diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5338 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 44361,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.6934,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 0.5953,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 0.5041,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.5038,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4902,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-05,
+      "loss": 0.4758,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-05,
+      "loss": 0.5029,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 0.4621,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-05,
+      "loss": 0.477,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-05,
+      "loss": 0.4989,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994300175554593e-05,
+      "loss": 0.4836,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988600351109186e-05,
+      "loss": 0.4955,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982900526663779e-05,
+      "loss": 0.4802,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.977200702218372e-05,
+      "loss": 0.4741,
+      "step": 700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971500877772965e-05,
+      "loss": 0.4728,
+      "step": 750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965801053327558e-05,
+      "loss": 0.4439,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9601012288821506e-05,
+      "loss": 0.4795,
+      "step": 850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954401404436744e-05,
+      "loss": 0.4612,
+      "step": 900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948701579991337e-05,
+      "loss": 0.4623,
+      "step": 950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9430017555459296e-05,
+      "loss": 0.4743,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9373019311005225e-05,
+      "loss": 0.434,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931602106655115e-05,
+      "loss": 0.461,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925902282209708e-05,
+      "loss": 0.4438,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9202024577643015e-05,
+      "loss": 0.4663,
+      "step": 1200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9145026333188944e-05,
+      "loss": 0.4321,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908802808873487e-05,
+      "loss": 0.4614,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903102984428079e-05,
+      "loss": 0.4656,
+      "step": 1350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.897403159982673e-05,
+      "loss": 0.4643,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8917033355372656e-05,
+      "loss": 0.4566,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8860035110918584e-05,
+      "loss": 0.4545,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880303686646451e-05,
+      "loss": 0.4477,
+      "step": 1550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.874603862201044e-05,
+      "loss": 0.4497,
+      "step": 1600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8689040377556375e-05,
+      "loss": 0.4629,
+      "step": 1650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.86320421331023e-05,
+      "loss": 0.4539,
+      "step": 1700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.857504388864823e-05,
+      "loss": 0.4724,
+      "step": 1750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.851804564419416e-05,
+      "loss": 0.4619,
+      "step": 1800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.846104739974009e-05,
+      "loss": 0.444,
+      "step": 1850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.840404915528602e-05,
+      "loss": 0.4532,
+      "step": 1900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.834705091083195e-05,
+      "loss": 0.4819,
+      "step": 1950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.829005266637788e-05,
+      "loss": 0.4672,
+      "step": 2000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8233054421923806e-05,
+      "loss": 0.4373,
+      "step": 2050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8176056177469734e-05,
+      "loss": 0.4425,
+      "step": 2100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.811905793301567e-05,
+      "loss": 0.442,
+      "step": 2150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.80620596885616e-05,
+      "loss": 0.4365,
+      "step": 2200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8005061444107525e-05,
+      "loss": 0.4481,
+      "step": 2250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.794806319965345e-05,
+      "loss": 0.4639,
+      "step": 2300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.789106495519938e-05,
+      "loss": 0.4527,
+      "step": 2350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.783406671074531e-05,
+      "loss": 0.4575,
+      "step": 2400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7777068466291244e-05,
+      "loss": 0.4626,
+      "step": 2450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.772007022183717e-05,
+      "loss": 0.4235,
+      "step": 2500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.76630719773831e-05,
+      "loss": 0.4496,
+      "step": 2550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.760607373292903e-05,
+      "loss": 0.4355,
+      "step": 2600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7549075488474956e-05,
+      "loss": 0.4361,
+      "step": 2650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.749207724402089e-05,
+      "loss": 0.4655,
+      "step": 2700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.743507899956682e-05,
+      "loss": 0.4295,
+      "step": 2750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.737808075511275e-05,
+      "loss": 0.4567,
+      "step": 2800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7321082510658675e-05,
+      "loss": 0.4501,
+      "step": 2850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.72640842662046e-05,
+      "loss": 0.4518,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.720708602175054e-05,
+      "loss": 0.4625,
+      "step": 2950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.715008777729646e-05,
+      "loss": 0.4563,
+      "step": 3000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.709308953284239e-05,
+      "loss": 0.4593,
+      "step": 3050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7036091288388315e-05,
+      "loss": 0.4316,
+      "step": 3100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.697909304393425e-05,
+      "loss": 0.4465,
+      "step": 3150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.692209479948018e-05,
+      "loss": 0.4427,
+      "step": 3200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6865096555026106e-05,
+      "loss": 0.4623,
+      "step": 3250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6808098310572034e-05,
+      "loss": 0.4505,
+      "step": 3300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.675110006611796e-05,
+      "loss": 0.446,
+      "step": 3350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.669410182166389e-05,
+      "loss": 0.4549,
+      "step": 3400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6637103577209825e-05,
+      "loss": 0.4467,
+      "step": 3450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.658010533275575e-05,
+      "loss": 0.4508,
+      "step": 3500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.652310708830168e-05,
+      "loss": 0.4378,
+      "step": 3550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.646610884384761e-05,
+      "loss": 0.4326,
+      "step": 3600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.640911059939354e-05,
+      "loss": 0.4312,
+      "step": 3650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.635211235493947e-05,
+      "loss": 0.4426,
+      "step": 3700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.62951141104854e-05,
+      "loss": 0.4246,
+      "step": 3750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.623811586603133e-05,
+      "loss": 0.4327,
+      "step": 3800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6181117621577257e-05,
+      "loss": 0.4526,
+      "step": 3850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6124119377123185e-05,
+      "loss": 0.4427,
+      "step": 3900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.606712113266912e-05,
+      "loss": 0.45,
+      "step": 3950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.601012288821505e-05,
+      "loss": 0.4537,
+      "step": 4000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5953124643760975e-05,
+      "loss": 0.4487,
+      "step": 4050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5896126399306904e-05,
+      "loss": 0.4305,
+      "step": 4100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.583912815485283e-05,
+      "loss": 0.4434,
+      "step": 4150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5782129910398766e-05,
+      "loss": 0.4619,
+      "step": 4200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5725131665944694e-05,
+      "loss": 0.4481,
+      "step": 4250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.566813342149062e-05,
+      "loss": 0.4493,
+      "step": 4300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.561113517703655e-05,
+      "loss": 0.4438,
+      "step": 4350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.555413693258248e-05,
+      "loss": 0.4334,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.549713868812841e-05,
+      "loss": 0.433,
+      "step": 4450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.544014044367434e-05,
+      "loss": 0.4493,
+      "step": 4500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.538314219922027e-05,
+      "loss": 0.449,
+      "step": 4550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.53261439547662e-05,
+      "loss": 0.4464,
+      "step": 4600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.526914571031212e-05,
+      "loss": 0.4241,
+      "step": 4650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5212147465858054e-05,
+      "loss": 0.4091,
+      "step": 4700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.515514922140398e-05,
+      "loss": 0.4402,
+      "step": 4750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.509815097694991e-05,
+      "loss": 0.4245,
+      "step": 4800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.504115273249584e-05,
+      "loss": 0.4507,
+      "step": 4850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4984154488041766e-05,
+      "loss": 0.4107,
+      "step": 4900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.49271562435877e-05,
+      "loss": 0.4454,
+      "step": 4950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.487015799913363e-05,
+      "loss": 0.4451,
+      "step": 5000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.481315975467956e-05,
+      "loss": 0.4273,
+      "step": 5050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4756161510225485e-05,
+      "loss": 0.4523,
+      "step": 5100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.469916326577141e-05,
+      "loss": 0.4469,
+      "step": 5150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.464216502131735e-05,
+      "loss": 0.4264,
+      "step": 5200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4585166776863276e-05,
+      "loss": 0.4442,
+      "step": 5250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4528168532409204e-05,
+      "loss": 0.4361,
+      "step": 5300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.447117028795513e-05,
+      "loss": 0.4309,
+      "step": 5350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.441417204350106e-05,
+      "loss": 0.4252,
+      "step": 5400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.4357173799046995e-05,
+      "loss": 0.4442,
+      "step": 5450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.430017555459292e-05,
+      "loss": 0.4425,
+      "step": 5500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.424317731013885e-05,
+      "loss": 0.4469,
+      "step": 5550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.418617906568478e-05,
+      "loss": 0.4222,
+      "step": 5600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.412918082123071e-05,
+      "loss": 0.4431,
+      "step": 5650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.4072182576776635e-05,
+      "loss": 0.4301,
+      "step": 5700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.401518433232257e-05,
+      "loss": 0.4565,
+      "step": 5750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.39581860878685e-05,
+      "loss": 0.4332,
+      "step": 5800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3901187843414426e-05,
+      "loss": 0.4352,
