diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15793 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 20.0,
+  "global_step": 65700,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.998599695585997e-05,
+      "loss": 3.8582,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.997138508371385e-05,
+      "loss": 3.284,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.995616438356165e-05,
+      "loss": 3.1009,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.994094368340944e-05,
+      "loss": 3.1079,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9925722983257234e-05,
+      "loss": 3.1098,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.991050228310503e-05,
+      "loss": 3.1987,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.989528158295282e-05,
+      "loss": 2.9982,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9880060882800616e-05,
+      "loss": 2.9164,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.9864840182648406e-05,
+      "loss": 2.9484,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.9849619482496195e-05,
+      "loss": 2.8611,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.983439878234399e-05,
+      "loss": 2.9438,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.981917808219178e-05,
+      "loss": 3.0905,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.980395738203958e-05,
+      "loss": 3.0313,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.978873668188737e-05,
+      "loss": 2.9305,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.977351598173516e-05,
+      "loss": 2.9258,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.975829528158296e-05,
+      "loss": 2.9008,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.974307458143075e-05,
+      "loss": 2.7581,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.9727853881278545e-05,
+      "loss": 2.8324,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.9712633181126334e-05,
+      "loss": 2.6921,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9697412480974123e-05,
+      "loss": 2.8431,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.968219178082192e-05,
+      "loss": 2.7828,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.9666971080669716e-05,
+      "loss": 2.8666,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9651750380517505e-05,
+      "loss": 2.8547,
+      "step": 575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.9637138508371385e-05,
+      "loss": 2.6909,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.962191780821918e-05,
+      "loss": 2.7679,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.960669710806698e-05,
+      "loss": 2.7284,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.959147640791477e-05,
+      "loss": 2.902,
+      "step": 675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.957625570776256e-05,
+      "loss": 2.8097,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.956103500761035e-05,
+      "loss": 2.7846,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.954581430745815e-05,
+      "loss": 2.7917,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.953059360730594e-05,
+      "loss": 2.7978,
+      "step": 775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.951537290715373e-05,
+      "loss": 2.7402,
+      "step": 800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.9500152207001524e-05,
+      "loss": 2.7104,
+      "step": 825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.948493150684932e-05,
+      "loss": 2.7742,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.946971080669711e-05,
+      "loss": 2.6598,
+      "step": 875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.9454490106544906e-05,
+      "loss": 2.6493,
+      "step": 900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.9439269406392695e-05,
+      "loss": 2.6935,
+      "step": 925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.942404870624049e-05,
+      "loss": 2.6664,
+      "step": 950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.940882800608828e-05,
+      "loss": 2.7125,
+      "step": 975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.939360730593608e-05,
+      "loss": 2.77,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9378386605783867e-05,
+      "loss": 2.7562,
+      "step": 1025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.936316590563166e-05,
+      "loss": 2.699,
+      "step": 1050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.934794520547946e-05,
+      "loss": 2.8319,
+      "step": 1075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.933272450532725e-05,
+      "loss": 2.6888,
+      "step": 1100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9317503805175045e-05,
+      "loss": 2.7843,
+      "step": 1125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9302283105022834e-05,
+      "loss": 2.6738,
+      "step": 1150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9287062404870624e-05,
+      "loss": 2.7412,
+      "step": 1175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.927184170471842e-05,
+      "loss": 2.6732,
+      "step": 1200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.925662100456621e-05,
+      "loss": 2.7467,
+      "step": 1225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9241400304414005e-05,
+      "loss": 2.7866,
+      "step": 1250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.92261796042618e-05,
+      "loss": 2.8271,
+      "step": 1275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.921095890410959e-05,
+      "loss": 2.6714,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.919573820395739e-05,
+      "loss": 2.6909,
+      "step": 1325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.918051750380518e-05,
+      "loss": 2.7683,
+      "step": 1350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.916529680365297e-05,
+      "loss": 2.7829,
+      "step": 1375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.915007610350076e-05,
+      "loss": 2.6544,
+      "step": 1400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.913485540334855e-05,
+      "loss": 2.6491,
+      "step": 1425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9119634703196355e-05,
+      "loss": 2.7998,
+      "step": 1450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9104414003044144e-05,
+      "loss": 2.6747,
+      "step": 1475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9089193302891934e-05,
+      "loss": 2.8246,
+      "step": 1500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.907397260273973e-05,
+      "loss": 2.6864,
+      "step": 1525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.905875190258752e-05,
+      "loss": 2.7112,
+      "step": 1550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9043531202435316e-05,
+      "loss": 2.641,
+      "step": 1575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9028310502283105e-05,
+      "loss": 2.7254,
+      "step": 1600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.90130898021309e-05,
+      "loss": 2.6282,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.89978691019787e-05,
+      "loss": 2.6655,
+      "step": 1650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.898264840182649e-05,
+      "loss": 2.646,
+      "step": 1675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.896742770167428e-05,
+      "loss": 2.6284,
+      "step": 1700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.895220700152207e-05,
+      "loss": 2.6256,
+      "step": 1725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.893698630136987e-05,
+      "loss": 2.5699,
+      "step": 1750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.892176560121766e-05,
+      "loss": 2.6774,
+      "step": 1775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.890654490106545e-05,
+      "loss": 2.7368,
+      "step": 1800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.8891324200913244e-05,
+      "loss": 2.6614,
+      "step": 1825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.887610350076104e-05,
+      "loss": 2.6495,
+      "step": 1850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.886088280060883e-05,
+      "loss": 2.6243,
+      "step": 1875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8845662100456626e-05,
+      "loss": 2.5812,
+      "step": 1900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8830441400304416e-05,
+      "loss": 2.6791,
+      "step": 1925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.881522070015221e-05,
+      "loss": 2.5731,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.88e-05,
+      "loss": 2.5387,
+      "step": 1975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.87847792998478e-05,
+      "loss": 2.6743,
+      "step": 2000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.876955859969559e-05,
+      "loss": 2.5973,
