 27360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4044734388662426e-05,
+      "loss": 0.8922,
+      "step": 27380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3975775343343205e-05,
+      "loss": 0.8991,
+      "step": 27400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.390685025063633e-05,
+      "loss": 0.8822,
+      "step": 27420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.383795925658233e-05,
+      "loss": 0.9007,
+      "step": 27440
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.376910250714955e-05,
+      "loss": 0.9058,
+      "step": 27460
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.370028014823375e-05,
+      "loss": 0.9046,
+      "step": 27480
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.363149232565785e-05,
+      "loss": 0.9123,
+      "step": 27500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.356273918517153e-05,
+      "loss": 0.8856,
+      "step": 27520
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.349402087245104e-05,
+      "loss": 0.9146,
+      "step": 27540
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.342533753309887e-05,
+      "loss": 0.9106,
+      "step": 27560
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.335668931264327e-05,
+      "loss": 0.8902,
+      "step": 27580
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.328807635653822e-05,
+      "loss": 0.8881,
+      "step": 27600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.321949881016293e-05,
+      "loss": 0.9172,
+      "step": 27620
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.315095681882159e-05,
+      "loss": 0.9076,
+      "step": 27640
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.3082450527743014e-05,
+      "loss": 0.8837,
+      "step": 27660
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.301398008208042e-05,
+      "loss": 0.8725,
+      "step": 27680
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.294554562691108e-05,
+      "loss": 0.9202,
+      "step": 27700
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.287714730723596e-05,
+      "loss": 0.909,
+      "step": 27720
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.280878526797948e-05,
+      "loss": 0.9094,
+      "step": 27740
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.274045965398924e-05,
+      "loss": 0.8797,
+      "step": 27760
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.267217061003562e-05,
+      "loss": 0.8962,
+      "step": 27780
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.260391828081147e-05,
+      "loss": 0.8772,
+      "step": 27800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.253570281093192e-05,
+      "loss": 0.8907,
+      "step": 27820
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.246752434493398e-05,
+      "loss": 0.8898,
+      "step": 27840
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.239938302727622e-05,
+      "loss": 0.8992,
+      "step": 27860
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.233127900233855e-05,
+      "loss": 0.8948,
+      "step": 27880
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2263212414421846e-05,
+      "loss": 0.9386,
+      "step": 27900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.219518340774763e-05,
+      "loss": 0.915,
+      "step": 27920
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2127192126457815e-05,
+      "loss": 0.9026,
+      "step": 27940
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.205923871461442e-05,
+      "loss": 0.8793,
+      "step": 27960
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.19913233161992e-05,
+      "loss": 0.9182,
+      "step": 27980
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.192344607511329e-05,
+      "loss": 0.8803,
+      "step": 28000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.18556071351771e-05,
+      "loss": 0.872,
+      "step": 28020
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1787806640129826e-05,
+      "loss": 0.899,
+      "step": 28040
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1720044733629196e-05,
+      "loss": 0.9047,
+      "step": 28060
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.165232155925118e-05,
+      "loss": 0.8979,
+      "step": 28080
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.15846372604897e-05,
+      "loss": 0.8833,
+      "step": 28100
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.151699198075633e-05,
+      "loss": 0.908,
+      "step": 28120
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.1449385863379866e-05,
+      "loss": 0.8998,
+      "step": 28140
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.138181905160625e-05,
+      "loss": 0.8975,
+      "step": 28160
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.13142916885981e-05,
+      "loss": 0.8921,
+      "step": 28180
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.124680391743438e-05,
+      "loss": 0.9263,
+      "step": 28200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.117935588111026e-05,
+      "loss": 0.9153,
+      "step": 28220
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.111194772253668e-05,
+      "loss": 0.9274,
+      "step": 28240
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.104457958454009e-05,
+      "loss": 0.9159,
+      "step": 28260
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.097725160986212e-05,
+      "loss": 0.9314,
+      "step": 28280
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.090996394115933e-05,
+      "loss": 0.9059,
+      "step": 28300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.0842716721002894e-05,
+      "loss": 0.9248,
+      "step": 28320
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.077551009187821e-05,
+      "loss": 0.9125,
+      "step": 28340
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.0708344196184756e-05,
+      "loss": 0.9084,
+      "step": 28360
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.064121917623566e-05,
+      "loss": 0.9046,
+      "step": 28380
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.0574135174257444e-05,
+      "loss": 0.8961,
+      "step": 28400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.050709233238972e-05,
+      "loss": 0.9001,
+      "step": 28420
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0440090792684884e-05,
+      "loss": 0.91,
+      "step": 28440
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.037313069710784e-05,
+      "loss": 0.9225,
+      "step": 28460
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0306212187535653e-05,
+      "loss": 0.8991,
+      "step": 28480
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0239335405757275e-05,
+      "loss": 0.8773,
+      "step": 28500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0172500493473294e-05,
+      "loss": 0.8922,
+      "step": 28520
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.0105707592295528e-05,
+      "loss": 0.9035,
+      "step": 28540
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.003895684374679e-05,
+      "loss": 0.9027,
+      "step": 28560
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.9972248389260593e-05,
+      "loss": 0.9163,
+      "step": 28580
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.9905582370180836e-05,
+      "loss": 0.909,
+      "step": 28600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.9838958927761477e-05,
+      "loss": 0.8952,
+      "step": 28620
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.9772378203166307e-05,
+      "loss": 0.9269,
+      "step": 28640
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.9705840337468554e-05,
+      "loss": 0.8917,
+      "step": 28660
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.9639345471650716e-05,
+      "loss": 0.8882,
+      "step": 28680
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.9572893746604052e-05,
+      "loss": 0.9008,
+      "step": 28700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.950648530312854e-05,
+      "loss": 0.9153,
+      "step": 28720
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.9440120281932403e-05,
+      "loss": 0.8977,
+      "step": 28740
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.937379882363183e-05,
+      "loss": 0.9006,
+      "step": 28760
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.9307521068750748e-05,
+      "loss": 0.921,
+      "step": 28780
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.924128715772047e-05,
+      "loss": 0.8782,
+      "step": 28800
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.9175097230879423e-05,
+      "loss": 0.8994,
+      "step": 28820
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.9108951428472804e-05,
+      "loss": 0.8945,
+      "step": 28840
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.9042849890652352e-05,
+      "loss": 0.8867,
+      "step": 28860
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.8976792757476013e-05,
+      "loss": 0.8793,
+      "step": 28880
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.891078016890763e-05,
+      "loss": 0.9037,
+      "step": 28900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.8844812264816684e-05,
+      "loss": 0.9293,
+      "step": 28920
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.8778889184977986e-05,
+      "loss": 0.8962,
+      "step": 28940
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.8713011069071306e-05,
+      "loss": 0.886,
+      "step": 28960
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.8647178056681194e-05,
+      "loss": 0.8791,
+      "step": 28980
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.8581390287296672e-05,
+      "loss": 0.9162,
+      "step": 29000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.851564790031086e-05,
+      "loss": 0.9088,
+      "step": 29020
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.8449951035020672e-05,
+      "loss": 0.9208,
+      "step": 29040
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.8384299830626637e-05,
+      "loss": 0.8747,
+      "step": 29060
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.8318694426232516e-05,
+      "loss": 0.8721,
+      "step": 29080
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.825313496084503e-05,
+      "loss": 0.9281,
+      "step": 29100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.8187621573373544e-05,
+      "loss": 0.9045,
+      "step": 29120
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.8122154402629818e-05,
+      "loss": 0.8925,
+      "step": 29140
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.8056733587327694e-05,
+      "loss": 0.8958,
+      "step": 29160
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.7991359266082717e-05,
+      "loss": 0.9155,
+      "step": 29180
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.7926031577412038e-05,
+      "loss": 0.8971,
+      "step": 29200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.7860750659733938e-05,
+      "loss": 0.9249,
+      "step": 29220
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.779551665136756e-05,
+      "loss": 0.9045,
+      "step": 29240
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.773032969053273e-05,
+      "loss": 0.9207,
+      "step": 29260
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.7665189915349533e-05,
+      "loss": 0.8938,
+      "step": 29280
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.7600097463838114e-05,
+      "loss": 0.9088,
+      "step": 29300
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.753505247391832e-05,
+      "loss": 0.9085,
+      "step": 29320
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.7470055083409452e-05,
+      "loss": 0.8978,
+      "step": 29340
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.740510543002996e-05,
+      "loss": 0.8963,
+      "step": 29360
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.734020365139708e-05,
+      "loss": 0.9295,
+      "step": 29380
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.727534988502673e-05,
+      "loss": 0.9076,
+      "step": 29400
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.721054426833301e-05,
+      "loss": 0.9037,
+      "step": 29420
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.7145786938628036e-05,
+      "loss": 0.8821,
+      "step": 29440
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.7081078033121577e-05,
+      "loss": 0.8976,
+      "step": 29460
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.7016417688920815e-05,
+      "loss": 0.9083,
+      "step": 29480
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.695180604303007e-05,
+      "loss": 0.9198,
+      "step": 29500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.6887243232350434e-05,