+      "step": 5850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3844189598960354e-05,
+      "loss": 0.4307,
+      "step": 5900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.378719135450628e-05,
+      "loss": 0.4603,
+      "step": 5950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.373019311005222e-05,
+      "loss": 0.4314,
+      "step": 6000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3673194865598145e-05,
+      "loss": 0.4299,
+      "step": 6050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.361619662114407e-05,
+      "loss": 0.4427,
+      "step": 6100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.355919837669e-05,
+      "loss": 0.4358,
+      "step": 6150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.350220013223593e-05,
+      "loss": 0.4483,
+      "step": 6200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.344520188778186e-05,
+      "loss": 0.4326,
+      "step": 6250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3388203643327785e-05,
+      "loss": 0.4367,
+      "step": 6300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.333120539887371e-05,
+      "loss": 0.4305,
+      "step": 6350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.327420715441964e-05,
+      "loss": 0.4274,
+      "step": 6400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3217208909965576e-05,
+      "loss": 0.4265,
+      "step": 6450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3160210665511504e-05,
+      "loss": 0.4519,
+      "step": 6500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.310321242105743e-05,
+      "loss": 0.4335,
+      "step": 6550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.304621417660336e-05,
+      "loss": 0.448,
+      "step": 6600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.298921593214929e-05,
+      "loss": 0.4098,
+      "step": 6650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2932217687695217e-05,
+      "loss": 0.45,
+      "step": 6700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.287521944324115e-05,
+      "loss": 0.4218,
+      "step": 6750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.281822119878708e-05,
+      "loss": 0.446,
+      "step": 6800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.276122295433301e-05,
+      "loss": 0.4319,
+      "step": 6850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2704224709878936e-05,
+      "loss": 0.4232,
+      "step": 6900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2647226465424864e-05,
+      "loss": 0.4258,
+      "step": 6950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.25902282209708e-05,
+      "loss": 0.4277,
+      "step": 7000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2533229976516726e-05,
+      "loss": 0.4264,
+      "step": 7050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2476231732062655e-05,
+      "loss": 0.4419,
+      "step": 7100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.241923348760858e-05,
+      "loss": 0.4369,
+      "step": 7150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.236223524315451e-05,
+      "loss": 0.4532,
+      "step": 7200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2305236998700445e-05,
+      "loss": 0.4205,
+      "step": 7250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.2248238754246373e-05,
+      "loss": 0.406,
+      "step": 7300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.21912405097923e-05,
+      "loss": 0.4436,
+      "step": 7350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.213424226533823e-05,
+      "loss": 0.4443,
+      "step": 7400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.207724402088416e-05,
+      "loss": 0.4353,
+      "step": 7450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.202024577643009e-05,
+      "loss": 0.4181,
+      "step": 7500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.196324753197602e-05,
+      "loss": 0.4202,
+      "step": 7550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.190624928752195e-05,
+      "loss": 0.4327,
+      "step": 7600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.184925104306788e-05,
+      "loss": 0.4169,
+      "step": 7650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1792252798613805e-05,
+      "loss": 0.4559,
+      "step": 7700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.173525455415974e-05,
+      "loss": 0.4114,
+      "step": 7750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.167825630970567e-05,
+      "loss": 0.4371,
+      "step": 7800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1621258065251596e-05,
+      "loss": 0.4198,
+      "step": 7850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.156425982079752e-05,
+      "loss": 0.4338,
+      "step": 7900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1507261576343445e-05,
+      "loss": 0.4651,
+      "step": 7950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.145026333188938e-05,
+      "loss": 0.4341,
+      "step": 8000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.139326508743531e-05,
+      "loss": 0.4449,
+      "step": 8050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1336266842981236e-05,
+      "loss": 0.4415,
+      "step": 8100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.1279268598527164e-05,
+      "loss": 0.4214,
+      "step": 8150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.122227035407309e-05,
+      "loss": 0.4582,
+      "step": 8200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.116527210961903e-05,
+      "loss": 0.4339,
+      "step": 8250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1108273865164955e-05,
+      "loss": 0.4519,
+      "step": 8300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.105127562071088e-05,
+      "loss": 0.4483,
+      "step": 8350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.099427737625681e-05,
+      "loss": 0.4335,
+      "step": 8400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.093727913180274e-05,
+      "loss": 0.4199,
+      "step": 8450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0880280887348674e-05,
+      "loss": 0.4409,
+      "step": 8500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.08232826428946e-05,
+      "loss": 0.4459,
+      "step": 8550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.076628439844053e-05,
+      "loss": 0.4306,
+      "step": 8600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.070928615398646e-05,
+      "loss": 0.4147,
+      "step": 8650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0652287909532386e-05,
+      "loss": 0.4381,
+      "step": 8700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.059528966507832e-05,
+      "loss": 0.4233,
+      "step": 8750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.053829142062425e-05,
+      "loss": 0.4353,
+      "step": 8800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.048129317617018e-05,
+      "loss": 0.4322,
+      "step": 8850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0424294931716105e-05,
+      "loss": 0.4247,
+      "step": 8900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.036729668726203e-05,
+      "loss": 0.4309,
+      "step": 8950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.031029844280796e-05,
+      "loss": 0.4046,
+      "step": 9000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0253300198353896e-05,
+      "loss": 0.4704,
+      "step": 9050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.0196301953899824e-05,
+      "loss": 0.4778,
+      "step": 9100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.013930370944575e-05,
+      "loss": 0.4714,
+      "step": 9150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.008230546499168e-05,
+      "loss": 0.4971,
+      "step": 9200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.002530722053761e-05,
+      "loss": 0.4713,
+      "step": 9250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.996830897608354e-05,
+      "loss": 0.4536,
+      "step": 9300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.991131073162947e-05,
+      "loss": 0.4443,
+      "step": 9350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.98543124871754e-05,
+      "loss": 0.4303,
+      "step": 9400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.979731424272133e-05,
+      "loss": 0.4285,
+      "step": 9450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9740315998267255e-05,
+      "loss": 0.4376,
+      "step": 9500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.968331775381318e-05,
+      "loss": 0.4458,
+      "step": 9550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.962631950935911e-05,
+      "loss": 0.4414,
+      "step": 9600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.956932126490504e-05,
+      "loss": 0.4474,
+      "step": 9650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.951232302045097e-05,
+      "loss": 0.4276,
+      "step": 9700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.94553247759969e-05,
+      "loss": 0.4529,
+      "step": 9750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.939832653154283e-05,
+      "loss": 0.4629,
+      "step": 9800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.934132828708876e-05,
+      "loss": 0.4438,
+      "step": 9850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9284330042634686e-05,
+      "loss": 0.4331,
+      "step": 9900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9227331798180615e-05,
+      "loss": 0.4478,
+      "step": 9950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.917033355372654e-05,
+      "loss": 0.4499,
+      "step": 10000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.911333530927248e-05,
+      "loss": 0.4589,
+      "step": 10050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9056337064818405e-05,
+      "loss": 0.4636,
+      "step": 10100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8999338820364334e-05,
+      "loss": 0.4601,
+      "step": 10150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.894234057591026e-05,
+      "loss": 0.4391,
+      "step": 10200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.888534233145619e-05,
+      "loss": 0.4452,
+      "step": 10250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8828344087002124e-05,
+      "loss": 0.4603,
+      "step": 10300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.877134584254805e-05,
+      "loss": 0.4416,
+      "step": 10350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.871434759809398e-05,
+      "loss": 0.4419,
+      "step": 10400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.865734935363991e-05,
+      "loss": 0.4367,
+      "step": 10450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.860035110918584e-05,
+      "loss": 0.4265,
+      "step": 10500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.854335286473177e-05,
+      "loss": 0.4387,
+      "step": 10550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84863546202777e-05,
+      "loss": 0.4327,
+      "step": 10600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842935637582363e-05,
+      "loss": 0.4645,
+      "step": 10650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8372358131369556e-05,
+      "loss": 0.439,
+      "step": 10700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8315359886915484e-05,
+      "loss": 0.4391,
+      "step": 10750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.825836164246142e-05,
+      "loss": 0.4795,
+      "step": 10800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.8201363398007347e-05,
+      "loss": 0.485,
+      "step": 10850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8144365153553275e-05,
+      "loss": 0.4669,
+      "step": 10900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.80873669090992e-05,
+      "loss": 0.4674,
+      "step": 10950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.803036866464513e-05,
+      "loss": 0.4926,
+      "step": 11000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7973370420191066e-05,
+      "loss": 0.4585,
+      "step": 11050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7916372175736994e-05,
+      "loss": 0.4782,
+      "step": 11100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.785937393128292e-05,
+      "loss": 0.4851,
+      "step": 11150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.780237568682884e-05,
+      "loss": 0.4782,
+      "step": 11200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.774537744237477e-05,
+      "loss": 0.4744,
+      "step": 11250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7688379197920706e-05,
+      "loss": 0.4659,
+      "step": 11300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7631380953466634e-05,
+      "loss": 0.471,
+      "step": 11350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.757438270901256e-05,
+      "loss": 0.4871,