+      "step": 2025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.875433789954338e-05,
+      "loss": 2.6038,
+      "step": 2050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.873911719939118e-05,
+      "loss": 2.7553,
+      "step": 2075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.872389649923897e-05,
+      "loss": 2.7192,
+      "step": 2100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.870867579908676e-05,
+      "loss": 2.6565,
+      "step": 2125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8693455098934554e-05,
+      "loss": 2.5753,
+      "step": 2150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8678234398782344e-05,
+      "loss": 2.501,
+      "step": 2175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.866301369863014e-05,
+      "loss": 2.72,
+      "step": 2200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.864779299847793e-05,
+      "loss": 2.5979,
+      "step": 2225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8632572298325726e-05,
+      "loss": 2.587,
+      "step": 2250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.861735159817352e-05,
+      "loss": 2.7236,
+      "step": 2275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.860213089802131e-05,
+      "loss": 2.6234,
+      "step": 2300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.858691019786911e-05,
+      "loss": 2.652,
+      "step": 2325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85716894977169e-05,
+      "loss": 2.5885,
+      "step": 2350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8556468797564693e-05,
+      "loss": 2.6084,
+      "step": 2375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.854124809741248e-05,
+      "loss": 2.693,
+      "step": 2400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.852602739726027e-05,
+      "loss": 2.6778,
+      "step": 2425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.851080669710807e-05,
+      "loss": 2.591,
+      "step": 2450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.8495585996955865e-05,
+      "loss": 2.5932,
+      "step": 2475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.8480365296803654e-05,
+      "loss": 2.5999,
+      "step": 2500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.846514459665145e-05,
+      "loss": 2.5795,
+      "step": 2525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.844992389649924e-05,
+      "loss": 2.543,
+      "step": 2550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.8434703196347036e-05,
+      "loss": 2.5534,
+      "step": 2575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8419482496194826e-05,
+      "loss": 2.6031,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.840426179604262e-05,
+      "loss": 2.6382,
+      "step": 2625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.838904109589042e-05,
+      "loss": 2.5381,
+      "step": 2650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.837382039573821e-05,
+      "loss": 2.4354,
+      "step": 2675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8358599695586004e-05,
+      "loss": 2.6449,
+      "step": 2700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.834337899543379e-05,
+      "loss": 2.5669,
+      "step": 2725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.832815829528158e-05,
+      "loss": 2.6498,
+      "step": 2750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.831293759512938e-05,
+      "loss": 2.6078,
+      "step": 2775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.829771689497717e-05,
+      "loss": 2.6951,
+      "step": 2800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8282496194824965e-05,
+      "loss": 2.5454,
+      "step": 2825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.826727549467276e-05,
+      "loss": 2.6792,
+      "step": 2850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.825205479452055e-05,
+      "loss": 2.5385,
+      "step": 2875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8236834094368346e-05,
+      "loss": 2.502,
+      "step": 2900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8221613394216136e-05,
+      "loss": 2.5359,
+      "step": 2925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.820639269406393e-05,
+      "loss": 2.5942,
+      "step": 2950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.819117199391172e-05,
+      "loss": 2.5167,
+      "step": 2975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.817595129375952e-05,
+      "loss": 2.4514,
+      "step": 3000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.816073059360731e-05,
+      "loss": 2.6363,
+      "step": 3025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8145509893455103e-05,
+      "loss": 2.39,
+      "step": 3050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.813028919330289e-05,
+      "loss": 2.5625,
+      "step": 3075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.811506849315069e-05,
+      "loss": 2.6096,
+      "step": 3100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.809984779299848e-05,
+      "loss": 2.7414,
+      "step": 3125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8084627092846275e-05,
+      "loss": 2.6735,
+      "step": 3150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8069406392694064e-05,
+      "loss": 2.6245,
+      "step": 3175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.805418569254186e-05,
+      "loss": 2.5353,
+      "step": 3200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.803896499238965e-05,
+      "loss": 2.4748,
+      "step": 3225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8023744292237446e-05,
+      "loss": 2.6444,
+      "step": 3250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.800852359208524e-05,
+      "loss": 2.6216,
+      "step": 3275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.799330289193303e-05,
+      "loss": 2.6074,
+      "step": 3300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.797808219178083e-05,
+      "loss": 2.5025,
+      "step": 3325
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.796286149162862e-05,
+      "loss": 2.62,
+      "step": 3350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.794764079147641e-05,
+      "loss": 2.521,
+      "step": 3375
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.79324200913242e-05,
+      "loss": 2.4202,
+      "step": 3400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.791719939117199e-05,
+      "loss": 2.5935,
+      "step": 3425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.790197869101979e-05,
+      "loss": 2.5516,
+      "step": 3450
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.7886757990867585e-05,
+      "loss": 2.5356,
+      "step": 3475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.7871537290715375e-05,
+      "loss": 2.6087,
+      "step": 3500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.785631659056317e-05,
+      "loss": 2.4974,
+      "step": 3525
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.784109589041096e-05,
+      "loss": 2.4709,
+      "step": 3550
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.7825875190258757e-05,
+      "loss": 2.5459,
+      "step": 3575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.7810654490106546e-05,
+      "loss": 2.5447,
+      "step": 3600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.779543378995434e-05,
+      "loss": 2.5014,
+      "step": 3625
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.778021308980214e-05,
+      "loss": 2.56,
+      "step": 3650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.776499238964993e-05,
+      "loss": 2.5625,
+      "step": 3675
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.774977168949772e-05,
+      "loss": 2.5072,
+      "step": 3700
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.7734550989345514e-05,
+      "loss": 2.5928,
+      "step": 3725
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.77193302891933e-05,
+      "loss": 2.4591,
+      "step": 3750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.77041095890411e-05,
+      "loss": 2.3873,
+      "step": 3775
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.768888888888889e-05,
+      "loss": 2.468,
+      "step": 3800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.7673668188736685e-05,
+      "loss": 2.5052,
+      "step": 3825
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.765844748858448e-05,
+      "loss": 2.5028,
+      "step": 3850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.764322678843227e-05,
+      "loss": 2.516,
+      "step": 3875
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.762800608828007e-05,
+      "loss": 2.514,
+      "step": 3900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.7612785388127856e-05,
+      "loss": 2.5943,
+      "step": 3925
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.759756468797565e-05,