+      "loss": 0.9052,
+      "step": 29520
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.6822729393679558e-05,
+      "loss": 0.9033,
+      "step": 29540
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.6758264663711306e-05,
+      "loss": 0.9058,
+      "step": 29560
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.6693849179035513e-05,
+      "loss": 0.889,
+      "step": 29580
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.662948307613764e-05,
+      "loss": 0.9118,
+      "step": 29600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.6565166491398553e-05,
+      "loss": 0.8994,
+      "step": 29620
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.6500899561094184e-05,
+      "loss": 0.9048,
+      "step": 29640
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.643668242139522e-05,
+      "loss": 0.9,
+      "step": 29660
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.63725152083669e-05,
+      "loss": 0.9199,
+      "step": 29680
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.630839805796863e-05,
+      "loss": 0.9024,
+      "step": 29700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.624433110605383e-05,
+      "loss": 0.9102,
+      "step": 29720
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.6180314488369452e-05,
+      "loss": 0.8843,
+      "step": 29740
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.611634834055585e-05,
+      "loss": 0.8905,
+      "step": 29760
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.6052432798146436e-05,
+      "loss": 0.8834,
+      "step": 29780
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.5988567996567402e-05,
+      "loss": 0.8973,
+      "step": 29800
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.5924754071137415e-05,
+      "loss": 0.887,
+      "step": 29820
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.5860991157067356e-05,
+      "loss": 0.9162,
+      "step": 29840
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.5797279389460037e-05,
+      "loss": 0.8948,
+      "step": 29860
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.5733618903309843e-05,
+      "loss": 0.897,
+      "step": 29880
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.567000983350254e-05,
+      "loss": 0.875,
+      "step": 29900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.5606452314815e-05,
+      "loss": 0.9287,
+      "step": 29920
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.554294648191477e-05,
+      "loss": 0.8996,
+      "step": 29940
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.5479492469359944e-05,
+      "loss": 0.8989,
+      "step": 29960
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.5416090411598813e-05,
+      "loss": 0.91,
+      "step": 29980
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.535274044296957e-05,
+      "loss": 0.9104,
+      "step": 30000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.5289442697700043e-05,
+      "loss": 0.8902,
+      "step": 30020
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.5226197309907418e-05,
+      "loss": 0.8907,
+      "step": 30040
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.5163004413597955e-05,
+      "loss": 0.9099,
+      "step": 30060
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.5099864142666642e-05,
+      "loss": 0.8979,
+      "step": 30080
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.5036776630896985e-05,
+      "loss": 0.9008,
+      "step": 30100
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.4973742011960775e-05,
+      "loss": 0.9105,
+      "step": 30120
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.4910760419417616e-05,
+      "loss": 0.9075,
+      "step": 30140
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.4847831986714837e-05,
+      "loss": 0.9141,
+      "step": 30160
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.47849568471871e-05,
+      "loss": 0.9281,
+      "step": 30180
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.472213513405615e-05,
+      "loss": 0.9085,
+      "step": 30200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4659366980430547e-05,
+      "loss": 0.9308,
+      "step": 30220
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4596652519305346e-05,
+      "loss": 0.9147,
+      "step": 30240
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4533991883561868e-05,
+      "loss": 0.91,
+      "step": 30260
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4471385205967323e-05,
+      "loss": 0.8888,
+      "step": 30280
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4408832619174644e-05,
+      "loss": 0.894,
+      "step": 30300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.4346334255722168e-05,
+      "loss": 0.8859,
+      "step": 30320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.4283890248033337e-05,
+      "loss": 0.9136,
+      "step": 30340
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.4221500728416356e-05,
+      "loss": 0.9353,
+      "step": 30360
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.415916582906405e-05,
+      "loss": 0.9154,
+      "step": 30380
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.409688568205349e-05,
+      "loss": 0.9022,
+      "step": 30400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.403466041934574e-05,
+      "loss": 0.9071,
+      "step": 30420
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.3972490172785567e-05,
+      "loss": 0.8839,
+      "step": 30440
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.3910375074101172e-05,
+      "loss": 0.9102,
+      "step": 30460
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.3848315254903924e-05,
+      "loss": 0.9196,
+      "step": 30480
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.3786310846688e-05,
+      "loss": 0.8863,
+      "step": 30500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.3724361980830257e-05,
+      "loss": 0.9043,
+      "step": 30520
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.366246878858984e-05,
+      "loss": 0.8813,
+      "step": 30540
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.3600631401107882e-05,
+      "loss": 0.9023,
+      "step": 30560
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.353884994940732e-05,
+      "loss": 0.8872,
+      "step": 30580
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.3477124564392572e-05,
+      "loss": 0.9056,
+      "step": 30600
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.3415455376849248e-05,
+      "loss": 0.8708,
+      "step": 30620
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.3353842517443898e-05,
+      "loss": 0.8727,
+      "step": 30640
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.32922861167237e-05,
+      "loss": 0.8796,
+      "step": 30660
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.3230786305116253e-05,
+      "loss": 0.9369,
+      "step": 30680
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.316934321292915e-05,
+      "loss": 0.9044,
+      "step": 30700
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.3107956970349942e-05,
+      "loss": 0.8945,
+      "step": 30720
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.3046627707445635e-05,
+      "loss": 0.9097,
+      "step": 30740
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.2985355554162546e-05,
+      "loss": 0.908,
+      "step": 30760
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.292414064032593e-05,
+      "loss": 0.8604,
+      "step": 30780
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.2862983095639823e-05,
+      "loss": 0.8698,
+      "step": 30800
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.2801883049686678e-05,
+      "loss": 0.8969,
+      "step": 30820
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2740840631927118e-05,
+      "loss": 0.8927,
+      "step": 30840
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2679855971699676e-05,
+      "loss": 0.9017,
+      "step": 30860
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2618929198220513e-05,
+      "loss": 0.924,
+      "step": 30880
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2558060440583057e-05,
+      "loss": 0.8936,
+      "step": 30900
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2497249827757933e-05,
+      "loss": 0.9073,
+      "step": 30920
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.2436497488592497e-05,
+      "loss": 0.9292,
+      "step": 30940
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.2375803551810654e-05,
+      "loss": 0.9278,
+      "step": 30960
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.2315168146012527e-05,
+      "loss": 0.8894,
+      "step": 30980
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.225459139967426e-05,
+      "loss": 0.8936,
+      "step": 31000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.21940734411477e-05,
+      "loss": 0.9102,
+      "step": 31020
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.213361439866013e-05,
+      "loss": 0.8849,
+      "step": 31040
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.2073214400313997e-05,
+      "loss": 0.8884,
+      "step": 31060
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.201287357408665e-05,
+      "loss": 0.8864,
+      "step": 31080
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.1952592047830055e-05,
+      "loss": 0.94,
+      "step": 31100
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.189236994927054e-05,
+      "loss": 0.8892,
+      "step": 31120
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.1832207406008502e-05,
+      "loss": 0.8934,
+      "step": 31140
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.1772104545518185e-05,
+      "loss": 0.8911,
+      "step": 31160
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.171206149514731e-05,
+      "loss": 0.9029,
+      "step": 31180
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.165207838211693e-05,
+      "loss": 0.9103,
+      "step": 31200
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.159215533352106e-05,
+      "loss": 0.9166,
+      "step": 31220
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.153229247632652e-05,
+      "loss": 0.8993,
+      "step": 31240
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.14724899373725e-05,
+      "loss": 0.9114,
+      "step": 31260
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.141274784337044e-05,
+      "loss": 0.9053,
+      "step": 31280
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1353066320903698e-05,
+      "loss": 0.8942,
+      "step": 31300
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1293445496427296e-05,
+      "loss": 0.8935,
+      "step": 31320
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1233885496267634e-05,
+      "loss": 0.8798,
+      "step": 31340
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.117438644662226e-05,
+      "loss": 0.9204,
+      "step": 31360
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.1114948473559554e-05,
+      "loss": 0.8907,
+      "step": 31380
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.1055571703018474e-05,
+      "loss": 0.8935,
+      "step": 31400
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.0996256260808316e-05,
+      "loss": 0.8761,
+      "step": 31420
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.0937002272608493e-05,
+      "loss": 0.9049,
+      "step": 31440
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.087780986396808e-05,
+      "loss": 0.9202,
+      "step": 31460
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0818679160305776e-05,
+      "loss": 0.8871,
+      "step": 31480
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0759610286909508e-05,
+      "loss": 0.8833,
+      "step": 31500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0700603368936182e-05,
+      "loss": 0.917,
+      "step": 31520
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.064165853141145e-05,
+      "loss": 0.9222,
+      "step": 31540
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.058277589922942e-05,
+      "loss": 0.9001,
+      "step": 31560
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.05239555971524e-05,
+      "loss": 0.906,
+      "step": 31580
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.0465197749810604e-05,
+      "loss": 0.9021,
+      "step": 31600
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.040650248170194e-05,
+      "loss": 0.8962,
+      "step": 31620
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.034786991719174e-05,
+      "loss": 0.901,
+      "step": 31640
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.0289300180512478e-05,
+      "loss": 0.9213,