+      "step": 11400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.751738446455849e-05,
+      "loss": 0.4729,
+      "step": 11450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.746038622010442e-05,
+      "loss": 0.4772,
+      "step": 11500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.740338797565035e-05,
+      "loss": 0.4957,
+      "step": 11550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.734638973119628e-05,
+      "loss": 0.4931,
+      "step": 11600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.728939148674221e-05,
+      "loss": 0.4642,
+      "step": 11650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.723239324228814e-05,
+      "loss": 0.4725,
+      "step": 11700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.7175394997834065e-05,
+      "loss": 0.4871,
+      "step": 11750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.711839675338e-05,
+      "loss": 0.4686,
+      "step": 11800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.706139850892593e-05,
+      "loss": 0.4776,
+      "step": 11850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7004400264471856e-05,
+      "loss": 0.4852,
+      "step": 11900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6947402020017784e-05,
+      "loss": 0.4928,
+      "step": 11950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.689040377556371e-05,
+      "loss": 0.4728,
+      "step": 12000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.683340553110965e-05,
+      "loss": 0.462,
+      "step": 12050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6776407286655575e-05,
+      "loss": 0.4577,
+      "step": 12100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.67194090422015e-05,
+      "loss": 0.4697,
+      "step": 12150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.666241079774743e-05,
+      "loss": 0.4914,
+      "step": 12200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.660541255329336e-05,
+      "loss": 0.4493,
+      "step": 12250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.654841430883929e-05,
+      "loss": 0.4601,
+      "step": 12300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.649141606438522e-05,
+      "loss": 0.4577,
+      "step": 12350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.643441781993115e-05,
+      "loss": 0.4645,
+      "step": 12400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.637741957547708e-05,
+      "loss": 0.4722,
+      "step": 12450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6320421331023006e-05,
+      "loss": 0.4707,
+      "step": 12500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6263423086568934e-05,
+      "loss": 0.4611,
+      "step": 12550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.620642484211487e-05,
+      "loss": 0.4907,
+      "step": 12600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.61494265976608e-05,
+      "loss": 0.4848,
+      "step": 12650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6092428353206725e-05,
+      "loss": 0.5054,
+      "step": 12700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.603543010875265e-05,
+      "loss": 0.4926,
+      "step": 12750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.597843186429858e-05,
+      "loss": 0.4608,
+      "step": 12800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.592143361984451e-05,
+      "loss": 0.4791,
+      "step": 12850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.586443537539044e-05,
+      "loss": 0.4826,
+      "step": 12900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5807437130936365e-05,
+      "loss": 0.4558,
+      "step": 12950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5750438886482294e-05,
+      "loss": 0.4822,
+      "step": 13000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.569344064202823e-05,
+      "loss": 0.4668,
+      "step": 13050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5636442397574156e-05,
+      "loss": 0.4707,
+      "step": 13100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5579444153120084e-05,
+      "loss": 0.4704,
+      "step": 13150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.552244590866601e-05,
+      "loss": 0.4502,
+      "step": 13200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.546544766421194e-05,
+      "loss": 0.4552,
+      "step": 13250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.540844941975787e-05,
+      "loss": 0.4639,
+      "step": 13300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5351451175303803e-05,
+      "loss": 0.4518,
+      "step": 13350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.529445293084973e-05,
+      "loss": 0.4547,
+      "step": 13400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.523745468639566e-05,
+      "loss": 0.4349,
+      "step": 13450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.518045644194159e-05,
+      "loss": 0.4243,
+      "step": 13500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5123458197487516e-05,
+      "loss": 0.4471,
+      "step": 13550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.506645995303345e-05,
+      "loss": 0.4511,
+      "step": 13600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.500946170857938e-05,
+      "loss": 0.4786,
+      "step": 13650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.495246346412531e-05,
+      "loss": 0.4378,
+      "step": 13700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4895465219671235e-05,
+      "loss": 0.4542,
+      "step": 13750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.483846697521716e-05,
+      "loss": 0.4533,
+      "step": 13800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.47814687307631e-05,
+      "loss": 0.453,
+      "step": 13850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4724470486309026e-05,
+      "loss": 0.4585,
+      "step": 13900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4667472241854954e-05,
+      "loss": 0.4277,
+      "step": 13950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.461047399740088e-05,
+      "loss": 0.4447,
+      "step": 14000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.455347575294681e-05,
+      "loss": 0.4586,
+      "step": 14050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4496477508492745e-05,
+      "loss": 0.4695,
+      "step": 14100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.443947926403867e-05,
+      "loss": 0.4866,
+      "step": 14150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.43824810195846e-05,
+      "loss": 0.4645,
+      "step": 14200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.432548277513053e-05,
+      "loss": 0.468,
+      "step": 14250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.426848453067646e-05,
+      "loss": 0.4599,
+      "step": 14300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.421148628622239e-05,
+      "loss": 0.4653,
+      "step": 14350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.415448804176832e-05,
+      "loss": 0.4466,
+      "step": 14400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.409748979731425e-05,
+      "loss": 0.4781,
+      "step": 14450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.404049155286017e-05,
+      "loss": 0.4575,
+      "step": 14500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.39834933084061e-05,
+      "loss": 0.4547,
+      "step": 14550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.392649506395203e-05,
+      "loss": 0.436,
+      "step": 14600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.386949681949796e-05,
+      "loss": 0.432,
+      "step": 14650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.381249857504389e-05,
+      "loss": 0.4602,
+      "step": 14700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3755500330589816e-05,
+      "loss": 0.4339,
+      "step": 14750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3698502086135744e-05,
+      "loss": 0.4424,
+      "step": 14800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.364150384168168e-05,
+      "loss": 0.4588,
+      "step": 14850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.358450559722761e-05,
+      "loss": 0.4564,
+      "step": 14900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3527507352773535e-05,
+      "loss": 0.4571,
+      "step": 14950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.347050910831946e-05,
+      "loss": 0.4501,
+      "step": 15000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.341351086386539e-05,
+      "loss": 0.4238,
+      "step": 15050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3356512619411326e-05,
+      "loss": 0.4521,
+      "step": 15100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3299514374957254e-05,
+      "loss": 0.4545,
+      "step": 15150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.324251613050318e-05,
+      "loss": 0.4586,
+      "step": 15200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.318551788604911e-05,
+      "loss": 0.4197,
+      "step": 15250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.312851964159504e-05,
+      "loss": 0.4576,
+      "step": 15300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.307152139714097e-05,
+      "loss": 0.4315,
+      "step": 15350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.30145231526869e-05,
+      "loss": 0.4138,
+      "step": 15400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.295752490823283e-05,
+      "loss": 0.4436,
+      "step": 15450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.290052666377876e-05,
+      "loss": 0.4423,
+      "step": 15500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2843528419324685e-05,
+      "loss": 0.4696,
+      "step": 15550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.278653017487061e-05,
+      "loss": 0.434,
+      "step": 15600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.272953193041655e-05,
+      "loss": 0.4509,
+      "step": 15650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2672533685962476e-05,
+      "loss": 0.4651,
+      "step": 15700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2615535441508404e-05,
+      "loss": 0.5966,
+      "step": 15750
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.255853719705433e-05,
+      "loss": 0.4684,
+      "step": 15800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.250153895260026e-05,
+      "loss": 0.4602,
+      "step": 15850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2444540708146195e-05,
+      "loss": 0.4792,
+      "step": 15900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.238754246369212e-05,
+      "loss": 0.489,
+      "step": 15950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.233054421923805e-05,
+      "loss": 0.4577,
+      "step": 16000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.227354597478398e-05,
+      "loss": 0.474,
+      "step": 16050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.221654773032991e-05,
+      "loss": 0.4908,
+      "step": 16100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2159549485875835e-05,
+      "loss": 0.4544,
+      "step": 16150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.2102551241421763e-05,
+      "loss": 0.4703,
+      "step": 16200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.204555299696769e-05,
+      "loss": 0.4626,
+      "step": 16250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.198855475251362e-05,
+      "loss": 0.452,
+      "step": 16300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1931556508059554e-05,
+      "loss": 0.463,
+      "step": 16350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.187455826360548e-05,
+      "loss": 0.4837,
+      "step": 16400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.181756001915141e-05,
+      "loss": 0.4514,
+      "step": 16450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.176056177469734e-05,
+      "loss": 0.4509,
+      "step": 16500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.170356353024327e-05,
+      "loss": 0.4866,
+      "step": 16550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.16465652857892e-05,
+      "loss": 0.4715,
+      "step": 16600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.158956704133513e-05,
+      "loss": 0.4567,
+      "step": 16650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.153256879688106e-05,
+      "loss": 0.4849,
+      "step": 16700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1475570552426986e-05,
+      "loss": 0.4728,
+      "step": 16750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1418572307972914e-05,
+      "loss": 0.4786,
+      "step": 16800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.136157406351884e-05,
+      "loss": 0.4524,
+      "step": 16850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1304575819064777e-05,
+      "loss": 0.4636,