+      "loss": 2.4263,
+      "step": 3950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.758234398782344e-05,
+      "loss": 2.3991,
+      "step": 3975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.756712328767123e-05,
+      "loss": 2.3838,
+      "step": 4000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.755190258751903e-05,
+      "loss": 2.4018,
+      "step": 4025
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7536681887366824e-05,
+      "loss": 2.4734,
+      "step": 4050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.752146118721461e-05,
+      "loss": 2.4167,
+      "step": 4075
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.750624048706241e-05,
+      "loss": 2.3863,
+      "step": 4100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.74910197869102e-05,
+      "loss": 2.5197,
+      "step": 4125
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7475799086757995e-05,
+      "loss": 2.4539,
+      "step": 4150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7460578386605785e-05,
+      "loss": 2.4153,
+      "step": 4175
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.744535768645358e-05,
+      "loss": 2.5164,
+      "step": 4200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.743013698630137e-05,
+      "loss": 2.4194,
+      "step": 4225
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7414916286149167e-05,
+      "loss": 2.4861,
+      "step": 4250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.739969558599696e-05,
+      "loss": 2.4588,
+      "step": 4275
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.738447488584475e-05,
+      "loss": 2.5212,
+      "step": 4300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.736925418569254e-05,
+      "loss": 2.3907,
+      "step": 4325
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.735403348554034e-05,
+      "loss": 2.4816,
+      "step": 4350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.733881278538813e-05,
+      "loss": 2.5161,
+      "step": 4375
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7323592085235924e-05,
+      "loss": 2.5058,
+      "step": 4400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.730837138508371e-05,
+      "loss": 2.4269,
+      "step": 4425
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.729315068493151e-05,
+      "loss": 2.5356,
+      "step": 4450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.7277929984779306e-05,
+      "loss": 2.509,
+      "step": 4475
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.7262709284627095e-05,
+      "loss": 2.397,
+      "step": 4500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.724748858447489e-05,
+      "loss": 2.5031,
+      "step": 4525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.723226788432268e-05,
+      "loss": 2.4124,
+      "step": 4550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.721704718417048e-05,
+      "loss": 2.5102,
+      "step": 4575
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.7201826484018266e-05,
+      "loss": 2.486,
+      "step": 4600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.7186605783866056e-05,
+      "loss": 2.407,
+      "step": 4625
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.717138508371386e-05,
+      "loss": 2.464,
+      "step": 4650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.715616438356165e-05,
+      "loss": 2.5452,
+      "step": 4675
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.714094368340944e-05,
+      "loss": 2.4468,
+      "step": 4700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7125722983257234e-05,
+      "loss": 2.4068,
+      "step": 4725
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.711050228310502e-05,
+      "loss": 2.4681,
+      "step": 4750
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.709528158295282e-05,
+      "loss": 2.3566,
+      "step": 4775
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.70806697108067e-05,
+      "loss": 2.4723,
+      "step": 4800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.7065449010654495e-05,
+      "loss": 2.4191,
+      "step": 4825
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.7050228310502285e-05,
+      "loss": 2.5294,
+      "step": 4850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.703500761035008e-05,
+      "loss": 2.7049,
+      "step": 4875
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.701978691019787e-05,
+      "loss": 2.5428,
+      "step": 4900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.700456621004567e-05,
+      "loss": 2.4997,
+      "step": 4925
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.6989345509893456e-05,
+      "loss": 2.3452,
+      "step": 4950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.697412480974125e-05,
+      "loss": 2.5259,
+      "step": 4975
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.695890410958904e-05,
+      "loss": 2.3965,
+      "step": 5000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.694368340943684e-05,
+      "loss": 2.501,
+      "step": 5025
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.692846270928463e-05,
+      "loss": 2.4793,
+      "step": 5050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.6913242009132424e-05,
+      "loss": 2.5491,
+      "step": 5075
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.689802130898021e-05,
+      "loss": 2.4435,
+      "step": 5100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.688280060882801e-05,
+      "loss": 2.4705,
+      "step": 5125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.6867579908675806e-05,
+      "loss": 2.475,
+      "step": 5150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6852359208523595e-05,
+      "loss": 2.5521,
+      "step": 5175
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.683713850837139e-05,
+      "loss": 2.5303,
+      "step": 5200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.682191780821918e-05,
+      "loss": 2.4951,
+      "step": 5225
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.680669710806697e-05,
+      "loss": 2.4559,
+      "step": 5250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6791476407914766e-05,
+      "loss": 2.5772,
+      "step": 5275
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6776255707762556e-05,
+      "loss": 2.4943,
+      "step": 5300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.676164383561644e-05,
+      "loss": 2.4351,
+      "step": 5325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.674642313546423e-05,
+      "loss": 2.5454,
+      "step": 5350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.673120243531203e-05,
+      "loss": 2.492,
+      "step": 5375
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6715981735159824e-05,
+      "loss": 2.5228,
+      "step": 5400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.6700761035007614e-05,
+      "loss": 2.4511,
+      "step": 5425
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.668554033485541e-05,
+      "loss": 2.4755,
+      "step": 5450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.66703196347032e-05,
+      "loss": 2.4568,
+      "step": 5475
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6655098934550995e-05,
+      "loss": 2.4765,
+      "step": 5500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.6639878234398785e-05,
+      "loss": 2.4977,
+      "step": 5525
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.6624657534246574e-05,
+      "loss": 2.5427,
+      "step": 5550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.660943683409437e-05,
+      "loss": 2.6176,
+      "step": 5575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.659421613394217e-05,
+      "loss": 2.5213,
+      "step": 5600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.6578995433789956e-05,
+      "loss": 2.5476,
+      "step": 5625
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.656377473363775e-05,
+      "loss": 2.4277,
+      "step": 5650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.654855403348554e-05,
+      "loss": 2.5481,
+      "step": 5675
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.653333333333334e-05,
+      "loss": 2.3107,
+      "step": 5700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.651811263318113e-05,
+      "loss": 2.4638,
+      "step": 5725
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.6502891933028924e-05,
+      "loss": 2.4764,
+      "step": 5750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.648767123287671e-05,
+      "loss": 2.5169,
+      "step": 5775
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.647245053272451e-05,
+      "loss": 2.5419,
+      "step": 5800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.6457229832572306e-05,
+      "loss": 2.469,
+      "step": 5825