+      "step": 31660
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.0230793395763426e-05,
+      "loss": 0.9018,
+      "step": 31680
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.0172349686910568e-05,
+      "loss": 0.8924,
+      "step": 31700
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.011396917778619e-05,
+      "loss": 0.8989,
+      "step": 31720
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.0055651992088692e-05,
+      "loss": 0.8872,
+      "step": 31740
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9997398253382287e-05,
+      "loss": 0.9258,
+      "step": 31760
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.993920808509676e-05,
+      "loss": 0.9025,
+      "step": 31780
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.988108161052722e-05,
+      "loss": 0.8995,
+      "step": 31800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9823018952833748e-05,
+      "loss": 0.9158,
+      "step": 31820
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9765020235041322e-05,
+      "loss": 0.9092,
+      "step": 31840
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9707085580039365e-05,
+      "loss": 0.9003,
+      "step": 31860
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9649215110581553e-05,
+      "loss": 0.8768,
+      "step": 31880
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9591408949285605e-05,
+      "loss": 0.9147,
+      "step": 31900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.953366721863297e-05,
+      "loss": 0.9055,
+      "step": 31920
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.947599004096856e-05,
+      "loss": 0.9076,
+      "step": 31940
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9418377538500543e-05,
+      "loss": 0.8803,
+      "step": 31960
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.936082983330002e-05,
+      "loss": 0.9346,
+      "step": 31980
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9303347047300834e-05,
+      "loss": 0.9033,
+      "step": 32000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9245929302299202e-05,
+      "loss": 0.8857,
+      "step": 32020
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9188576719953633e-05,
+      "loss": 0.9229,
+      "step": 32040
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.913128942178451e-05,
+      "loss": 0.9216,
+      "step": 32060
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.907406752917386e-05,
+      "loss": 0.8786,
+      "step": 32080
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9016911163365185e-05,
+      "loss": 0.896,
+      "step": 32100
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.89598204454631e-05,
+      "loss": 0.9214,
+      "step": 32120
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.89027954964332e-05,
+      "loss": 0.8923,
+      "step": 32140
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.8845836437101622e-05,
+      "loss": 0.9045,
+      "step": 32160
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8788943388154962e-05,
+      "loss": 0.8815,
+      "step": 32180
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.873211647013995e-05,
+      "loss": 0.9081,
+      "step": 32200
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.867535580346313e-05,
+      "loss": 0.8995,
+      "step": 32220
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.861866150839078e-05,
+      "loss": 0.9083,
+      "step": 32240
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.856203370504846e-05,
+      "loss": 0.8916,
+      "step": 32260
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.850547251342089e-05,
+      "loss": 0.9364,
+      "step": 32280
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.844897805335162e-05,
+      "loss": 0.8781,
+      "step": 32300
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8392550444542793e-05,
+      "loss": 0.9038,
+      "step": 32320
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8336189806555014e-05,
+      "loss": 0.8806,
+      "step": 32340
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8279896258806844e-05,
+      "loss": 0.9133,
+      "step": 32360
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8223669920574772e-05,
+      "loss": 0.9114,
+      "step": 32380
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8167510910992875e-05,
+      "loss": 0.8787,
+      "step": 32400
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.811141934905255e-05,
+      "loss": 0.8737,
+      "step": 32420
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8055395353602306e-05,
+      "loss": 0.9233,
+      "step": 32440
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7999439043347476e-05,
+      "loss": 0.9049,
+      "step": 32460
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7943550536850006e-05,
+      "loss": 0.893,
+      "step": 32480
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7887729952528117e-05,
+      "loss": 0.9078,
+      "step": 32500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7831977408656153e-05,
+      "loss": 0.9135,
+      "step": 32520
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7776293023364325e-05,
+      "loss": 0.8959,
+      "step": 32540
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7720676914638407e-05,
+      "loss": 0.9199,
+      "step": 32560
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.766512920031944e-05,
+      "loss": 0.9057,
+      "step": 32580
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7609649998103634e-05,
+      "loss": 0.8747,
+      "step": 32600
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.755423942554199e-05,
+      "loss": 0.911,
+      "step": 32620
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.749889760004012e-05,
+      "loss": 0.8998,
+      "step": 32640
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7443624638857954e-05,
+      "loss": 0.9014,
+      "step": 32660
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7388420659109515e-05,
+      "loss": 0.8887,
+      "step": 32680
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7333285777762682e-05,
+      "loss": 0.9168,
+      "step": 32700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.727822011163886e-05,
+      "loss": 0.9106,
+      "step": 32720
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7223223777412905e-05,
+      "loss": 0.907,
+      "step": 32740
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7168296891612707e-05,
+      "loss": 0.906,
+      "step": 32760
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.711343957061899e-05,
+      "loss": 0.8978,
+      "step": 32780
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7058651930665114e-05,
+      "loss": 0.8732,
+      "step": 32800
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.70039340878368e-05,
+      "loss": 0.909,
+      "step": 32820
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6949286158071858e-05,
+      "loss": 0.8888,
+      "step": 32840
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.689470825715998e-05,
+      "loss": 0.9104,
+      "step": 32860
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6840200500742482e-05,
+      "loss": 0.878,
+      "step": 32880
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6785763004312055e-05,
+      "loss": 0.9154,
+      "step": 32900
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.673139588321247e-05,
+      "loss": 0.8783,
+      "step": 32920
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6677099252638477e-05,
+      "loss": 0.8909,
+      "step": 32940
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6622873227635428e-05,
+      "loss": 0.908,
+      "step": 32960
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.656871792309902e-05,
+      "loss": 0.8948,
+      "step": 32980
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.651463345377518e-05,
+      "loss": 0.888,
+      "step": 33000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.6460619934259707e-05,
+      "loss": 0.9002,
+      "step": 33020
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.6406677478998094e-05,
+      "loss": 0.9047,
+      "step": 33040
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.6352806202285244e-05,
+      "loss": 0.8803,
+      "step": 33060
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.6299006218265246e-05,
+      "loss": 0.8966,
+      "step": 33080
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.624527764093115e-05,
+      "loss": 0.8901,
+      "step": 33100
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.619162058412465e-05,
+      "loss": 0.9045,
+      "step": 33120
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.6138035161535986e-05,
+      "loss": 0.8999,
+      "step": 33140
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.608452148670356e-05,
+      "loss": 0.8836,
+      "step": 33160
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.603107967301378e-05,
+      "loss": 0.9097,
+      "step": 33180
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.597770983370075e-05,
+      "loss": 0.9269,
+      "step": 33200
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5924412081846113e-05,
+      "loss": 0.9053,
+      "step": 33220
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5871186530378763e-05,
+      "loss": 0.8981,
+      "step": 33240
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.581803329207461e-05,
+      "loss": 0.8572,
+      "step": 33260
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5764952479556334e-05,
+      "loss": 0.8934,
+      "step": 33280
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5711944205293185e-05,
+      "loss": 0.8725,
+      "step": 33300
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.565900858160068e-05,
+      "loss": 0.8948,
+      "step": 33320
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5606145720640442e-05,
+      "loss": 0.8983,
+      "step": 33340
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.555335573441989e-05,
+      "loss": 0.9201,
+      "step": 33360
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5500638734792055e-05,
+      "loss": 0.8854,
+      "step": 33380
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5447994833455292e-05,
+      "loss": 0.8975,
+      "step": 33400
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.53954241419531e-05,
+      "loss": 0.8831,
+      "step": 33420
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5342926771673842e-05,
+      "loss": 0.8813,
+      "step": 33440
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5290502833850578e-05,
+      "loss": 0.8986,
+      "step": 33460
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5238152439560693e-05,
+      "loss": 0.8652,
+      "step": 33480
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5185875699725793e-05,
+      "loss": 0.9025,
+      "step": 33500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.5133672725111425e-05,
+      "loss": 0.8958,
+      "step": 33520
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.5081543626326833e-05,
+      "loss": 0.9204,
+      "step": 33540
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.5029488513824724e-05,
+      "loss": 0.8805,
+      "step": 33560
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4977507497901055e-05,
+      "loss": 0.9058,
+      "step": 33580
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.492560068869478e-05,
+      "loss": 0.9166,
+      "step": 33600
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4873768196187593e-05,
+      "loss": 0.9029,
+      "step": 33620
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.482201013020375e-05,
+      "loss": 0.883,
+      "step": 33640
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4770326600409851e-05,
+      "loss": 0.9034,
+      "step": 33660
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.471871771631448e-05,
+      "loss": 0.8594,
+      "step": 33680
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4667183587268118e-05,
+      "loss": 0.9064,
+      "step": 33700
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4615724322462838e-05,
+      "loss": 0.9083,
+      "step": 33720
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4564340030932083e-05,
+      "loss": 0.8579,
+      "step": 33740
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4513030821550449e-05,
+      "loss": 0.8899,
+      "step": 33760
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4461796803033445e-05,
+      "loss": 0.9189,
+      "step": 33780
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4410638083937272e-05,
+      "loss": 0.9012,
+      "step": 33800
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4359554772658552e-05,