+      "step": 16900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1247577574610705e-05,
+      "loss": 0.4702,
+      "step": 16950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.119057933015663e-05,
+      "loss": 0.4745,
+      "step": 17000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.113358108570256e-05,
+      "loss": 0.4628,
+      "step": 17050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.107658284124849e-05,
+      "loss": 0.4564,
+      "step": 17100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.1019584596794424e-05,
+      "loss": 0.441,
+      "step": 17150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.096258635234035e-05,
+      "loss": 0.4533,
+      "step": 17200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.090558810788628e-05,
+      "loss": 0.4554,
+      "step": 17250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.084858986343221e-05,
+      "loss": 0.4478,
+      "step": 17300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0791591618978136e-05,
+      "loss": 0.4402,
+      "step": 17350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.073459337452407e-05,
+      "loss": 0.4424,
+      "step": 17400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.067759513007e-05,
+      "loss": 0.442,
+      "step": 17450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.062059688561593e-05,
+      "loss": 0.4467,
+      "step": 17500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0563598641161855e-05,
+      "loss": 0.4426,
+      "step": 17550
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0506600396707786e-05,
+      "loss": 0.4461,
+      "step": 17600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0449602152253714e-05,
+      "loss": 0.4512,
+      "step": 17650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0392603907799642e-05,
+      "loss": 0.4496,
+      "step": 17700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0335605663345567e-05,
+      "loss": 0.4622,
+      "step": 17750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.02786074188915e-05,
+      "loss": 0.4519,
+      "step": 17800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0221609174437427e-05,
+      "loss": 0.4418,
+      "step": 17850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0164610929983355e-05,
+      "loss": 0.4574,
+      "step": 17900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0107612685529286e-05,
+      "loss": 0.4453,
+      "step": 17950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.0050614441075214e-05,
+      "loss": 0.4571,
+      "step": 18000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9993616196621142e-05,
+      "loss": 0.4594,
+      "step": 18050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9936617952167074e-05,
+      "loss": 0.4406,
+      "step": 18100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9879619707713e-05,
+      "loss": 0.4556,
+      "step": 18150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9822621463258933e-05,
+      "loss": 0.4724,
+      "step": 18200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.976562321880486e-05,
+      "loss": 0.4933,
+      "step": 18250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.970862497435079e-05,
+      "loss": 0.469,
+      "step": 18300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.965162672989672e-05,
+      "loss": 0.4717,
+      "step": 18350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.959462848544265e-05,
+      "loss": 0.4661,
+      "step": 18400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.953763024098858e-05,
+      "loss": 0.4688,
+      "step": 18450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9480631996534508e-05,
+      "loss": 0.4663,
+      "step": 18500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9423633752080436e-05,
+      "loss": 0.4566,
+      "step": 18550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9366635507626368e-05,
+      "loss": 0.4506,
+      "step": 18600
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9309637263172296e-05,
+      "loss": 0.474,
+      "step": 18650
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9252639018718224e-05,
+      "loss": 0.4725,
+      "step": 18700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9195640774264155e-05,
+      "loss": 0.4634,
+      "step": 18750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9138642529810083e-05,
+      "loss": 0.4481,
+      "step": 18800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9081644285356015e-05,
+      "loss": 0.4626,
+      "step": 18850
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.9024646040901943e-05,
+      "loss": 0.481,
+      "step": 18900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.896764779644787e-05,
+      "loss": 0.4722,
+      "step": 18950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8910649551993802e-05,
+      "loss": 0.442,
+      "step": 19000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.885365130753973e-05,
+      "loss": 0.4898,
+      "step": 19050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8796653063085662e-05,
+      "loss": 0.4528,
+      "step": 19100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.873965481863159e-05,
+      "loss": 0.4724,
+      "step": 19150
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8682656574177518e-05,
+      "loss": 0.4491,
+      "step": 19200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.862565832972345e-05,
+      "loss": 0.4485,
+      "step": 19250
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8568660085269377e-05,
+      "loss": 0.4482,
+      "step": 19300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8511661840815305e-05,
+      "loss": 0.4829,
+      "step": 19350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.845466359636123e-05,
+      "loss": 0.4445,
+      "step": 19400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.839766535190716e-05,
+      "loss": 0.4405,
+      "step": 19450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.834066710745309e-05,
+      "loss": 0.4384,
+      "step": 19500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8283668862999018e-05,
+      "loss": 0.4467,
+      "step": 19550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.822667061854495e-05,
+      "loss": 0.4562,
+      "step": 19600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8169672374090877e-05,
+      "loss": 0.4699,
+      "step": 19650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8112674129636805e-05,
+      "loss": 0.4678,
+      "step": 19700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8055675885182737e-05,
+      "loss": 0.4582,
+      "step": 19750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7998677640728665e-05,
+      "loss": 0.4573,
+      "step": 19800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7941679396274596e-05,
+      "loss": 0.4693,
+      "step": 19850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7884681151820524e-05,
+      "loss": 0.4623,
+      "step": 19900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7827682907366452e-05,
+      "loss": 0.4467,
+      "step": 19950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7770684662912384e-05,
+      "loss": 0.4454,
+      "step": 20000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.771368641845831e-05,
+      "loss": 0.4502,
+      "step": 20050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7656688174004243e-05,
+      "loss": 0.4671,
+      "step": 20100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.759968992955017e-05,
+      "loss": 0.4675,
+      "step": 20150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.75426916850961e-05,
+      "loss": 0.4585,
+      "step": 20200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.748569344064203e-05,
+      "loss": 0.4725,
+      "step": 20250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.742869519618796e-05,
+      "loss": 0.4883,
+      "step": 20300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7371696951733887e-05,
+      "loss": 0.4526,
+      "step": 20350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7314698707279818e-05,
+      "loss": 0.432,
+      "step": 20400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7257700462825746e-05,
+      "loss": 0.4647,
+      "step": 20450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7200702218371678e-05,
+      "loss": 0.4803,
+      "step": 20500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7143703973917606e-05,
+      "loss": 0.4536,
+      "step": 20550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7086705729463534e-05,
+      "loss": 0.4601,
+      "step": 20600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7029707485009465e-05,
+      "loss": 0.4664,
+      "step": 20650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6972709240555393e-05,
+      "loss": 0.4715,
+      "step": 20700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6915710996101325e-05,
+      "loss": 0.451,
+      "step": 20750
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6858712751647253e-05,
+      "loss": 0.4567,
+      "step": 20800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.680171450719318e-05,
+      "loss": 0.4773,
+      "step": 20850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6744716262739112e-05,
+      "loss": 0.4746,
+      "step": 20900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.668771801828504e-05,
+      "loss": 0.4636,
+      "step": 20950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.663071977383097e-05,
+      "loss": 0.4715,
+      "step": 21000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6573721529376893e-05,
+      "loss": 0.4631,
+      "step": 21050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6516723284922825e-05,
+      "loss": 0.4626,
+      "step": 21100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6459725040468753e-05,
+      "loss": 0.458,
+      "step": 21150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.640272679601468e-05,
+      "loss": 0.4568,
+      "step": 21200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6345728551560612e-05,
+      "loss": 0.4642,
+      "step": 21250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.628873030710654e-05,
+      "loss": 0.4558,
+      "step": 21300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6231732062652468e-05,
+      "loss": 0.4539,
+      "step": 21350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.61747338181984e-05,
+      "loss": 0.4756,
+      "step": 21400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6117735573744328e-05,
+      "loss": 0.4646,
+      "step": 21450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.606073732929026e-05,
+      "loss": 0.4461,
+      "step": 21500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6003739084836187e-05,
+      "loss": 0.4643,
+      "step": 21550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5946740840382115e-05,
+      "loss": 0.4728,
+      "step": 21600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5889742595928047e-05,
+      "loss": 0.4643,
+      "step": 21650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5832744351473975e-05,
+      "loss": 0.4754,
+      "step": 21700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5775746107019906e-05,
+      "loss": 0.5094,
+      "step": 21750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5718747862565834e-05,
+      "loss": 0.548,
+      "step": 21800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5661749618111762e-05,
+      "loss": 0.5629,
+      "step": 21850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5604751373657694e-05,
+      "loss": 0.5656,
+      "step": 21900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5547753129203622e-05,
+      "loss": 0.5313,
+      "step": 21950
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.549075488474955e-05,
+      "loss": 0.5195,
+      "step": 22000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.543375664029548e-05,
+      "loss": 0.5275,
+      "step": 22050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.537675839584141e-05,
+      "loss": 0.5084,
+      "step": 22100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.531976015138734e-05,
+      "loss": 0.4969,
+      "step": 22150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.526276190693327e-05,
+      "loss": 0.4924,
+      "step": 22200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5205763662479197e-05,
+      "loss": 0.5039,
+      "step": 22250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5148765418025128e-05,
+      "loss": 0.4906,
+      "step": 22300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5091767173571056e-05,
+      "loss": 0.4634,
+      "step": 22350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5034768929116988e-05,
+      "loss": 0.4807,
+      "step": 22400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4977770684662912e-05,