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.6442009132420095e-05,
+      "loss": 2.5355,
+      "step": 5850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.6426788432267885e-05,
+      "loss": 2.4811,
+      "step": 5875
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.641156773211568e-05,
+      "loss": 2.3855,
+      "step": 5900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.639634703196347e-05,
+      "loss": 2.5192,
+      "step": 5925
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6381126331811267e-05,
+      "loss": 2.5408,
+      "step": 5950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6365905631659056e-05,
+      "loss": 2.4392,
+      "step": 5975
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.635068493150685e-05,
+      "loss": 2.33,
+      "step": 6000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.633546423135465e-05,
+      "loss": 2.3823,
+      "step": 6025
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.632024353120244e-05,
+      "loss": 2.5043,
+      "step": 6050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6305022831050234e-05,
+      "loss": 2.5216,
+      "step": 6075
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6289802130898024e-05,
+      "loss": 2.5068,
+      "step": 6100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.627458143074582e-05,
+      "loss": 2.4264,
+      "step": 6125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.625936073059361e-05,
+      "loss": 2.3439,
+      "step": 6150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.62441400304414e-05,
+      "loss": 2.467,
+      "step": 6175
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.62289193302892e-05,
+      "loss": 2.4853,
+      "step": 6200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.621369863013699e-05,
+      "loss": 2.4064,
+      "step": 6225
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.619847792998478e-05,
+      "loss": 2.4467,
+      "step": 6250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.618325722983258e-05,
+      "loss": 2.4435,
+      "step": 6275
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.6168036529680366e-05,
+      "loss": 2.4629,
+      "step": 6300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.615281582952816e-05,
+      "loss": 2.5577,
+      "step": 6325
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.613759512937595e-05,
+      "loss": 2.5897,
+      "step": 6350
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.612237442922375e-05,
+      "loss": 2.4605,
+      "step": 6375
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6107153729071544e-05,
+      "loss": 2.4198,
+      "step": 6400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.6091933028919334e-05,
+      "loss": 2.5914,
+      "step": 6425
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.607671232876713e-05,
+      "loss": 2.4329,
+      "step": 6450
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.606149162861492e-05,
+      "loss": 2.3896,
+      "step": 6475
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.604627092846271e-05,
+      "loss": 2.4753,
+      "step": 6500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.6031050228310505e-05,
+      "loss": 2.4698,
+      "step": 6525
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.6015829528158295e-05,
+      "loss": 2.4075,
+      "step": 6550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.600060882800609e-05,
+      "loss": 2.4441,
+      "step": 6575
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.598538812785389e-05,
+      "loss": 2.3363,
+      "step": 6600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.597016742770168e-05,
+      "loss": 2.3688,
+      "step": 6625
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.595494672754947e-05,
+      "loss": 2.362,
+      "step": 6650
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.593972602739726e-05,
+      "loss": 2.4022,
+      "step": 6675
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.592450532724506e-05,
+      "loss": 2.295,
+      "step": 6700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.590928462709285e-05,
+      "loss": 2.4906,
+      "step": 6725
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.5894063926940644e-05,
+      "loss": 2.3863,
+      "step": 6750
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.5878843226788434e-05,
+      "loss": 2.4918,
+      "step": 6775
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.586362252663623e-05,
+      "loss": 2.3376,
+      "step": 6800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.5848401826484026e-05,
+      "loss": 2.4073,
+      "step": 6825
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.5833181126331816e-05,
+      "loss": 2.4083,
+      "step": 6850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.5817960426179605e-05,
+      "loss": 2.3808,
+      "step": 6875
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.58027397260274e-05,
+      "loss": 2.389,
+      "step": 6900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.578751902587519e-05,
+      "loss": 2.4459,
+      "step": 6925
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.577229832572299e-05,
+      "loss": 2.3535,
+      "step": 6950
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.5757077625570776e-05,
+      "loss": 2.3977,
+      "step": 6975
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.574185692541857e-05,
+      "loss": 2.4717,
+      "step": 7000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.572663622526637e-05,
+      "loss": 2.4242,
+      "step": 7025
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.571141552511416e-05,
+      "loss": 2.5278,
+      "step": 7050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.5696194824961955e-05,
+      "loss": 2.3942,
+      "step": 7075
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.5680974124809744e-05,
+      "loss": 2.4924,
+      "step": 7100
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5665753424657533e-05,
+      "loss": 2.4798,
+      "step": 7125
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.565053272450533e-05,
+      "loss": 2.4777,
+      "step": 7150
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.563531202435312e-05,
+      "loss": 2.4038,
+      "step": 7175
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5620091324200915e-05,
+      "loss": 2.4612,
+      "step": 7200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.560487062404871e-05,
+      "loss": 2.4008,
+      "step": 7225
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.55896499238965e-05,
+      "loss": 2.4106,
+      "step": 7250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.55744292237443e-05,
+      "loss": 2.3472,
+      "step": 7275
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.555920852359209e-05,
+      "loss": 2.3121,
+      "step": 7300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.554398782343988e-05,
+      "loss": 2.3585,
+      "step": 7325
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.552876712328767e-05,
+      "loss": 2.421,
+      "step": 7350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.551354642313547e-05,
+      "loss": 2.3866,
+      "step": 7375
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5498325722983265e-05,
+      "loss": 2.3897,
+      "step": 7400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.5483105022831054e-05,
+      "loss": 2.3006,
+      "step": 7425
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.546788432267885e-05,
+      "loss": 2.4119,
+      "step": 7450
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.545266362252664e-05,
+      "loss": 2.3586,
+      "step": 7475
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.543744292237443e-05,
+      "loss": 2.4986,
+      "step": 7500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.5422222222222226e-05,
+      "loss": 2.401,
+      "step": 7525
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5407001522070015e-05,
+      "loss": 2.4442,
+      "step": 7550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.539178082191781e-05,
+      "loss": 2.4115,
+      "step": 7575
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.537656012176561e-05,
+      "loss": 2.3774,
+      "step": 7600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.53613394216134e-05,
+      "loss": 2.4354,
+      "step": 7625
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.534611872146119e-05,
+      "loss": 2.4084,
+      "step": 7650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.533089802130898e-05,
+      "loss": 2.3302,
+      "step": 7675
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.531567732115678e-05,
+      "loss": 2.4794,
+      "step": 7700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.530045662100457e-05,
+      "loss": 2.5226,
+      "step": 7725