+      "loss": 0.9,
+      "step": 33820
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4308546977434135e-05,
+      "loss": 0.957,
+      "step": 33840
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4257614806340919e-05,
+      "loss": 0.8913,
+      "step": 33860
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4206758367295537e-05,
+      "loss": 0.9182,
+      "step": 33880
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4155977768054113e-05,
+      "loss": 0.9013,
+      "step": 33900
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4105273116212136e-05,
+      "loss": 0.9113,
+      "step": 33920
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4054644519204157e-05,
+      "loss": 0.8801,
+      "step": 33940
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4004092084303583e-05,
+      "loss": 0.9287,
+      "step": 33960
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3953615918622443e-05,
+      "loss": 0.9068,
+      "step": 33980
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3903216129111174e-05,
+      "loss": 0.8831,
+      "step": 34000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.385289282255835e-05,
+      "loss": 0.9099,
+      "step": 34020
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.380264610559051e-05,
+      "loss": 0.9004,
+      "step": 34040
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3752476084671962e-05,
+      "loss": 0.8964,
+      "step": 34060
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3702382866104457e-05,
+      "loss": 0.8801,
+      "step": 34080
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3652366556026996e-05,
+      "loss": 0.919,
+      "step": 34100
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3602427260415663e-05,
+      "loss": 0.8887,
+      "step": 34120
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3552565085083352e-05,
+      "loss": 0.9103,
+      "step": 34140
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.350278013567955e-05,
+      "loss": 0.9001,
+      "step": 34160
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3453072517690107e-05,
+      "loss": 0.8922,
+      "step": 34180
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3403442336437039e-05,
+      "loss": 0.8959,
+      "step": 34200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3353889697078287e-05,
+      "loss": 0.8825,
+      "step": 34220
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3304414704607443e-05,
+      "loss": 0.9026,
+      "step": 34240
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3255017463853659e-05,
+      "loss": 0.8999,
+      "step": 34260
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3205698079481298e-05,
+      "loss": 0.8748,
+      "step": 34280
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3156456655989746e-05,
+      "loss": 0.882,
+      "step": 34300
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3107293297713236e-05,
+      "loss": 0.8574,
+      "step": 34320
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3058208108820574e-05,
+      "loss": 0.9052,
+      "step": 34340
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3009201193314947e-05,
+      "loss": 0.9069,
+      "step": 34360
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2960272655033689e-05,
+      "loss": 0.8821,
+      "step": 34380
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2911422597648077e-05,
+      "loss": 0.8953,
+      "step": 34400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2862651124663095e-05,
+      "loss": 0.8736,
+      "step": 34420
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2813958339417176e-05,
+      "loss": 0.9102,
+      "step": 34440
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2765344345082114e-05,
+      "loss": 0.8673,
+      "step": 34460
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2716809244662691e-05,
+      "loss": 0.9082,
+      "step": 34480
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.266835314099657e-05,
+      "loss": 0.8893,
+      "step": 34500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.261997613675398e-05,
+      "loss": 0.8961,
+      "step": 34520
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2571678334437591e-05,
+      "loss": 0.8572,
+      "step": 34540
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.252345983638225e-05,
+      "loss": 0.9225,
+      "step": 34560
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2475320744754776e-05,
+      "loss": 0.9017,
+      "step": 34580
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2427261161553732e-05,
+      "loss": 0.9022,
+      "step": 34600
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2379281188609209e-05,
+      "loss": 0.9,
+      "step": 34620
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2331380927582642e-05,
+      "loss": 0.8776,
+      "step": 34640
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2283560479966538e-05,
+      "loss": 0.8804,
+      "step": 34660
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.223581994708432e-05,
+      "loss": 0.8881,
+      "step": 34680
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2188159430090085e-05,
+      "loss": 0.8949,
+      "step": 34700
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2140579029968352e-05,
+      "loss": 0.8953,
+      "step": 34720
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2093078847533922e-05,
+      "loss": 0.8937,
+      "step": 34740
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2045658983431612e-05,
+      "loss": 0.9329,
+      "step": 34760
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.199831953813611e-05,
+      "loss": 0.8943,
+      "step": 34780
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1951060611951615e-05,
+      "loss": 0.8774,
+      "step": 34800
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1903882305011793e-05,
+      "loss": 0.9075,
+      "step": 34820
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1856784717279462e-05,
+      "loss": 0.8714,
+      "step": 34840
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1809767948546419e-05,
+      "loss": 0.8841,
+      "step": 34860
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1762832098433219e-05,
+      "loss": 0.8719,
+      "step": 34880
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1715977266388961e-05,
+      "loss": 0.8972,
+      "step": 34900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1669203551691093e-05,
+      "loss": 0.8943,
+      "step": 34920
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1622511053445156e-05,
+      "loss": 0.8861,
+      "step": 34940
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1575899870584621e-05,
+      "loss": 0.9284,
+      "step": 34960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1529370101870723e-05,
+      "loss": 0.8943,
+      "step": 34980
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1482921845892098e-05,
+      "loss": 0.8904,
+      "step": 35000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.143655520106473e-05,
+      "loss": 0.8703,
+      "step": 35020
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1390270265631675e-05,
+      "loss": 0.9096,
+      "step": 35040
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.134406713766285e-05,
+      "loss": 0.9049,
+      "step": 35060
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1297945915054842e-05,
+      "loss": 0.8983,
+      "step": 35080
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1251906695530701e-05,
+      "loss": 0.9089,
+      "step": 35100
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1205949576639723e-05,
+      "loss": 0.8768,
+      "step": 35120
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.116007465575722e-05,
+      "loss": 0.9167,
+      "step": 35140
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1114282030084361e-05,
+      "loss": 0.9169,
+      "step": 35160
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1068571796647992e-05,
+      "loss": 0.903,
+      "step": 35180
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.1022944052300293e-05,
+      "loss": 0.8746,
+      "step": 35200
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0977398893718732e-05,
+      "loss": 0.9006,
+      "step": 35220
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0931936417405764e-05,
+      "loss": 0.8895,
+      "step": 35240
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0886556719688662e-05,
+      "loss": 0.8928,
+      "step": 35260
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0841259896719297e-05,
+      "loss": 0.9004,
+      "step": 35280
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0796046044473962e-05,
+      "loss": 0.9078,
+      "step": 35300
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0750915258753141e-05,
+      "loss": 0.8804,
+      "step": 35320
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0705867635181278e-05,
+      "loss": 0.902,
+      "step": 35340
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0660903269206652e-05,
+      "loss": 0.898,
+      "step": 35360
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0616022256101143e-05,
+      "loss": 0.8605,
+      "step": 35380
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0571224690960002e-05,
+      "loss": 0.8795,
+      "step": 35400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0526510668701633e-05,
+      "loss": 0.8985,
+      "step": 35420
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0481880284067485e-05,
+      "loss": 0.9146,
+      "step": 35440
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0437333631621765e-05,
+      "loss": 0.8778,
+      "step": 35460
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0392870805751265e-05,
+      "loss": 0.9129,
+      "step": 35480
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0348491900665164e-05,
+      "loss": 0.9134,
+      "step": 35500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.030419701039484e-05,
+      "loss": 0.9043,
+      "step": 35520
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.025998622879365e-05,
+      "loss": 0.8896,
+      "step": 35540
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0215859649536702e-05,
+      "loss": 0.9067,
+      "step": 35560
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0171817366120767e-05,
+      "loss": 0.9166,
+      "step": 35580
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.012785947186397e-05,
+      "loss": 0.8699,
+      "step": 35600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0083986059905598e-05,
+      "loss": 0.8923,
+      "step": 35620
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0040197223205978e-05,
+      "loss": 0.8432,
+      "step": 35640
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.996493054546214e-06,
+      "loss": 0.8849,
+      "step": 35660
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.95287364652806e-06,
+      "loss": 0.8686,
+      "step": 35680
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.909339091573594e-06,
+      "loss": 0.8646,
+      "step": 35700
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.865889481925167e-06,
+      "loss": 0.8976,
+      "step": 35720
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.822524909645137e-06,
+      "loss": 0.8732,
+      "step": 35740
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.779245466615639e-06,
+      "loss": 0.8954,
+      "step": 35760
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.736051244538497e-06,
+      "loss": 0.912,
+      "step": 35780
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.692942334934935e-06,
+      "loss": 0.891,
+      "step": 35800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.649918829145415e-06,
+      "loss": 0.8954,
+      "step": 35820
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.60698081832943e-06,
+      "loss": 0.9018,
+      "step": 35840
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.564128393465332e-06,
+      "loss": 0.8627,
+      "step": 35860
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.52136164535018e-06,
+      "loss": 0.9076,
+      "step": 35880
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.478680664599404e-06,
+      "loss": 0.93,
+      "step": 35900
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.436085541646783e-06,
+      "loss": 0.8731,
+      "step": 35920
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.393576366744162e-06,
+      "loss": 0.8885,
+      "step": 35940
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.351153229961223e-06,
+      "loss": 0.8715,
+      "step": 35960
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.308816221185441e-06,