+      "loss": 0.4635,
+      "step": 22450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.492077244020884e-05,
+      "loss": 0.4851,
+      "step": 22500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4863774195754772e-05,
+      "loss": 0.4647,
+      "step": 22550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.48067759513007e-05,
+      "loss": 0.4764,
+      "step": 22600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.474977770684663e-05,
+      "loss": 0.4688,
+      "step": 22650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.469277946239256e-05,
+      "loss": 0.4592,
+      "step": 22700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4635781217938488e-05,
+      "loss": 0.4809,
+      "step": 22750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.457878297348442e-05,
+      "loss": 0.4754,
+      "step": 22800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4521784729030347e-05,
+      "loss": 0.4727,
+      "step": 22850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.446478648457628e-05,
+      "loss": 0.4643,
+      "step": 22900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4407788240122207e-05,
+      "loss": 0.4628,
+      "step": 22950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4350789995668135e-05,
+      "loss": 0.4587,
+      "step": 23000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4293791751214066e-05,
+      "loss": 0.4665,
+      "step": 23050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.423679350675999e-05,
+      "loss": 0.4624,
+      "step": 23100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.4179795262305922e-05,
+      "loss": 0.4667,
+      "step": 23150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.412279701785185e-05,
+      "loss": 0.4898,
+      "step": 23200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.4065798773397778e-05,
+      "loss": 0.4808,
+      "step": 23250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.400880052894371e-05,
+      "loss": 0.4781,
+      "step": 23300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3951802284489638e-05,
+      "loss": 0.4926,
+      "step": 23350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.389480404003557e-05,
+      "loss": 0.4804,
+      "step": 23400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3837805795581497e-05,
+      "loss": 0.4645,
+      "step": 23450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3780807551127425e-05,
+      "loss": 0.4867,
+      "step": 23500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3723809306673357e-05,
+      "loss": 0.4763,
+      "step": 23550
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3666811062219285e-05,
+      "loss": 0.4924,
+      "step": 23600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3609812817765213e-05,
+      "loss": 0.4916,
+      "step": 23650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3552814573311144e-05,
+      "loss": 0.4834,
+      "step": 23700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3495816328857072e-05,
+      "loss": 0.4842,
+      "step": 23750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3438818084403004e-05,
+      "loss": 0.4999,
+      "step": 23800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3381819839948932e-05,
+      "loss": 0.4782,
+      "step": 23850
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.332482159549486e-05,
+      "loss": 0.4867,
+      "step": 23900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3267823351040788e-05,
+      "loss": 0.473,
+      "step": 23950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3210825106586716e-05,
+      "loss": 0.4925,
+      "step": 24000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.3153826862132647e-05,
+      "loss": 0.4853,
+      "step": 24050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.3096828617678575e-05,
+      "loss": 0.4928,
+      "step": 24100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.3039830373224504e-05,
+      "loss": 0.4842,
+      "step": 24150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2982832128770435e-05,
+      "loss": 0.4831,
+      "step": 24200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2925833884316363e-05,
+      "loss": 0.4822,
+      "step": 24250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2868835639862294e-05,
+      "loss": 0.4623,
+      "step": 24300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2811837395408223e-05,
+      "loss": 0.4582,
+      "step": 24350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.275483915095415e-05,
+      "loss": 0.4681,
+      "step": 24400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2697840906500082e-05,
+      "loss": 0.467,
+      "step": 24450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.264084266204601e-05,
+      "loss": 0.4718,
+      "step": 24500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.258384441759194e-05,
+      "loss": 0.4712,
+      "step": 24550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.252684617313787e-05,
+      "loss": 0.472,
+      "step": 24600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2469847928683798e-05,
+      "loss": 0.4705,
+      "step": 24650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.241284968422973e-05,
+      "loss": 0.4553,
+      "step": 24700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2355851439775654e-05,
+      "loss": 0.4804,
+      "step": 24750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2298853195321585e-05,
+      "loss": 0.4814,
+      "step": 24800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2241854950867513e-05,
+      "loss": 0.4678,
+      "step": 24850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.218485670641344e-05,
+      "loss": 0.4675,
+      "step": 24900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2127858461959373e-05,
+      "loss": 0.4655,
+      "step": 24950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.20708602175053e-05,
+      "loss": 0.4687,
+      "step": 25000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2013861973051232e-05,
+      "loss": 0.4589,
+      "step": 25050
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.195686372859716e-05,
+      "loss": 0.4671,
+      "step": 25100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.189986548414309e-05,
+      "loss": 0.4679,
+      "step": 25150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.184286723968902e-05,
+      "loss": 0.4652,
+      "step": 25200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1785868995234948e-05,
+      "loss": 0.4703,
+      "step": 25250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1728870750780876e-05,
+      "loss": 0.4699,
+      "step": 25300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1671872506326807e-05,
+      "loss": 0.4767,
+      "step": 25350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1614874261872735e-05,
+      "loss": 0.4657,
+      "step": 25400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1557876017418667e-05,
+      "loss": 0.4697,
+      "step": 25450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1500877772964595e-05,
+      "loss": 0.4573,
+      "step": 25500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1443879528510523e-05,
+      "loss": 0.4732,
+      "step": 25550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.138688128405645e-05,
+      "loss": 0.4723,
+      "step": 25600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.132988303960238e-05,
+      "loss": 0.483,
+      "step": 25650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.127288479514831e-05,
+      "loss": 0.4727,
+      "step": 25700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.121588655069424e-05,
+      "loss": 0.4577,
+      "step": 25750
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.115888830624017e-05,
+      "loss": 0.4557,
+      "step": 25800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1101890061786098e-05,
+      "loss": 0.46,
+      "step": 25850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1044891817332026e-05,
+      "loss": 0.4819,
+      "step": 25900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0987893572877958e-05,
+      "loss": 0.4792,
+      "step": 25950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0930895328423886e-05,
+      "loss": 0.4655,
+      "step": 26000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0873897083969814e-05,
+      "loss": 0.4756,
+      "step": 26050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0816898839515745e-05,
+      "loss": 0.4788,
+      "step": 26100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0759900595061673e-05,
+      "loss": 0.4684,
+      "step": 26150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0702902350607605e-05,
+      "loss": 0.4458,
+      "step": 26200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0645904106153533e-05,
+      "loss": 0.4838,
+      "step": 26250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.058890586169946e-05,
+      "loss": 0.4629,
+      "step": 26300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0531907617245392e-05,
+      "loss": 0.471,
+      "step": 26350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0474909372791317e-05,
+      "loss": 0.4667,
+      "step": 26400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0417911128337248e-05,
+      "loss": 0.4684,
+      "step": 26450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0360912883883176e-05,
+      "loss": 0.4879,
+      "step": 26500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0303914639429104e-05,
+      "loss": 0.4789,
+      "step": 26550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0246916394975036e-05,
+      "loss": 0.4743,
+      "step": 26600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0189918150520964e-05,
+      "loss": 0.4621,
+      "step": 26650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0132919906066895e-05,
+      "loss": 0.4463,
+      "step": 26700
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0075921661612823e-05,
+      "loss": 0.4662,
+      "step": 26750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.001892341715875e-05,
+      "loss": 0.4968,
+      "step": 26800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9961925172704683e-05,
+      "loss": 0.4652,
+      "step": 26850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.990492692825061e-05,
+      "loss": 0.4772,
+      "step": 26900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.984792868379654e-05,
+      "loss": 0.4855,
+      "step": 26950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.979093043934247e-05,
+      "loss": 0.4771,
+      "step": 27000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.97339321948884e-05,
+      "loss": 0.4758,
+      "step": 27050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.967693395043433e-05,
+      "loss": 0.4901,
+      "step": 27100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9619935705980258e-05,
+      "loss": 0.4627,
+      "step": 27150
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9562937461526186e-05,
+      "loss": 0.4656,
+      "step": 27200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9505939217072114e-05,
+      "loss": 0.4754,
+      "step": 27250
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9448940972618042e-05,
+      "loss": 0.4963,
+      "step": 27300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9391942728163973e-05,
+      "loss": 0.4665,
+      "step": 27350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.93349444837099e-05,
+      "loss": 0.4733,
+      "step": 27400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9277946239255833e-05,
+      "loss": 0.4678,
+      "step": 27450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.922094799480176e-05,
+      "loss": 0.4889,
+      "step": 27500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.916394975034769e-05,
+      "loss": 0.4806,
+      "step": 27550
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.910695150589362e-05,
+      "loss": 0.4547,
+      "step": 27600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.904995326143955e-05,
+      "loss": 0.492,
+      "step": 27650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8992955016985477e-05,
+      "loss": 0.4799,
+      "step": 27700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8935956772531408e-05,
+      "loss": 0.4741,
+      "step": 27750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8878958528077336e-05,
+      "loss": 0.4816,
+      "step": 27800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8821960283623268e-05,
+      "loss": 0.4761,
+      "step": 27850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8764962039169196e-05,
+      "loss": 0.4596,
+      "step": 27900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8707963794715124e-05,