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.528523592085236e-05,
+      "loss": 2.4108,
+      "step": 7750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5270015220700154e-05,
+      "loss": 2.3983,
+      "step": 7775
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.525479452054795e-05,
+      "loss": 2.355,
+      "step": 7800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.523957382039574e-05,
+      "loss": 2.4036,
+      "step": 7825
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5224353120243536e-05,
+      "loss": 2.3806,
+      "step": 7850
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5209132420091325e-05,
+      "loss": 2.397,
+      "step": 7875
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.519391171993912e-05,
+      "loss": 2.4013,
+      "step": 7900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.517869101978691e-05,
+      "loss": 2.3221,
+      "step": 7925
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.516347031963471e-05,
+      "loss": 2.4978,
+      "step": 7950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.51482496194825e-05,
+      "loss": 2.3336,
+      "step": 7975
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.513302891933029e-05,
+      "loss": 2.3407,
+      "step": 8000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.511780821917809e-05,
+      "loss": 2.3885,
+      "step": 8025
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.510258751902588e-05,
+      "loss": 2.3823,
+      "step": 8050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5087366818873675e-05,
+      "loss": 2.3793,
+      "step": 8075
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5072146118721464e-05,
+      "loss": 2.5151,
+      "step": 8100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5056925418569254e-05,
+      "loss": 2.4551,
+      "step": 8125
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.504170471841705e-05,
+      "loss": 2.2755,
+      "step": 8150
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.502648401826484e-05,
+      "loss": 2.3501,
+      "step": 8175
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5011263318112636e-05,
+      "loss": 2.1867,
+      "step": 8200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.499604261796043e-05,
+      "loss": 2.4121,
+      "step": 8225
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.498082191780822e-05,
+      "loss": 2.3982,
+      "step": 8250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.496560121765602e-05,
+      "loss": 2.3759,
+      "step": 8275
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.495038051750381e-05,
+      "loss": 2.392,
+      "step": 8300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.49351598173516e-05,
+      "loss": 2.3885,
+      "step": 8325
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.491993911719939e-05,
+      "loss": 2.385,
+      "step": 8350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.490471841704718e-05,
+      "loss": 2.3473,
+      "step": 8375
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4889497716894985e-05,
+      "loss": 2.3964,
+      "step": 8400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4874277016742775e-05,
+      "loss": 2.4064,
+      "step": 8425
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4859056316590564e-05,
+      "loss": 2.418,
+      "step": 8450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.484383561643836e-05,
+      "loss": 2.3923,
+      "step": 8475
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.482861491628615e-05,
+      "loss": 2.3131,
+      "step": 8500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4813394216133946e-05,
+      "loss": 2.28,
+      "step": 8525
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4798173515981735e-05,
+      "loss": 2.2658,
+      "step": 8550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.478295281582953e-05,
+      "loss": 2.3469,
+      "step": 8575
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.476773211567733e-05,
+      "loss": 2.3742,
+      "step": 8600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.475251141552512e-05,
+      "loss": 2.2876,
+      "step": 8625
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4737290715372914e-05,
+      "loss": 2.3574,
+      "step": 8650
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.47220700152207e-05,
+      "loss": 2.3034,
+      "step": 8675
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.47068493150685e-05,
+      "loss": 2.4737,
+      "step": 8700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.469162861491629e-05,
+      "loss": 2.4021,
+      "step": 8725
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.467640791476408e-05,
+      "loss": 2.3528,
+      "step": 8750
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4661187214611874e-05,
+      "loss": 2.3844,
+      "step": 8775
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.464596651445967e-05,
+      "loss": 2.3548,
+      "step": 8800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.463074581430746e-05,
+      "loss": 2.3506,
+      "step": 8825
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4615525114155256e-05,
+      "loss": 2.2949,
+      "step": 8850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4600304414003046e-05,
+      "loss": 2.4181,
+      "step": 8875
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.458508371385084e-05,
+      "loss": 2.3763,
+      "step": 8900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.456986301369863e-05,
+      "loss": 2.3273,
+      "step": 8925
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.455464231354643e-05,
+      "loss": 2.2915,
+      "step": 8950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.453942161339422e-05,
+      "loss": 2.2817,
+      "step": 8975
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.452420091324201e-05,
+      "loss": 2.4281,
+      "step": 9000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.450898021308981e-05,
+      "loss": 2.3455,
+      "step": 9025
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.44937595129376e-05,
+      "loss": 2.3443,
+      "step": 9050
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.447853881278539e-05,
+      "loss": 2.3764,
+      "step": 9075
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.4463318112633185e-05,
+      "loss": 2.3281,
+      "step": 9100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.4448097412480974e-05,
+      "loss": 2.3462,
+      "step": 9125
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.443287671232877e-05,
+      "loss": 2.3027,
+      "step": 9150
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.441765601217656e-05,
+      "loss": 2.3474,
+      "step": 9175
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4402435312024356e-05,
+      "loss": 2.4758,
+      "step": 9200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.438721461187215e-05,
+      "loss": 2.4488,
+      "step": 9225
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.437199391171994e-05,
+      "loss": 2.2771,
+      "step": 9250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.435677321156774e-05,
+      "loss": 2.3559,
+      "step": 9275
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.434155251141553e-05,
+      "loss": 2.4645,
+      "step": 9300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4326331811263324e-05,
+      "loss": 2.3265,
+      "step": 9325
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.431111111111111e-05,
+      "loss": 2.3063,
+      "step": 9350
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.42958904109589e-05,
+      "loss": 2.3427,
+      "step": 9375
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.42806697108067e-05,
+      "loss": 2.372,
+      "step": 9400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4265449010654495e-05,
+      "loss": 2.4457,
+      "step": 9425
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4250228310502284e-05,
+      "loss": 2.4012,
+      "step": 9450
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.423500761035008e-05,
+      "loss": 2.4501,
+      "step": 9475
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.421978691019787e-05,
+      "loss": 2.4032,
+      "step": 9500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.4204566210045666e-05,
+      "loss": 2.2623,
+      "step": 9525
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.4189345509893456e-05,
+      "loss": 2.3914,
+      "step": 9550
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.417412480974125e-05,
+      "loss": 2.2566,
+      "step": 9575
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.415890410958904e-05,
+      "loss": 2.3409,
+      "step": 9600
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.414368340943684e-05,
+      "loss": 2.4972,
+      "step": 9625