+      "loss": 0.8789,
+      "step": 35980
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.266565430121733e-06,
+      "loss": 0.886,
+      "step": 36000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.224400946292367e-06,
+      "loss": 0.8862,
+      "step": 36020
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.182322859036702e-06,
+      "loss": 0.9107,
+      "step": 36040
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.14033125751107e-06,
+      "loss": 0.877,
+      "step": 36060
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.098426230688578e-06,
+      "loss": 0.8937,
+      "step": 36080
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.056607867358829e-06,
+      "loss": 0.8663,
+      "step": 36100
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.014876256127852e-06,
+      "loss": 0.8781,
+      "step": 36120
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.973231485417849e-06,
+      "loss": 0.866,
+      "step": 36140
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.931673643467014e-06,
+      "loss": 0.8693,
+      "step": 36160
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.890202818329368e-06,
+      "loss": 0.8629,
+      "step": 36180
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.84881909787455e-06,
+      "loss": 0.8731,
+      "step": 36200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.807522569787653e-06,
+      "loss": 0.8921,
+      "step": 36220
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.76631332156898e-06,
+      "loss": 0.8864,
+      "step": 36240
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.725191440533936e-06,
+      "loss": 0.8866,
+      "step": 36260
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.684157013812839e-06,
+      "loss": 0.8669,
+      "step": 36280
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.643210128350665e-06,
+      "loss": 0.8555,
+      "step": 36300
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.602350870906895e-06,
+      "loss": 0.8809,
+      "step": 36320
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.561579328055375e-06,
+      "loss": 0.865,
+      "step": 36340
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.52089558618408e-06,
+      "loss": 0.888,
+      "step": 36360
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.480299731494956e-06,
+      "loss": 0.8966,
+      "step": 36380
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.439791850003726e-06,
+      "loss": 0.8947,
+      "step": 36400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.39937202753972e-06,
+      "loss": 0.8662,
+      "step": 36420
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.35904034974569e-06,
+      "loss": 0.8828,
+      "step": 36440
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.31879690207758e-06,
+      "loss": 0.8679,
+      "step": 36460
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.278641769804469e-06,
+      "loss": 0.8961,
+      "step": 36480
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.23857503800825e-06,
+      "loss": 0.8836,
+      "step": 36500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.198596791583523e-06,
+      "loss": 0.8672,
+      "step": 36520
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.158707115237407e-06,
+      "loss": 0.8916,
+      "step": 36540
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.118906093489358e-06,
+      "loss": 0.9051,
+      "step": 36560
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.079193810670988e-06,
+      "loss": 0.8754,
+      "step": 36580
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.039570350925878e-06,
+      "loss": 0.895,
+      "step": 36600
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.000035798209421e-06,
+      "loss": 0.8794,
+      "step": 36620
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.960590236288633e-06,
+      "loss": 0.8761,
+      "step": 36640
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.921233748741934e-06,
+      "loss": 0.8677,
+      "step": 36660
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.88196641895907e-06,
+      "loss": 0.8771,
+      "step": 36680
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.842788330140838e-06,
+      "loss": 0.8851,
+      "step": 36700
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.803699565298972e-06,
+      "loss": 0.9103,
+      "step": 36720
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.764700207255903e-06,
+      "loss": 0.8621,
+      "step": 36740
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.725790338644673e-06,
+      "loss": 0.8558,
+      "step": 36760
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.686970041908675e-06,
+      "loss": 0.8793,
+      "step": 36780
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.648239399301544e-06,
+      "loss": 0.9105,
+      "step": 36800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.6095984928869265e-06,
+      "loss": 0.879,
+      "step": 36820
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.571047404538351e-06,
+      "loss": 0.8657,
+      "step": 36840
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.532586215939025e-06,
+      "loss": 0.8624,
+      "step": 36860
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.49421500858169e-06,
+      "loss": 0.8568,
+      "step": 36880
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.45593386376841e-06,
+      "loss": 0.8805,
+      "step": 36900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.41774286261045e-06,
+      "loss": 0.8731,
+      "step": 36920
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.379642086028038e-06,
+      "loss": 0.9025,
+      "step": 36940
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.341631614750266e-06,
+      "loss": 0.867,
+      "step": 36960
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.303711529314861e-06,
+      "loss": 0.877,
+      "step": 36980
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.265881910068062e-06,
+      "loss": 0.8611,
+      "step": 37000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.228142837164404e-06,
+      "loss": 0.8895,
+      "step": 37020
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.190494390566571e-06,
+      "loss": 0.9216,
+      "step": 37040
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.152936650045245e-06,
+      "loss": 0.8817,
+      "step": 37060
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.115469695178895e-06,
+      "loss": 0.8688,
+      "step": 37080
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.078093605353642e-06,
+      "loss": 0.8903,
+      "step": 37100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.040808459763082e-06,
+      "loss": 0.8687,
+      "step": 37120
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.003614337408099e-06,
+      "loss": 0.8761,
+      "step": 37140
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.96651131709673e-06,
+      "loss": 0.88,
+      "step": 37160
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.929499477443962e-06,
+      "loss": 0.8919,
+      "step": 37180
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.892578896871643e-06,
+      "loss": 0.9064,
+      "step": 37200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.855749653608179e-06,
+      "loss": 0.8838,
+      "step": 37220
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.819011825688498e-06,
+      "loss": 0.8945,
+      "step": 37240
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.782365490953824e-06,
+      "loss": 0.8609,
+      "step": 37260
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.745810727051521e-06,
+      "loss": 0.8978,
+      "step": 37280
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.709347611434924e-06,
+      "loss": 0.8814,
+      "step": 37300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.672976221363186e-06,
+      "loss": 0.8896,
+      "step": 37320
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.636696633901124e-06,
+      "loss": 0.9108,
+      "step": 37340
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.600508925919008e-06,
+      "loss": 0.9018,
+      "step": 37360
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.564413174092443e-06,
+      "loss": 0.9047,
+      "step": 37380
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.528409454902235e-06,
+      "loss": 0.8608,
+      "step": 37400
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.492497844634121e-06,
+      "loss": 0.8941,
+      "step": 37420
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.4566784193787255e-06,
+      "loss": 0.8743,
+      "step": 37440
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.4209512550313215e-06,
+      "loss": 0.8991,
+      "step": 37460
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.38531642729171e-06,
+      "loss": 0.8968,
+      "step": 37480
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.3497740116640396e-06,
+      "loss": 0.8719,
+      "step": 37500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.314324083456663e-06,
+      "loss": 0.9034,
+      "step": 37520
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.2789667177819755e-06,
+      "loss": 0.8603,
+      "step": 37540
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.2437019895561995e-06,
+      "loss": 0.9047,
+      "step": 37560
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.208529973499316e-06,
+      "loss": 0.9057,
+      "step": 37580
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.1734507441348785e-06,
+      "loss": 0.8938,
+      "step": 37600
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.138464375789821e-06,
+      "loss": 0.8755,
+      "step": 37620
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.1035709425943e-06,
+      "loss": 0.8896,
+      "step": 37640
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.068770518481582e-06,
+      "loss": 0.8586,
+      "step": 37660
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.034063177187865e-06,
+      "loss": 0.8803,
+      "step": 37680
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.9994489922521155e-06,
+      "loss": 0.9121,
+      "step": 37700
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.96492803701591e-06,
+      "loss": 0.8782,
+      "step": 37720
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.9305003846233e-06,
+      "loss": 0.9016,
+      "step": 37740
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.8961661080206454e-06,
+      "loss": 0.8687,
+      "step": 37760
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.861925279956415e-06,
+      "loss": 0.885,
+      "step": 37780
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.827777972981152e-06,
+      "loss": 0.853,
+      "step": 37800
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.793724259447203e-06,
+      "loss": 0.8716,
+      "step": 37820
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.759764211508578e-06,
+      "loss": 0.868,
+      "step": 37840
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.7258979011208746e-06,
+      "loss": 0.8453,
+      "step": 37860
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.692125400041049e-06,
+      "loss": 0.8788,
+      "step": 37880
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.658446779827309e-06,
+      "loss": 0.8741,
+      "step": 37900
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.624862111838919e-06,
+      "loss": 0.8998,
+      "step": 37920
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.5913714672361065e-06,
+      "loss": 0.8717,
+      "step": 37940
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.557974916979863e-06,
+      "loss": 0.8747,
+      "step": 37960
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.5246725318317815e-06,
+      "loss": 0.8667,
+      "step": 37980
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.491464382354e-06,
+      "loss": 0.8788,
+      "step": 38000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.458350538908946e-06,
+      "loss": 0.8948,
+      "step": 38020
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.425331071659212e-06,
+      "loss": 0.8609,
+      "step": 38040
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.392406050567455e-06,
+      "loss": 0.8659,
+      "step": 38060
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.3595755453962115e-06,
+      "loss": 0.8705,
+      "step": 38080
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.3268396257077465e-06,
+      "loss": 0.8849,
+      "step": 38100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.294198360863917e-06,
+      "loss": 0.8659,
+      "step": 38120