+      "loss": 0.4823,
+      "step": 27950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8650965550261055e-05,
+      "loss": 0.4787,
+      "step": 28000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.859396730580698e-05,
+      "loss": 0.464,
+      "step": 28050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.853696906135291e-05,
+      "loss": 0.4784,
+      "step": 28100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.847997081689884e-05,
+      "loss": 0.4931,
+      "step": 28150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8422972572444767e-05,
+      "loss": 0.4663,
+      "step": 28200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.83659743279907e-05,
+      "loss": 0.468,
+      "step": 28250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8308976083536627e-05,
+      "loss": 0.4927,
+      "step": 28300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8251977839082558e-05,
+      "loss": 0.4759,
+      "step": 28350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8194979594628486e-05,
+      "loss": 0.4773,
+      "step": 28400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8137981350174414e-05,
+      "loss": 0.4849,
+      "step": 28450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8080983105720346e-05,
+      "loss": 0.4894,
+      "step": 28500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8023984861266274e-05,
+      "loss": 0.4584,
+      "step": 28550
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7966986616812205e-05,
+      "loss": 0.4739,
+      "step": 28600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7909988372358133e-05,
+      "loss": 0.4971,
+      "step": 28650
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.785299012790406e-05,
+      "loss": 0.4737,
+      "step": 28700
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7795991883449993e-05,
+      "loss": 0.4674,
+      "step": 28750
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.773899363899592e-05,
+      "loss": 0.5095,
+      "step": 28800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.768199539454185e-05,
+      "loss": 0.6674,
+      "step": 28850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7624997150087777e-05,
+      "loss": 0.683,
+      "step": 28900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7567998905633705e-05,
+      "loss": 0.554,
+      "step": 28950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7511000661179637e-05,
+      "loss": 0.529,
+      "step": 29000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7454002416725565e-05,
+      "loss": 0.5206,
+      "step": 29050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7397004172271496e-05,
+      "loss": 0.5223,
+      "step": 29100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7340005927817424e-05,
+      "loss": 0.5134,
+      "step": 29150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7283007683363352e-05,
+      "loss": 0.5048,
+      "step": 29200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7226009438909284e-05,
+      "loss": 0.5268,
+      "step": 29250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.716901119445521e-05,
+      "loss": 0.5021,
+      "step": 29300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.711201295000114e-05,
+      "loss": 0.507,
+      "step": 29350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.705501470554707e-05,
+      "loss": 0.5111,
+      "step": 29400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6998016461093e-05,
+      "loss": 0.5057,
+      "step": 29450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.694101821663893e-05,
+      "loss": 0.5187,
+      "step": 29500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.688401997218486e-05,
+      "loss": 0.5258,
+      "step": 29550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6827021727730787e-05,
+      "loss": 0.5441,
+      "step": 29600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6770023483276718e-05,
+      "loss": 0.513,
+      "step": 29650
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6713025238822643e-05,
+      "loss": 0.5211,
+      "step": 29700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6656026994368574e-05,
+      "loss": 0.5033,
+      "step": 29750
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6599028749914502e-05,
+      "loss": 0.5205,
+      "step": 29800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.654203050546043e-05,
+      "loss": 0.5044,
+      "step": 29850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6485032261006362e-05,
+      "loss": 0.5123,
+      "step": 29900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.642803401655229e-05,
+      "loss": 0.5146,
+      "step": 29950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.637103577209822e-05,
+      "loss": 0.5034,
+      "step": 30000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.631403752764415e-05,
+      "loss": 0.5002,
+      "step": 30050
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6257039283190077e-05,
+      "loss": 0.5213,
+      "step": 30100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.620004103873601e-05,
+      "loss": 0.5083,
+      "step": 30150
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6143042794281937e-05,
+      "loss": 0.5096,
+      "step": 30200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.608604454982787e-05,
+      "loss": 0.5147,
+      "step": 30250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.6029046305373796e-05,
+      "loss": 0.4939,
+      "step": 30300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5972048060919724e-05,
+      "loss": 0.5027,
+      "step": 30350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5915049816465656e-05,
+      "loss": 0.5152,
+      "step": 30400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5858051572011584e-05,
+      "loss": 0.517,
+      "step": 30450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5801053327557512e-05,
+      "loss": 0.4968,
+      "step": 30500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.574405508310344e-05,
+      "loss": 0.5201,
+      "step": 30550
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5687056838649368e-05,
+      "loss": 0.5068,
+      "step": 30600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.56300585941953e-05,
+      "loss": 0.4969,
+      "step": 30650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5573060349741228e-05,
+      "loss": 0.5006,
+      "step": 30700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.551606210528716e-05,
+      "loss": 0.5208,
+      "step": 30750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5459063860833087e-05,
+      "loss": 0.5175,
+      "step": 30800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5402065616379015e-05,
+      "loss": 0.4906,
+      "step": 30850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5345067371924947e-05,
+      "loss": 0.5021,
+      "step": 30900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5288069127470875e-05,
+      "loss": 0.5058,
+      "step": 30950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5231070883016804e-05,
+      "loss": 0.5035,
+      "step": 31000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5174072638562734e-05,
+      "loss": 0.4971,
+      "step": 31050
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5117074394108662e-05,
+      "loss": 0.5042,
+      "step": 31100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5060076149654592e-05,
+      "loss": 0.5126,
+      "step": 31150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5003077905200522e-05,
+      "loss": 0.5091,
+      "step": 31200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4946079660746451e-05,
+      "loss": 0.5054,
+      "step": 31250
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.488908141629238e-05,
+      "loss": 0.5025,
+      "step": 31300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4832083171838308e-05,
+      "loss": 0.5099,
+      "step": 31350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4775084927384236e-05,
+      "loss": 0.4915,
+      "step": 31400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4718086682930165e-05,
+      "loss": 0.5073,
+      "step": 31450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4661088438476095e-05,
+      "loss": 0.5031,
+      "step": 31500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4604090194022025e-05,
+      "loss": 0.5181,
+      "step": 31550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4547091949567953e-05,
+      "loss": 0.5257,
+      "step": 31600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4490093705113883e-05,
+      "loss": 0.5142,
+      "step": 31650
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4433095460659812e-05,
+      "loss": 0.5039,
+      "step": 31700
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4376097216205742e-05,
+      "loss": 0.5099,
+      "step": 31750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.431909897175167e-05,
+      "loss": 0.5162,
+      "step": 31800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.42621007272976e-05,
+      "loss": 0.505,
+      "step": 31850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.420510248284353e-05,
+      "loss": 0.503,
+      "step": 31900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.414810423838946e-05,
+      "loss": 0.5164,
+      "step": 31950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.409110599393539e-05,
+      "loss": 0.5147,
+      "step": 32000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4034107749481317e-05,
+      "loss": 0.4865,
+      "step": 32050
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3977109505027247e-05,
+      "loss": 0.5074,
+      "step": 32100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3920111260573173e-05,
+      "loss": 0.4984,
+      "step": 32150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3863113016119103e-05,
+      "loss": 0.4858,
+      "step": 32200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3806114771665033e-05,
+      "loss": 0.5167,
+      "step": 32250
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3749116527210963e-05,
+      "loss": 0.5001,
+      "step": 32300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.369211828275689e-05,
+      "loss": 0.5178,
+      "step": 32350
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.363512003830282e-05,
+      "loss": 0.5179,
+      "step": 32400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.357812179384875e-05,
+      "loss": 0.4965,
+      "step": 32450
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.352112354939468e-05,
+      "loss": 0.503,
+      "step": 32500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3464125304940608e-05,
+      "loss": 0.5151,
+      "step": 32550
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3407127060486538e-05,
+      "loss": 0.4917,
+      "step": 32600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3350128816032467e-05,
+      "loss": 0.4839,
+      "step": 32650
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3293130571578397e-05,
+      "loss": 0.5002,
+      "step": 32700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3236132327124325e-05,
+      "loss": 0.5082,
+      "step": 32750
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3179134082670255e-05,
+      "loss": 0.5221,
+      "step": 32800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3122135838216185e-05,
+      "loss": 0.5093,
+      "step": 32850
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3065137593762114e-05,
+      "loss": 0.5144,
+      "step": 32900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.300813934930804e-05,
+      "loss": 0.4986,
+      "step": 32950
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.295114110485397e-05,
+      "loss": 0.5137,
+      "step": 33000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2894142860399899e-05,
+      "loss": 0.5076,
+      "step": 33050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2837144615945828e-05,
+      "loss": 0.5125,
+      "step": 33100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2780146371491758e-05,
+      "loss": 0.4967,
+      "step": 33150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2723148127037688e-05,
+      "loss": 0.5228,
+      "step": 33200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2666149882583616e-05,
+      "loss": 0.4989,
+      "step": 33250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2609151638129546e-05,
+      "loss": 0.5216,
+      "step": 33300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2552153393675475e-05,
+      "loss": 0.5039,
+      "step": 33350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2495155149221405e-05,
+      "loss": 0.5053,
+      "step": 33400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2438156904767335e-05,