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4128462709284634e-05,
+      "loss": 2.4752,
+      "step": 9650
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4113242009132423e-05,
+      "loss": 2.3794,
+      "step": 9675
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.40986301369863e-05,
+      "loss": 2.3816,
+      "step": 9700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.40834094368341e-05,
+      "loss": 2.3872,
+      "step": 9725
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.406818873668189e-05,
+      "loss": 2.3142,
+      "step": 9750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4052968036529685e-05,
+      "loss": 2.3174,
+      "step": 9775
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4037747336377474e-05,
+      "loss": 2.2765,
+      "step": 9800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.402252663622527e-05,
+      "loss": 2.3761,
+      "step": 9825
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.400730593607306e-05,
+      "loss": 2.2079,
+      "step": 9850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.3992085235920856e-05,
+      "loss": 2.4002,
+      "step": 9875
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.397686453576865e-05,
+      "loss": 2.3128,
+      "step": 9900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.396164383561644e-05,
+      "loss": 2.3319,
+      "step": 9925
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.394642313546424e-05,
+      "loss": 2.3102,
+      "step": 9950
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.393120243531203e-05,
+      "loss": 2.3264,
+      "step": 9975
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.391598173515982e-05,
+      "loss": 2.285,
+      "step": 10000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.390076103500761e-05,
+      "loss": 2.3575,
+      "step": 10025
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.38855403348554e-05,
+      "loss": 2.3238,
+      "step": 10050
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.38703196347032e-05,
+      "loss": 2.3473,
+      "step": 10075
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.3855098934550995e-05,
+      "loss": 2.3622,
+      "step": 10100
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3839878234398785e-05,
+      "loss": 2.3262,
+      "step": 10125
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.382465753424658e-05,
+      "loss": 2.2765,
+      "step": 10150
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.380943683409437e-05,
+      "loss": 2.3726,
+      "step": 10175
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.3794216133942167e-05,
+      "loss": 2.3951,
+      "step": 10200
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.3778995433789956e-05,
+      "loss": 2.2942,
+      "step": 10225
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.3763774733637745e-05,
+      "loss": 2.2152,
+      "step": 10250
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.374855403348554e-05,
+      "loss": 2.2103,
+      "step": 10275
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.373333333333334e-05,
+      "loss": 2.3017,
+      "step": 10300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.371811263318113e-05,
+      "loss": 2.34,
+      "step": 10325
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.3702891933028924e-05,
+      "loss": 2.411,
+      "step": 10350
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.368767123287671e-05,
+      "loss": 2.3578,
+      "step": 10375
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.367245053272451e-05,
+      "loss": 2.3459,
+      "step": 10400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.36572298325723e-05,
+      "loss": 2.37,
+      "step": 10425
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.3642009132420095e-05,
+      "loss": 2.2468,
+      "step": 10450
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.3626788432267884e-05,
+      "loss": 2.347,
+      "step": 10475
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.361156773211568e-05,
+      "loss": 2.2949,
+      "step": 10500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.359634703196348e-05,
+      "loss": 2.2398,
+      "step": 10525
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3581126331811266e-05,
+      "loss": 2.3733,
+      "step": 10550
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.356590563165906e-05,
+      "loss": 2.3245,
+      "step": 10575
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.355068493150685e-05,
+      "loss": 2.3472,
+      "step": 10600
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.353546423135464e-05,
+      "loss": 2.2422,
+      "step": 10625
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.352024353120244e-05,
+      "loss": 2.2771,
+      "step": 10650
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.350502283105023e-05,
+      "loss": 2.3022,
+      "step": 10675
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.348980213089802e-05,
+      "loss": 2.363,
+      "step": 10700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.347458143074582e-05,
+      "loss": 2.3812,
+      "step": 10725
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.345936073059361e-05,
+      "loss": 2.3246,
+      "step": 10750
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3444140030441405e-05,
+      "loss": 2.2674,
+      "step": 10775
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3428919330289195e-05,
+      "loss": 2.2517,
+      "step": 10800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.341369863013699e-05,
+      "loss": 2.3662,
+      "step": 10825
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.339847792998478e-05,
+      "loss": 2.3391,
+      "step": 10850
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.338325722983257e-05,
+      "loss": 2.3473,
+      "step": 10875
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.336803652968037e-05,
+      "loss": 2.34,
+      "step": 10900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.335281582952816e-05,
+      "loss": 2.3697,
+      "step": 10925
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.333759512937595e-05,
+      "loss": 2.3801,
+      "step": 10950
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.332237442922375e-05,
+      "loss": 2.2578,
+      "step": 10975
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.330715372907154e-05,
+      "loss": 2.287,
+      "step": 11000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3291933028919334e-05,
+      "loss": 2.3065,
+      "step": 11025
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.327671232876712e-05,
+      "loss": 2.3242,
+      "step": 11050
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.326149162861492e-05,
+      "loss": 2.381,
+      "step": 11075
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3246270928462716e-05,
+      "loss": 2.3044,
+      "step": 11100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3231050228310505e-05,
+      "loss": 2.3529,
+      "step": 11125
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.32158295281583e-05,
+      "loss": 2.3205,
+      "step": 11150
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.320060882800609e-05,
+      "loss": 2.2909,
+      "step": 11175
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.318538812785389e-05,
+      "loss": 2.2733,
+      "step": 11200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.3170167427701676e-05,
+      "loss": 2.3524,
+      "step": 11225
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.3154946727549466e-05,
+      "loss": 2.4237,
+      "step": 11250
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.313972602739726e-05,
+      "loss": 2.3506,
+      "step": 11275
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.312450532724506e-05,
+      "loss": 2.239,
+      "step": 11300
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.310928462709285e-05,
+      "loss": 2.2446,
+      "step": 11325
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.3094063926940644e-05,
+      "loss": 2.4092,
+      "step": 11350
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.307884322678843e-05,
+      "loss": 2.2258,
+      "step": 11375
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.306362252663623e-05,
+      "loss": 2.3767,
+      "step": 11400
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.304840182648402e-05,
+      "loss": 2.3656,
+      "step": 11425
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.3033181126331815e-05,
+      "loss": 2.3565,
+      "step": 11450
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.3017960426179605e-05,
+      "loss": 2.3451,
+      "step": 11475
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.30027397260274e-05,
+      "loss": 2.3693,
+      "step": 11500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.29875190258752e-05,