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.26165182002602e-06,
+      "loss": 0.8674,
+      "step": 38140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.229200072154672e-06,
+      "loss": 0.8786,
+      "step": 38160
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.196843186009581e-06,
+      "loss": 0.8893,
+      "step": 38180
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.164581230149529e-06,
+      "loss": 0.899,
+      "step": 38200
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.132414272932107e-06,
+      "loss": 0.8991,
+      "step": 38220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.100342382513662e-06,
+      "loss": 0.867,
+      "step": 38240
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.068365626849058e-06,
+      "loss": 0.8965,
+      "step": 38260
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.036484073691622e-06,
+      "loss": 0.9204,
+      "step": 38280
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.004697790592961e-06,
+      "loss": 0.9037,
+      "step": 38300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.9730068449028e-06,
+      "loss": 0.8833,
+      "step": 38320
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.941411303768889e-06,
+      "loss": 0.88,
+      "step": 38340
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.90991123413681e-06,
+      "loss": 0.8873,
+      "step": 38360
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.878506702749869e-06,
+      "loss": 0.8802,
+      "step": 38380
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.847197776148932e-06,
+      "loss": 0.8771,
+      "step": 38400
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.815984520672301e-06,
+      "loss": 0.8883,
+      "step": 38420
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.784867002455584e-06,
+      "loss": 0.8629,
+      "step": 38440
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.753845287431491e-06,
+      "loss": 0.8824,
+      "step": 38460
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.722919441329782e-06,
+      "loss": 0.882,
+      "step": 38480
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.692089529677074e-06,
+      "loss": 0.8704,
+      "step": 38500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.661355617796742e-06,
+      "loss": 0.8956,
+      "step": 38520
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.630717770808696e-06,
+      "loss": 0.8867,
+      "step": 38540
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.600176053629346e-06,
+      "loss": 0.8825,
+      "step": 38560
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.569730530971411e-06,
+      "loss": 0.8755,
+      "step": 38580
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.5393812673437844e-06,
+      "loss": 0.8778,
+      "step": 38600
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.5091283270513985e-06,
+      "loss": 0.8921,
+      "step": 38620
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.4789717741951065e-06,
+      "loss": 0.8814,
+      "step": 38640
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.448911672671535e-06,
+      "loss": 0.8655,
+      "step": 38660
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.418948086172914e-06,
+      "loss": 0.9148,
+      "step": 38680
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.389081078186996e-06,
+      "loss": 0.8551,
+      "step": 38700
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.359310711996939e-06,
+      "loss": 0.8897,
+      "step": 38720
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.329637050681057e-06,
+      "loss": 0.859,
+      "step": 38740
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.300060157112817e-06,
+      "loss": 0.8528,
+      "step": 38760
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.270580093960641e-06,
+      "loss": 0.8661,
+      "step": 38780
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.241196923687774e-06,
+      "loss": 0.8779,
+      "step": 38800
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2119107085521636e-06,
+      "loss": 0.8813,
+      "step": 38820
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.18272151060633e-06,
+      "loss": 0.865,
+      "step": 38840
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.153629391697244e-06,
+      "loss": 0.8753,
+      "step": 38860
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.12463441346615e-06,
+      "loss": 0.9042,
+      "step": 38880
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.09573663734848e-06,
+      "loss": 0.8962,
+      "step": 38900
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.066936124573734e-06,
+      "loss": 0.8484,
+      "step": 38920
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.03823293616532e-06,
+      "loss": 0.8848,
+      "step": 38940
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.009627132940397e-06,
+      "loss": 0.8626,
+      "step": 38960
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.981118775509812e-06,
+      "loss": 0.8792,
+      "step": 38980
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.952707924277949e-06,
+      "loss": 0.8841,
+      "step": 39000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.9243946394425635e-06,
+      "loss": 0.8447,
+      "step": 39020
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.896178980994714e-06,
+      "loss": 0.8927,
+      "step": 39040
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.868061008718593e-06,
+      "loss": 0.8913,
+      "step": 39060
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.840040782191401e-06,
+      "loss": 0.889,
+      "step": 39080
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.8121183607832344e-06,
+      "loss": 0.869,
+      "step": 39100
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.7842938036569854e-06,
+      "loss": 0.9043,
+      "step": 39120
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.756567169768166e-06,
+      "loss": 0.8772,
+      "step": 39140
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.728938517864794e-06,
+      "loss": 0.8899,
+      "step": 39160
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.701407906487303e-06,
+      "loss": 0.8797,
+      "step": 39180
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.6739753939683817e-06,
+      "loss": 0.8753,
+      "step": 39200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.6466410384328685e-06,
+      "loss": 0.8659,
+      "step": 39220
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.6194048977976237e-06,
+      "loss": 0.875,
+      "step": 39240
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5922670297714124e-06,
+      "loss": 0.8816,
+      "step": 39260
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5652274918547724e-06,
+      "loss": 0.8792,
+      "step": 39280
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.5382863413398694e-06,
+      "loss": 0.872,
+      "step": 39300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.5114436353104574e-06,
+      "loss": 0.8681,
+      "step": 39320
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4846994306416746e-06,
+      "loss": 0.855,
+      "step": 39340
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4580537839999374e-06,
+      "loss": 0.8619,
+      "step": 39360
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.43150675184285e-06,
+      "loss": 0.8799,
+      "step": 39380
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.405058390419069e-06,
+      "loss": 0.8626,
+      "step": 39400
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3787087557681895e-06,
+      "loss": 0.8733,
+      "step": 39420
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.352457903720613e-06,
+      "loss": 0.8642,
+      "step": 39440
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.326305889897435e-06,
+      "loss": 0.8841,
+      "step": 39460
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3002527697103435e-06,
+      "loss": 0.8708,
+      "step": 39480
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.274298598361458e-06,
+      "loss": 0.896,
+      "step": 39500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.2484434308432843e-06,
+      "loss": 0.8654,
+      "step": 39520
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.2226873219385224e-06,
+      "loss": 0.8616,
+      "step": 39540
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.197030326220013e-06,
+      "loss": 0.8735,
+      "step": 39560
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.1714724980505484e-06,
+      "loss": 0.8782,
+      "step": 39580
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.146013891582844e-06,
+      "loss": 0.8788,
+      "step": 39600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.1206545607593616e-06,
+      "loss": 0.8564,
+      "step": 39620
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.095394559312226e-06,
+      "loss": 0.8777,
+      "step": 39640
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.070233940763084e-06,
+      "loss": 0.8988,
+      "step": 39660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.0451727584230207e-06,
+      "loss": 0.871,
+      "step": 39680
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.020211065392431e-06,
+      "loss": 0.885,
+      "step": 39700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.995348914560897e-06,
+      "loss": 0.8604,
+      "step": 39720
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.9705863586071027e-06,
+      "loss": 0.8535,
+      "step": 39740
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.945923449998711e-06,
+      "loss": 0.9172,
+      "step": 39760
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.921360240992216e-06,
+      "loss": 0.8873,
+      "step": 39780
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.8968967836329077e-06,
+      "loss": 0.8798,
+      "step": 39800
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.8725331297546786e-06,
+      "loss": 0.8649,
+      "step": 39820
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.8482693309800136e-06,
+      "loss": 0.8986,
+      "step": 39840
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.8241054387197487e-06,
+      "loss": 0.92,
+      "step": 39860
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.8000415041730845e-06,
+      "loss": 0.8747,
+      "step": 39880
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7760775783274127e-06,
+      "loss": 0.8853,
+      "step": 39900
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7522137119582238e-06,
+      "loss": 0.8646,
+      "step": 39920
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7284499556290033e-06,
+      "loss": 0.8972,
+      "step": 39940
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7047863596911248e-06,
+      "loss": 0.8619,
+      "step": 39960
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.681222974283715e-06,
+      "loss": 0.8512,
+      "step": 39980
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.657759849333591e-06,
+      "loss": 0.868,
+      "step": 40000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.6343970345551363e-06,
+      "loss": 0.8721,
+      "step": 40020
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.6111345794502027e-06,
+      "loss": 0.8967,
+      "step": 40040
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.587972533307964e-06,
+      "loss": 0.8751,
+      "step": 40060
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.5649109452048735e-06,
+      "loss": 0.8797,
+      "step": 40080
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.541949864004528e-06,
+      "loss": 0.8924,
+      "step": 40100
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.5190893383575498e-06,
+      "loss": 0.855,
+      "step": 40120
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.4963294167015204e-06,
+      "loss": 0.8679,
+      "step": 40140
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.473670147260848e-06,
+      "loss": 0.8751,
+      "step": 40160
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.451111578046689e-06,
+      "loss": 0.8775,
+      "step": 40180
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.4286537568568023e-06,
+      "loss": 0.8797,
+      "step": 40200
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.4062967312755037e-06,
+      "loss": 0.8527,
+      "step": 40220
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.3840405486735607e-06,
+      "loss": 0.8736,
+      "step": 40240
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.3618852562080187e-06,
+      "loss": 0.8782,
+      "step": 40260
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.339830900822193e-06,
+      "loss": 0.9045,