+      "loss": 0.5013,
+      "step": 33450
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2381158660313263e-05,
+      "loss": 0.513,
+      "step": 33500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2324160415859193e-05,
+      "loss": 0.4902,
+      "step": 33550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.226716217140512e-05,
+      "loss": 0.519,
+      "step": 33600
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.221016392695105e-05,
+      "loss": 0.4956,
+      "step": 33650
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.215316568249698e-05,
+      "loss": 0.5088,
+      "step": 33700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2096167438042908e-05,
+      "loss": 0.5174,
+      "step": 33750
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.2039169193588838e-05,
+      "loss": 0.5183,
+      "step": 33800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1982170949134768e-05,
+      "loss": 0.4934,
+      "step": 33850
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1925172704680698e-05,
+      "loss": 0.5109,
+      "step": 33900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1868174460226626e-05,
+      "loss": 0.5163,
+      "step": 33950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1811176215772554e-05,
+      "loss": 0.5088,
+      "step": 34000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1754177971318483e-05,
+      "loss": 0.5143,
+      "step": 34050
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1697179726864413e-05,
+      "loss": 0.5087,
+      "step": 34100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1640181482410343e-05,
+      "loss": 0.5253,
+      "step": 34150
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1583183237956271e-05,
+      "loss": 0.5104,
+      "step": 34200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.15261849935022e-05,
+      "loss": 0.513,
+      "step": 34250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.146918674904813e-05,
+      "loss": 0.5103,
+      "step": 34300
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.141218850459406e-05,
+      "loss": 0.4975,
+      "step": 34350
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1355190260139988e-05,
+      "loss": 0.5092,
+      "step": 34400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1298192015685916e-05,
+      "loss": 0.5011,
+      "step": 34450
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1241193771231846e-05,
+      "loss": 0.5027,
+      "step": 34500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1184195526777776e-05,
+      "loss": 0.5034,
+      "step": 34550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1127197282323706e-05,
+      "loss": 0.4982,
+      "step": 34600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1070199037869634e-05,
+      "loss": 0.5037,
+      "step": 34650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1013200793415563e-05,
+      "loss": 0.4958,
+      "step": 34700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0956202548961493e-05,
+      "loss": 0.5199,
+      "step": 34750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0899204304507421e-05,
+      "loss": 0.5108,
+      "step": 34800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0842206060053351e-05,
+      "loss": 0.505,
+      "step": 34850
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0785207815599279e-05,
+      "loss": 0.5053,
+      "step": 34900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0728209571145209e-05,
+      "loss": 0.5072,
+      "step": 34950
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0671211326691138e-05,
+      "loss": 0.523,
+      "step": 35000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0614213082237068e-05,
+      "loss": 0.5121,
+      "step": 35050
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0557214837782998e-05,
+      "loss": 0.5072,
+      "step": 35100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0500216593328926e-05,
+      "loss": 0.4977,
+      "step": 35150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0443218348874856e-05,
+      "loss": 0.5165,
+      "step": 35200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0386220104420784e-05,
+      "loss": 0.5042,
+      "step": 35250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0329221859966714e-05,
+      "loss": 0.516,
+      "step": 35300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0272223615512643e-05,
+      "loss": 0.4964,
+      "step": 35350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0215225371058571e-05,
+      "loss": 0.4903,
+      "step": 35400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0158227126604501e-05,
+      "loss": 0.5009,
+      "step": 35450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.010122888215043e-05,
+      "loss": 0.5015,
+      "step": 35500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.004423063769636e-05,
+      "loss": 0.4986,
+      "step": 35550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.987232393242289e-06,
+      "loss": 0.5018,
+      "step": 35600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.930234148788217e-06,
+      "loss": 0.5197,
+      "step": 35650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.873235904334146e-06,
+      "loss": 0.5129,
+      "step": 35700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.816237659880076e-06,
+      "loss": 0.5091,
+      "step": 35750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.759239415426006e-06,
+      "loss": 0.5197,
+      "step": 35800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.702241170971934e-06,
+      "loss": 0.4794,
+      "step": 35850
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.645242926517864e-06,
+      "loss": 0.5027,
+      "step": 35900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.588244682063793e-06,
+      "loss": 0.5051,
+      "step": 35950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.531246437609723e-06,
+      "loss": 0.5007,
+      "step": 36000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.474248193155651e-06,
+      "loss": 0.5028,
+      "step": 36050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.41724994870158e-06,
+      "loss": 0.5173,
+      "step": 36100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.360251704247509e-06,
+      "loss": 0.4957,
+      "step": 36150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.303253459793439e-06,
+      "loss": 0.4869,
+      "step": 36200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.246255215339369e-06,
+      "loss": 0.5006,
+      "step": 36250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.189256970885297e-06,
+      "loss": 0.5063,
+      "step": 36300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.132258726431226e-06,
+      "loss": 0.4872,
+      "step": 36350
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.075260481977156e-06,
+      "loss": 0.5157,
+      "step": 36400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.018262237523084e-06,
+      "loss": 0.4927,
+      "step": 36450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.961263993069014e-06,
+      "loss": 0.5193,
+      "step": 36500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.904265748614942e-06,
+      "loss": 0.4993,
+      "step": 36550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.847267504160872e-06,
+      "loss": 0.5108,
+      "step": 36600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.790269259706801e-06,
+      "loss": 0.4985,
+      "step": 36650
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.733271015252731e-06,
+      "loss": 0.5015,
+      "step": 36700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.676272770798661e-06,
+      "loss": 0.4881,
+      "step": 36750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.619274526344589e-06,
+      "loss": 0.4971,
+      "step": 36800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.562276281890517e-06,
+      "loss": 0.4989,
+      "step": 36850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.505278037436447e-06,
+      "loss": 0.505,
+      "step": 36900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.448279792982377e-06,
+      "loss": 0.5069,
+      "step": 36950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.391281548528306e-06,
+      "loss": 0.5175,
+      "step": 37000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.334283304074234e-06,
+      "loss": 0.4933,
+      "step": 37050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.277285059620164e-06,
+      "loss": 0.5073,
+      "step": 37100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.220286815166094e-06,
+      "loss": 0.5014,
+      "step": 37150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.163288570712024e-06,
+      "loss": 0.5197,
+      "step": 37200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.106290326257952e-06,
+      "loss": 0.5035,
+      "step": 37250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.04929208180388e-06,
+      "loss": 0.4902,
+      "step": 37300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.99229383734981e-06,
+      "loss": 0.5123,
+      "step": 37350
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.93529559289574e-06,
+      "loss": 0.5182,
+      "step": 37400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.878297348441669e-06,
+      "loss": 0.5093,
+      "step": 37450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.821299103987597e-06,
+      "loss": 0.4936,
+      "step": 37500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.764300859533527e-06,
+      "loss": 0.4883,
+      "step": 37550
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.707302615079457e-06,
+      "loss": 0.4968,
+      "step": 37600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.650304370625386e-06,
+      "loss": 0.4925,
+      "step": 37650
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.5933061261713135e-06,
+      "loss": 0.5,
+      "step": 37700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.536307881717243e-06,
+      "loss": 0.5009,
+      "step": 37750
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.479309637263172e-06,
+      "loss": 0.5161,
+      "step": 37800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.422311392809102e-06,
+      "loss": 0.5119,
+      "step": 37850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.365313148355031e-06,
+      "loss": 0.4972,
+      "step": 37900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.3083149039009605e-06,
+      "loss": 0.5116,
+      "step": 37950
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.251316659446889e-06,
+      "loss": 0.5006,
+      "step": 38000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.194318414992819e-06,
+      "loss": 0.5207,
+      "step": 38050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.137320170538747e-06,
+      "loss": 0.505,
+      "step": 38100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.080321926084676e-06,
+      "loss": 0.5112,
+      "step": 38150
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.023323681630606e-06,
+      "loss": 0.5169,
+      "step": 38200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.966325437176536e-06,
+      "loss": 0.5127,
+      "step": 38250
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.9093271927224645e-06,
+      "loss": 0.5065,
+      "step": 38300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.852328948268394e-06,
+      "loss": 0.4879,
+      "step": 38350
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.795330703814323e-06,
+      "loss": 0.5281,
+      "step": 38400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.738332459360253e-06,
+      "loss": 0.5051,
+      "step": 38450
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.681334214906181e-06,
+      "loss": 0.5217,
+      "step": 38500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.62433597045211e-06,
+      "loss": 0.4942,
+      "step": 38550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.56733772599804e-06,
+      "loss": 0.5241,
+      "step": 38600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5103394815439685e-06,
+      "loss": 0.4991,
+      "step": 38650
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.453341237089898e-06,
+      "loss": 0.5034,
+      "step": 38700
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.396342992635827e-06,
+      "loss": 0.5008,
+      "step": 38750
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.339344748181757e-06,
+      "loss": 0.5183,
+      "step": 38800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.282346503727686e-06,
+      "loss": 0.5006,
+      "step": 38850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.225348259273615e-06,
+      "loss": 0.4936,
+      "step": 38900
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1683500148195444e-06,