+      "loss": 2.2994,
+      "step": 11525
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.297229832572299e-05,
+      "loss": 2.3294,
+      "step": 11550
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.2957077625570776e-05,
+      "loss": 2.3281,
+      "step": 11575
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.294185692541857e-05,
+      "loss": 2.2991,
+      "step": 11600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.292663622526636e-05,
+      "loss": 2.4062,
+      "step": 11625
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.291141552511416e-05,
+      "loss": 2.2678,
+      "step": 11650
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.289619482496195e-05,
+      "loss": 2.3024,
+      "step": 11675
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2880974124809744e-05,
+      "loss": 2.2003,
+      "step": 11700
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.286575342465754e-05,
+      "loss": 2.3133,
+      "step": 11725
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.285053272450533e-05,
+      "loss": 2.347,
+      "step": 11750
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2835312024353126e-05,
+      "loss": 2.2283,
+      "step": 11775
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2820091324200915e-05,
+      "loss": 2.293,
+      "step": 11800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.280487062404871e-05,
+      "loss": 2.3133,
+      "step": 11825
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.27896499238965e-05,
+      "loss": 2.306,
+      "step": 11850
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.277442922374429e-05,
+      "loss": 2.3111,
+      "step": 11875
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.275920852359209e-05,
+      "loss": 2.2691,
+      "step": 11900
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.274398782343988e-05,
+      "loss": 2.4566,
+      "step": 11925
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.272876712328767e-05,
+      "loss": 2.3288,
+      "step": 11950
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.271354642313547e-05,
+      "loss": 2.2988,
+      "step": 11975
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.269832572298326e-05,
+      "loss": 2.2849,
+      "step": 12000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.2683105022831054e-05,
+      "loss": 2.2855,
+      "step": 12025
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.2667884322678843e-05,
+      "loss": 2.3684,
+      "step": 12050
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.265266362252664e-05,
+      "loss": 2.2872,
+      "step": 12075
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2637442922374436e-05,
+      "loss": 2.2523,
+      "step": 12100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.2622222222222225e-05,
+      "loss": 2.1726,
+      "step": 12125
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.260700152207002e-05,
+      "loss": 2.4131,
+      "step": 12150
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.259178082191781e-05,
+      "loss": 2.1719,
+      "step": 12175
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.25765601217656e-05,
+      "loss": 2.3665,
+      "step": 12200
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.25613394216134e-05,
+      "loss": 2.3097,
+      "step": 12225
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2546118721461186e-05,
+      "loss": 2.3808,
+      "step": 12250
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.253089802130898e-05,
+      "loss": 2.3309,
+      "step": 12275
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.251567732115678e-05,
+      "loss": 2.2463,
+      "step": 12300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.250045662100457e-05,
+      "loss": 2.2689,
+      "step": 12325
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2485235920852364e-05,
+      "loss": 2.301,
+      "step": 12350
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2470015220700154e-05,
+      "loss": 2.2508,
+      "step": 12375
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.245479452054795e-05,
+      "loss": 2.2932,
+      "step": 12400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.243957382039574e-05,
+      "loss": 2.2948,
+      "step": 12425
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.2424353120243536e-05,
+      "loss": 2.3646,
+      "step": 12450
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.2409132420091325e-05,
+      "loss": 2.279,
+      "step": 12475
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.239391171993912e-05,
+      "loss": 2.2907,
+      "step": 12500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.2379299847793e-05,
+      "loss": 2.2119,
+      "step": 12525
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.236407914764079e-05,
+      "loss": 2.24,
+      "step": 12550
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.2348858447488587e-05,
+      "loss": 2.3259,
+      "step": 12575
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.233363774733638e-05,
+      "loss": 2.2823,
+      "step": 12600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.231841704718417e-05,
+      "loss": 2.3152,
+      "step": 12625
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.230319634703197e-05,
+      "loss": 2.2235,
+      "step": 12650
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.228797564687976e-05,
+      "loss": 2.2921,
+      "step": 12675
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.2272754946727554e-05,
+      "loss": 2.2987,
+      "step": 12700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.2257534246575344e-05,
+      "loss": 2.2525,
+      "step": 12725
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.224231354642313e-05,
+      "loss": 2.31,
+      "step": 12750
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.2227092846270936e-05,
+      "loss": 2.2679,
+      "step": 12775
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.2211872146118725e-05,
+      "loss": 2.3119,
+      "step": 12800
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.2196651445966515e-05,
+      "loss": 2.2647,
+      "step": 12825
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.218143074581431e-05,
+      "loss": 2.2724,
+      "step": 12850
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.21662100456621e-05,
+      "loss": 2.1599,
+      "step": 12875
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.21509893455099e-05,
+      "loss": 2.307,
+      "step": 12900
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.2135768645357686e-05,
+      "loss": 2.282,
+      "step": 12925
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.212054794520548e-05,
+      "loss": 2.2976,
+      "step": 12950
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.210532724505328e-05,
+      "loss": 2.2577,
+      "step": 12975
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.209010654490107e-05,
+      "loss": 2.3745,
+      "step": 13000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.2074885844748864e-05,
+      "loss": 2.3284,
+      "step": 13025
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.2059665144596654e-05,
+      "loss": 2.2928,
+      "step": 13050
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.204444444444445e-05,
+      "loss": 2.3184,
+      "step": 13075
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.202922374429224e-05,
+      "loss": 2.377,
+      "step": 13100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.201400304414003e-05,
+      "loss": 2.2414,
+      "step": 13125
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.1998782343987825e-05,
+      "loss": 2.2791,
+      "step": 13150
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.198356164383562e-05,
+      "loss": 2.2645,
+      "step": 13175
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.196834094368341e-05,
+      "loss": 2.2783,
+      "step": 13200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.195312024353121e-05,
+      "loss": 2.2995,
+      "step": 13225
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.1937899543378997e-05,
+      "loss": 2.3162,
+      "step": 13250
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.192267884322679e-05,
+      "loss": 2.273,
+      "step": 13275
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.190745814307458e-05,
+      "loss": 2.1421,
+      "step": 13300
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.189223744292238e-05,
+      "loss": 2.2324,
+      "step": 13325
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.187701674277017e-05,
+      "loss": 2.148,
+      "step": 13350
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.1861796042617964e-05,
+      "loss": 2.2015,
+      "step": 13375
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.184657534246576e-05,