+      "step": 40280
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.3178775292455214e-06,
+      "loss": 0.8947,
+      "step": 40300
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.296025187993467e-06,
+      "loss": 0.8586,
+      "step": 40320
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.274273923367437e-06,
+      "loss": 0.8814,
+      "step": 40340
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.25262378145466e-06,
+      "loss": 0.8635,
+      "step": 40360
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.231074808128131e-06,
+      "loss": 0.8544,
+      "step": 40380
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.2096270490464476e-06,
+      "loss": 0.8556,
+      "step": 40400
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.1882805496537705e-06,
+      "loss": 0.9008,
+      "step": 40420
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.167035355179725e-06,
+      "loss": 0.8816,
+      "step": 40440
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.1458915106392697e-06,
+      "loss": 0.8555,
+      "step": 40460
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.1248490608326123e-06,
+      "loss": 0.8593,
+      "step": 40480
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.1039080503451325e-06,
+      "loss": 0.8759,
+      "step": 40500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.083068523547288e-06,
+      "loss": 0.8855,
+      "step": 40520
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.0623305245944913e-06,
+      "loss": 0.8577,
+      "step": 40540
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.0416940974270384e-06,
+      "loss": 0.8959,
+      "step": 40560
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.0211592857700303e-06,
+      "loss": 0.8824,
+      "step": 40580
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.0007261331332404e-06,
+      "loss": 0.9085,
+      "step": 40600
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.9803946828110375e-06,
+      "loss": 0.8982,
+      "step": 40620
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.9601649778823337e-06,
+      "loss": 0.863,
+      "step": 40640
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.940037061210437e-06,
+      "loss": 0.8765,
+      "step": 40660
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.920010975442976e-06,
+      "loss": 0.8977,
+      "step": 40680
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.9000867630118324e-06,
+      "loss": 0.8776,
+      "step": 40700
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8802644661330304e-06,
+      "loss": 0.862,
+      "step": 40720
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8605441268066625e-06,
+      "loss": 0.8683,
+      "step": 40740
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8409257868167718e-06,
+      "loss": 0.8883,
+      "step": 40760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8214094877312849e-06,
+      "loss": 0.8897,
+      "step": 40780
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8019952709019404e-06,
+      "loss": 0.8771,
+      "step": 40800
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7826831774641617e-06,
+      "loss": 0.8785,
+      "step": 40820
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7634732483369943e-06,
+      "loss": 0.8497,
+      "step": 40840
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7443655242230238e-06,
+      "loss": 0.8961,
+      "step": 40860
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7253600456082764e-06,
+      "loss": 0.8995,
+      "step": 40880
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.7064568527621228e-06,
+      "loss": 0.8858,
+      "step": 40900
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6876559857372243e-06,
+      "loss": 0.8618,
+      "step": 40920
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6689574843694433e-06,
+      "loss": 0.8773,
+      "step": 40940
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6503613882777101e-06,
+      "loss": 0.8614,
+      "step": 40960
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6318677368640067e-06,
+      "loss": 0.9059,
+      "step": 40980
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6134765693132382e-06,
+      "loss": 0.9023,
+      "step": 41000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5951879245931723e-06,
+      "loss": 0.8583,
+      "step": 41020
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5770018414543342e-06,
+      "loss": 0.8845,
+      "step": 41040
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5589183584299504e-06,
+      "loss": 0.8818,
+      "step": 41060
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5409375138358663e-06,
+      "loss": 0.8754,
+      "step": 41080
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5230593457704168e-06,
+      "loss": 0.8513,
+      "step": 41100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.505283892114412e-06,
+      "loss": 0.8949,
+      "step": 41120
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.4876111905310408e-06,
+      "loss": 0.8962,
+      "step": 41140
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.4700412784657336e-06,
+      "loss": 0.8671,
+      "step": 41160
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4525741931461612e-06,
+      "loss": 0.8822,
+      "step": 41180
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4352099715821133e-06,
+      "loss": 0.8698,
+      "step": 41200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4179486505654316e-06,
+      "loss": 0.8912,
+      "step": 41220
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4007902666699157e-06,
+      "loss": 0.9107,
+      "step": 41240
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.3837348562512842e-06,
+      "loss": 0.8674,
+      "step": 41260
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.3667824554470466e-06,
+      "loss": 0.8724,
+      "step": 41280
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.3499331001764592e-06,
+      "loss": 0.8747,
+      "step": 41300
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.3331868261404479e-06,
+      "loss": 0.8595,
+      "step": 41320
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.316543668821535e-06,
+      "loss": 0.8836,
+      "step": 41340
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3000036634837398e-06,
+      "loss": 0.8632,
+      "step": 41360
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2835668451725292e-06,
+      "loss": 0.8968,
+      "step": 41380
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.267233248714722e-06,
+      "loss": 0.8771,
+      "step": 41400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.251002908718446e-06,
+      "loss": 0.8829,
+      "step": 41420
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2348758595730314e-06,
+      "loss": 0.8628,
+      "step": 41440
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2188521354489613e-06,
+      "loss": 0.862,
+      "step": 41460
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2029317702977882e-06,
+      "loss": 0.8919,
+      "step": 41480
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.187114797852068e-06,
+      "loss": 0.8733,
+      "step": 41500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1714012516252648e-06,
+      "loss": 0.8723,
+      "step": 41520
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1557911649117293e-06,
+      "loss": 0.8721,
+      "step": 41540
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1402845707865928e-06,
+      "loss": 0.8917,
+      "step": 41560
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.124881502105679e-06,
+      "loss": 0.871,
+      "step": 41580
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1095819915054872e-06,
+      "loss": 0.8686,
+      "step": 41600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.094386071403075e-06,
+      "loss": 0.8621,
+      "step": 41620
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0792937739960262e-06,
+      "loss": 0.8884,
+      "step": 41640
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0643051312623553e-06,
+      "loss": 0.8675,
+      "step": 41660
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0494201749604525e-06,
+      "loss": 0.8882,
+      "step": 41680
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0346389366290122e-06,
+      "loss": 0.8953,
+      "step": 41700
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0199614475869646e-06,
+      "loss": 0.8781,
+      "step": 41720
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0053877389334277e-06,
+      "loss": 0.8719,
+      "step": 41740
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.909178415476116e-07,
+      "loss": 0.8847,
+      "step": 41760
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.765517860887808e-07,
+      "loss": 0.8803,
+      "step": 41780
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.62289602996158e-07,
+      "loss": 0.8687,
+      "step": 41800
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.481313224888877e-07,
+      "loss": 0.8566,
+      "step": 41820
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.340769745659672e-07,
+      "loss": 0.9047,
+      "step": 41840
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.201265890061816e-07,
+      "loss": 0.8534,
+      "step": 41860
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.062801953680145e-07,
+      "loss": 0.9076,
+      "step": 41880
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.925378229896364e-07,
+      "loss": 0.9027,
+      "step": 41900
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.788995009888002e-07,
+      "loss": 0.8772,
+      "step": 41920
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.653652582627958e-07,
+      "loss": 0.8791,
+      "step": 41940
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.519351234883787e-07,
+      "loss": 0.8871,
+      "step": 41960
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.386091251217365e-07,
+      "loss": 0.8808,
+      "step": 41980
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.253872913983884e-07,
+      "loss": 0.9046,
+      "step": 42000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.122696503331583e-07,
+      "loss": 0.9211,
+      "step": 42020
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.992562297201023e-07,
+      "loss": 0.8927,
+      "step": 42040
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.863470571324527e-07,
+      "loss": 0.8639,
+      "step": 42060
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.735421599225467e-07,
+      "loss": 0.9165,
+      "step": 42080
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.608415652217982e-07,
+      "loss": 0.8829,
+      "step": 42100
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.482452999406087e-07,
+      "loss": 0.8938,
+      "step": 42120
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.357533907683234e-07,
+      "loss": 0.8722,
+      "step": 42140
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.23365864173181e-07,
+      "loss": 0.8738,
+      "step": 42160
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.110827464022474e-07,
+      "loss": 0.8595,
+      "step": 42180
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.989040634813648e-07,
+      "loss": 0.8597,
+      "step": 42200
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.868298412150864e-07,
+      "loss": 0.9116,
+      "step": 42220
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.748601051866532e-07,
+      "loss": 0.8424,
+      "step": 42240
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.629948807579001e-07,
+      "loss": 0.8709,
+      "step": 42260
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.512341930692167e-07,
+      "loss": 0.9012,
+      "step": 42280
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.395780670395147e-07,
+      "loss": 0.869,
+      "step": 42300
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.280265273661379e-07,
+      "loss": 0.8905,
+      "step": 42320
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.165795985248413e-07,
+      "loss": 0.8933,
+      "step": 42340
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.052373047697236e-07,
+      "loss": 0.8983,
+      "step": 42360
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.939996701331884e-07,
+      "loss": 0.9021,
+      "step": 42380
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.828667184258673e-07,
+      "loss": 0.889,
+      "step": 42400
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.718384732365967e-07,
+      "loss": 0.8627,
+      "step": 42420
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.609149579323513e-07,