+      "loss": 0.4758,
+      "step": 38950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.111351770365473e-06,
+      "loss": 0.4654,
+      "step": 39000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.054353525911402e-06,
+      "loss": 0.4584,
+      "step": 39050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.997355281457331e-06,
+      "loss": 0.4929,
+      "step": 39100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.940357037003261e-06,
+      "loss": 0.4793,
+      "step": 39150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.88335879254919e-06,
+      "loss": 0.4625,
+      "step": 39200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.826360548095119e-06,
+      "loss": 0.462,
+      "step": 39250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7693623036410484e-06,
+      "loss": 0.4598,
+      "step": 39300
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.712364059186977e-06,
+      "loss": 0.4778,
+      "step": 39350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.655365814732907e-06,
+      "loss": 0.4707,
+      "step": 39400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.598367570278835e-06,
+      "loss": 0.4468,
+      "step": 39450
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.541369325824765e-06,
+      "loss": 0.454,
+      "step": 39500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.484371081370694e-06,
+      "loss": 0.4454,
+      "step": 39550
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.4273728369166235e-06,
+      "loss": 0.4493,
+      "step": 39600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3703745924625524e-06,
+      "loss": 0.451,
+      "step": 39650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.313376348008481e-06,
+      "loss": 0.4515,
+      "step": 39700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.256378103554411e-06,
+      "loss": 0.4714,
+      "step": 39750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.19937985910034e-06,
+      "loss": 0.4658,
+      "step": 39800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.142381614646269e-06,
+      "loss": 0.4694,
+      "step": 39850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.085383370192199e-06,
+      "loss": 0.4487,
+      "step": 39900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0283851257381275e-06,
+      "loss": 0.4528,
+      "step": 39950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.971386881284057e-06,
+      "loss": 0.4515,
+      "step": 40000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.914388636829985e-06,
+      "loss": 0.4878,
+      "step": 40050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.857390392375915e-06,
+      "loss": 0.4668,
+      "step": 40100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.800392147921844e-06,
+      "loss": 0.4651,
+      "step": 40150
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.743393903467774e-06,
+      "loss": 0.4722,
+      "step": 40200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.686395659013703e-06,
+      "loss": 0.47,
+      "step": 40250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.6293974145596315e-06,
+      "loss": 0.4629,
+      "step": 40300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.572399170105561e-06,
+      "loss": 0.4534,
+      "step": 40350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.51540092565149e-06,
+      "loss": 0.4816,
+      "step": 40400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.458402681197419e-06,
+      "loss": 0.4601,
+      "step": 40450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.401404436743348e-06,
+      "loss": 0.4642,
+      "step": 40500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.344406192289278e-06,
+      "loss": 0.4699,
+      "step": 40550
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2874079478352075e-06,
+      "loss": 0.4459,
+      "step": 40600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2304097033811355e-06,
+      "loss": 0.4494,
+      "step": 40650
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.173411458927065e-06,
+      "loss": 0.4565,
+      "step": 40700
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.116413214472994e-06,
+      "loss": 0.4567,
+      "step": 40750
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.059414970018924e-06,
+      "loss": 0.4602,
+      "step": 40800
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.002416725564853e-06,
+      "loss": 0.4396,
+      "step": 40850
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.945418481110782e-06,
+      "loss": 0.459,
+      "step": 40900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8884202366567115e-06,
+      "loss": 0.4686,
+      "step": 40950
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.83142199220264e-06,
+      "loss": 0.462,
+      "step": 41000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7744237477485697e-06,
+      "loss": 0.4658,
+      "step": 41050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7174255032944986e-06,
+      "loss": 0.4596,
+      "step": 41100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.660427258840428e-06,
+      "loss": 0.4557,
+      "step": 41150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6034290143863572e-06,
+      "loss": 0.4635,
+      "step": 41200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5464307699322866e-06,
+      "loss": 0.4548,
+      "step": 41250
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.489432525478215e-06,
+      "loss": 0.4472,
+      "step": 41300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4324342810241448e-06,
+      "loss": 0.4563,
+      "step": 41350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.375436036570074e-06,
+      "loss": 0.4541,
+      "step": 41400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3184377921160034e-06,
+      "loss": 0.4619,
+      "step": 41450
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.261439547661932e-06,
+      "loss": 0.4267,
+      "step": 41500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2044413032078612e-06,
+      "loss": 0.4374,
+      "step": 41550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1474430587537906e-06,
+      "loss": 0.4332,
+      "step": 41600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0904448142997195e-06,
+      "loss": 0.4487,
+      "step": 41650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.033446569845649e-06,
+      "loss": 0.4467,
+      "step": 41700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.976448325391578e-06,
+      "loss": 0.4488,
+      "step": 41750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9194500809375074e-06,
+      "loss": 0.4473,
+      "step": 41800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8624518364834363e-06,
+      "loss": 0.4497,
+      "step": 41850
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8054535920293657e-06,
+      "loss": 0.4458,
+      "step": 41900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7484553475752946e-06,
+      "loss": 0.4496,
+      "step": 41950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.691457103121224e-06,
+      "loss": 0.4375,
+      "step": 42000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.634458858667153e-06,
+      "loss": 0.4403,
+      "step": 42050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5774606142130825e-06,
+      "loss": 0.4317,
+      "step": 42100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5204623697590114e-06,
+      "loss": 0.4473,
+      "step": 42150
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4634641253049407e-06,
+      "loss": 0.4407,
+      "step": 42200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4064658808508696e-06,
+      "loss": 0.414,
+      "step": 42250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.349467636396799e-06,
+      "loss": 0.4204,
+      "step": 42300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2924693919427283e-06,
+      "loss": 0.4546,
+      "step": 42350
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2354711474886576e-06,
+      "loss": 0.4494,
+      "step": 42400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1784729030345865e-06,
+      "loss": 0.4347,
+      "step": 42450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.121474658580516e-06,
+      "loss": 0.461,
+      "step": 42500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0644764141264447e-06,
+      "loss": 0.4579,
+      "step": 42550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.007478169672374e-06,
+      "loss": 0.4155,
+      "step": 42600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9504799252183034e-06,
+      "loss": 0.4498,
+      "step": 42650
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8934816807642325e-06,
+      "loss": 0.4504,
+      "step": 42700
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8364834363101616e-06,
+      "loss": 0.4387,
+      "step": 42750
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.779485191856091e-06,
+      "loss": 0.4304,
+      "step": 42800
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7224869474020203e-06,
+      "loss": 0.4286,
+      "step": 42850
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6654887029479494e-06,
+      "loss": 0.4177,
+      "step": 42900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6084904584938787e-06,
+      "loss": 0.4518,
+      "step": 42950
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5514922140398076e-06,
+      "loss": 0.4462,
+      "step": 43000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4944939695857367e-06,
+      "loss": 0.432,
+      "step": 43050
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.437495725131666e-06,
+      "loss": 0.4284,
+      "step": 43100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3804974806775951e-06,
+      "loss": 0.4183,
+      "step": 43150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3234992362235243e-06,
+      "loss": 0.4254,
+      "step": 43200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2665009917694536e-06,
+      "loss": 0.4459,
+      "step": 43250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2095027473153827e-06,
+      "loss": 0.433,
+      "step": 43300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1525045028613118e-06,
+      "loss": 0.4376,
+      "step": 43350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0955062584072411e-06,
+      "loss": 0.4389,
+      "step": 43400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0385080139531702e-06,
+      "loss": 0.4332,
+      "step": 43450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.815097694990994e-07,
+      "loss": 0.4354,
+      "step": 43500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.245115250450286e-07,
+      "loss": 0.4511,
+      "step": 43550
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.675132805909578e-07,
+      "loss": 0.4149,
+      "step": 43600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.105150361368869e-07,
+      "loss": 0.4322,
+      "step": 43650
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.535167916828162e-07,
+      "loss": 0.4128,
+      "step": 43700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.965185472287453e-07,
+      "loss": 0.4194,
+      "step": 43750
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.395203027746746e-07,
+      "loss": 0.4325,
+      "step": 43800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.825220583206038e-07,
+      "loss": 0.4287,
+      "step": 43850
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.255238138665329e-07,
+      "loss": 0.4353,
+      "step": 43900
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.685255694124621e-07,
+      "loss": 0.4207,
+      "step": 43950
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.1152732495839127e-07,
+      "loss": 0.4465,
+      "step": 44000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.545290805043205e-07,
+      "loss": 0.4238,
+      "step": 44050
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9753083605024965e-07,
+      "loss": 0.4226,
+      "step": 44100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4053259159617887e-07,
+      "loss": 0.4465,
+      "step": 44150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8353434714210804e-07,
+      "loss": 0.4094,
+      "step": 44200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.265361026880372e-07,
+      "loss": 0.4491,
+      "step": 44250
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.95378582339664e-08,
+      "loss": 0.4208,
+      "step": 44300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.253961377989558e-08,
+      "loss": 0.4375,
+      "step": 44350
+    }
+  ],
+  "max_steps": 44361,
+  "num_train_epochs": 3,
+  "total_flos": 1.77824607798456e+17,
+  "trial_name": null,
+  "trial_params": null
+}