+      "loss": 0.8778,
+      "step": 42440
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.50096195658223e-07,
+      "loss": 0.8989,
+      "step": 42460
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.393822093373135e-07,
+      "loss": 0.8895,
+      "step": 42480
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.287730216707532e-07,
+      "loss": 0.8777,
+      "step": 42500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.182686551376048e-07,
+      "loss": 0.895,
+      "step": 42520
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.07869131994837e-07,
+      "loss": 0.8736,
+      "step": 42540
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.975744742772848e-07,
+      "loss": 0.8999,
+      "step": 42560
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.873847037975665e-07,
+      "loss": 0.8983,
+      "step": 42580
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.772998421460895e-07,
+      "loss": 0.8648,
+      "step": 42600
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.6731991069094984e-07,
+      "loss": 0.8712,
+      "step": 42620
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.574449305779327e-07,
+      "loss": 0.8866,
+      "step": 42640
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.4767492273045665e-07,
+      "loss": 0.9028,
+      "step": 42660
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.380099078495015e-07,
+      "loss": 0.88,
+      "step": 42680
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.284499064135916e-07,
+      "loss": 0.8929,
+      "step": 42700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.189949386787462e-07,
+      "loss": 0.8876,
+      "step": 42720
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.0964502467844e-07,
+      "loss": 0.9048,
+      "step": 42740
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.0040018422355385e-07,
+      "loss": 0.8726,
+      "step": 42760
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.9126043690234093e-07,
+      "loss": 0.8974,
+      "step": 42780
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.82225802080366e-07,
+      "loss": 0.8804,
+      "step": 42800
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.7329629890048846e-07,
+      "loss": 0.8947,
+      "step": 42820
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.6447194628281276e-07,
+      "loss": 0.8923,
+      "step": 42840
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.557527629246438e-07,
+      "loss": 0.8837,
+      "step": 42860
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.471387673004534e-07,
+      "loss": 0.8695,
+      "step": 42880
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.3862997766182515e-07,
+      "loss": 0.8827,
+      "step": 42900
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.302264120374543e-07,
+      "loss": 0.8632,
+      "step": 42920
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.219280882330644e-07,
+      "loss": 0.8822,
+      "step": 42940
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.137350238313963e-07,
+      "loss": 0.9121,
+      "step": 42960
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.0564723619215807e-07,
+      "loss": 0.9014,
+      "step": 42980
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.9766474245200847e-07,
+      "loss": 0.8743,
+      "step": 43000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.8978755952448475e-07,
+      "loss": 0.8561,
+      "step": 43020
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.8201570410000824e-07,
+      "loss": 0.8845,
+      "step": 43040
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.74349192645823e-07,
+      "loss": 0.8676,
+      "step": 43060
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6678804140596315e-07,
+      "loss": 0.8449,
+      "step": 43080
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5933226640121875e-07,
+      "loss": 0.8725,
+      "step": 43100
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5198188342912543e-07,
+      "loss": 0.876,
+      "step": 43120
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.447369080638806e-07,
+      "loss": 0.8781,
+      "step": 43140
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.3759735565637155e-07,
+      "loss": 0.8812,
+      "step": 43160
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.305632413340919e-07,
+      "loss": 0.8981,
+      "step": 43180
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.236345800011308e-07,
+      "loss": 0.886,
+      "step": 43200
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.1681138633816156e-07,
+      "loss": 0.8873,
+      "step": 43220
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.1009367480235298e-07,
+      "loss": 0.8964,
+      "step": 43240
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.0348145962740817e-07,
+      "loss": 0.8663,
+      "step": 43260
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.9697475482349238e-07,
+      "loss": 0.868,
+      "step": 43280
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.9057357417719968e-07,
+      "loss": 0.881,
+      "step": 43300
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8427793125154747e-07,
+      "loss": 0.8902,
+      "step": 43320
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7808783938593198e-07,
+      "loss": 0.8796,
+      "step": 43340
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7200331169611727e-07,
+      "loss": 0.8813,
+      "step": 43360
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6602436107417408e-07,
+      "loss": 0.8988,
+      "step": 43380
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6015100018849095e-07,
+      "loss": 0.8534,
+      "step": 43400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5438324148371875e-07,
+      "loss": 0.8964,
+      "step": 43420
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.487210971807651e-07,
+      "loss": 0.8934,
+      "step": 43440
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.4316457927674444e-07,
+      "loss": 0.8765,
+      "step": 43460
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.3771369954497793e-07,
+      "loss": 0.8458,
+      "step": 43480
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.323684695349603e-07,
+      "loss": 0.8656,
+      "step": 43500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2712890057232085e-07,
+      "loss": 0.8792,
+      "step": 43520
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2199500375881247e-07,
+      "loss": 0.8566,
+      "step": 43540
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.1696678997230038e-07,
+      "loss": 0.8954,
+      "step": 43560
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.1204426986671235e-07,
+      "loss": 0.8871,
+      "step": 43580
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0722745387203859e-07,
+      "loss": 0.875,
+      "step": 43600
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.025163521942818e-07,
+      "loss": 0.8808,
+      "step": 43620
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.791097481547939e-08,
+      "loss": 0.8818,
+      "step": 43640
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.34113314936369e-08,
+      "loss": 0.884,
+      "step": 43660
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.901743176273902e-08,
+      "loss": 0.8735,
+      "step": 43680
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.472928493271082e-08,
+      "loss": 0.883,
+      "step": 43700
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.054690008940657e-08,
+      "loss": 0.8736,
+      "step": 43720
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.647028609459317e-08,
+      "loss": 0.8778,
+      "step": 43740
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.249945158592231e-08,
+      "loss": 0.8444,
+      "step": 43760
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.863440497691942e-08,
+      "loss": 0.8991,
+      "step": 43780
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.487515445696146e-08,
+      "loss": 0.8613,
+      "step": 43800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.122170799126581e-08,
+      "loss": 0.8667,
+      "step": 43820
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.7674073320856945e-08,
+      "loss": 0.886,
+      "step": 43840
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.423225796257758e-08,
+      "loss": 0.9095,
+      "step": 43860
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.089626920904422e-08,
+      "loss": 0.8674,
+      "step": 43880
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.766611412865829e-08,
+      "loss": 0.9072,
+      "step": 43900
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.4541799565567255e-08,
+      "loss": 0.8989,
+      "step": 43920
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.1523332139664636e-08,
+      "loss": 0.885,
+      "step": 43940
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.861071824656226e-08,
+      "loss": 0.8705,
+      "step": 43960
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.5803964057606885e-08,
+      "loss": 0.8665,
+      "step": 43980
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.310307551983027e-08,
+      "loss": 0.8746,
+      "step": 44000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.050805835594917e-08,
+      "loss": 0.8846,
+      "step": 44020
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.8018918064376398e-08,
+      "loss": 0.8902,
+      "step": 44040
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.563565991916539e-08,
+      "loss": 0.8422,
+      "step": 44060
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.335828897004344e-08,
+      "loss": 0.8935,
+      "step": 44080
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1186810042372885e-08,
+      "loss": 0.8708,
+      "step": 44100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.912122773715108e-08,
+      "loss": 0.8663,
+      "step": 44120
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.7161546430988217e-08,
+      "loss": 0.8972,
+      "step": 44140
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5307770276123956e-08,
+      "loss": 0.8774,
+      "step": 44160
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.3559903200394131e-08,
+      "loss": 0.9105,
+      "step": 44180
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.1917948907225196e-08,
+      "loss": 0.8849,
+      "step": 44200
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.038191087565088e-08,
+      "loss": 0.8716,
+      "step": 44220
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.951792360267775e-09,
+      "loss": 0.8811,
+      "step": 44240
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.62759639125199e-09,
+      "loss": 0.8778,
+      "step": 44260
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.409325774359154e-09,
+      "loss": 0.8776,
+      "step": 44280
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.296983090891106e-09,
+      "loss": 0.8683,
+      "step": 44300
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.2905706977181e-09,
+      "loss": 0.8893,
+      "step": 44320
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3900907272510497e-09,
+      "loss": 0.8707,
+      "step": 44340
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5955450874581845e-09,
+      "loss": 0.879,
+      "step": 44360
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.90693546184284e-09,
+      "loss": 0.857,
+      "step": 44380
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3242633094545654e-09,
+      "loss": 0.8935,
+      "step": 44400
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.475298648835672e-10,
+      "loss": 0.9124,
+      "step": 44420
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.767361382329583e-10,
+      "loss": 0.9007,
+      "step": 44440
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.1188291516316404e-10,
+      "loss": 0.8689,
+      "step": 44460
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.297075684751463e-11,
+      "loss": 0.8644,
+      "step": 44480
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.8527,
+      "step": 44500
+    },
+    {
+      "epoch": 5.0,
+      "step": 44500,
+      "total_flos": 1.4745575028791706e+19,
+      "train_loss": 0.9734652058462079,
+      "train_runtime": 41271.4588,
+      "train_samples_per_second": 17.251,
+      "train_steps_per_second": 1.078
     }
   ],
-  "max_steps": 5640,
-  "num_train_epochs": 10,
-  "total_flos": 1.4686171215861645e+19,
+  "max_steps": 44500,
+  "num_train_epochs": 5,
+  "total_flos": 1.4745575028791706e+19,
   "trial_name": null,
   "trial_params": null
 }