+      "loss": 2.2374,
+      "step": 13400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.183135464231355e-05,
+      "loss": 2.172,
+      "step": 13425
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.181613394216134e-05,
+      "loss": 2.2591,
+      "step": 13450
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.1800913242009136e-05,
+      "loss": 2.3048,
+      "step": 13475
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.1785692541856925e-05,
+      "loss": 2.2699,
+      "step": 13500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.177047184170472e-05,
+      "loss": 2.2023,
+      "step": 13525
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.175525114155251e-05,
+      "loss": 2.2619,
+      "step": 13550
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.174003044140031e-05,
+      "loss": 2.2799,
+      "step": 13575
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.17248097412481e-05,
+      "loss": 2.2879,
+      "step": 13600
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.170958904109589e-05,
+      "loss": 2.2305,
+      "step": 13625
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.169436834094369e-05,
+      "loss": 2.2701,
+      "step": 13650
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.167914764079148e-05,
+      "loss": 2.1813,
+      "step": 13675
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.1663926940639274e-05,
+      "loss": 2.239,
+      "step": 13700
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.1648706240487064e-05,
+      "loss": 2.1826,
+      "step": 13725
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.1633485540334853e-05,
+      "loss": 2.2252,
+      "step": 13750
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.1618264840182656e-05,
+      "loss": 2.1614,
+      "step": 13775
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1603044140030446e-05,
+      "loss": 2.2616,
+      "step": 13800
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1587823439878235e-05,
+      "loss": 2.174,
+      "step": 13825
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.157260273972603e-05,
+      "loss": 2.176,
+      "step": 13850
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.155738203957382e-05,
+      "loss": 2.3032,
+      "step": 13875
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.154216133942162e-05,
+      "loss": 2.3267,
+      "step": 13900
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.152694063926941e-05,
+      "loss": 2.2344,
+      "step": 13925
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.15117199391172e-05,
+      "loss": 2.2329,
+      "step": 13950
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.1496499238965e-05,
+      "loss": 2.2776,
+      "step": 13975
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.148127853881279e-05,
+      "loss": 2.2616,
+      "step": 14000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.1466057838660585e-05,
+      "loss": 2.2726,
+      "step": 14025
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.1450837138508374e-05,
+      "loss": 2.2973,
+      "step": 14050
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.1435616438356164e-05,
+      "loss": 2.4038,
+      "step": 14075
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.142039573820396e-05,
+      "loss": 2.1422,
+      "step": 14100
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.140517503805175e-05,
+      "loss": 2.2385,
+      "step": 14125
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.1389954337899546e-05,
+      "loss": 2.3466,
+      "step": 14150
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.137473363774734e-05,
+      "loss": 2.3624,
+      "step": 14175
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.135951293759513e-05,
+      "loss": 2.1896,
+      "step": 14200
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.134429223744293e-05,
+      "loss": 2.2879,
+      "step": 14225
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.132907153729072e-05,
+      "loss": 2.2648,
+      "step": 14250
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.131385083713851e-05,
+      "loss": 2.1587,
+      "step": 14275
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.12986301369863e-05,
+      "loss": 2.2501,
+      "step": 14300
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.12834094368341e-05,
+      "loss": 2.3226,
+      "step": 14325
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.126818873668189e-05,
+      "loss": 2.2835,
+      "step": 14350
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.1252968036529685e-05,
+      "loss": 2.1606,
+      "step": 14375
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.123774733637748e-05,
+      "loss": 2.3472,
+      "step": 14400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.122252663622527e-05,
+      "loss": 2.3695,
+      "step": 14425
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.120730593607306e-05,
+      "loss": 2.3861,
+      "step": 14450
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.1192085235920856e-05,
+      "loss": 2.169,
+      "step": 14475
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.1176864535768645e-05,
+      "loss": 2.2452,
+      "step": 14500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.116164383561644e-05,
+      "loss": 2.3446,
+      "step": 14525
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.114642313546423e-05,
+      "loss": 2.265,
+      "step": 14550
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.113120243531203e-05,
+      "loss": 2.3011,
+      "step": 14575
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.1115981735159824e-05,
+      "loss": 2.1554,
+      "step": 14600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.110076103500761e-05,
+      "loss": 2.3185,
+      "step": 14625
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.108554033485541e-05,
+      "loss": 2.1786,
+      "step": 14650
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.10703196347032e-05,
+      "loss": 2.3914,
+      "step": 14675
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.105509893455099e-05,
+      "loss": 2.2981,
+      "step": 14700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1039878234398784e-05,
+      "loss": 2.1446,
+      "step": 14725
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1024657534246574e-05,
+      "loss": 2.2635,
+      "step": 14750
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.100943683409437e-05,
+      "loss": 2.3167,
+      "step": 14775
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0994216133942166e-05,
+      "loss": 2.161,
+      "step": 14800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0978995433789956e-05,
+      "loss": 2.2632,
+      "step": 14825
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.096377473363775e-05,
+      "loss": 2.2193,
+      "step": 14850
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.094855403348554e-05,
+      "loss": 2.2079,
+      "step": 14875
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.093333333333334e-05,
+      "loss": 2.2739,
+      "step": 14900
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.091811263318113e-05,
+      "loss": 2.2497,
+      "step": 14925
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.090289193302892e-05,
+      "loss": 2.3314,
+      "step": 14950
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.088767123287672e-05,
+      "loss": 2.14,
+      "step": 14975
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.087245053272451e-05,
+      "loss": 2.2944,
+      "step": 15000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.0857229832572305e-05,
+      "loss": 2.2942,
+      "step": 15025
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.0842009132420095e-05,
+      "loss": 2.2938,
+      "step": 15050
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.0826788432267884e-05,
+      "loss": 2.2439,
+      "step": 15075
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.081156773211568e-05,
+      "loss": 2.3085,
+      "step": 15100
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.079634703196347e-05,
+      "loss": 2.2249,
+      "step": 15125
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0781126331811266e-05,
+      "loss": 2.2429,
+      "step": 15150
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.076590563165906e-05,
+      "loss": 2.2207,
+      "step": 15175
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.075068493150685e-05,
+      "loss": 2.2028,
+      "step": 15200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.073546423135465e-05,
+      "loss": 2.1323,
+      "step": 15225
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.072024353120244e-05,
+      "loss": 2.2831,
+      "step": 15250
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.0705022831050234e-05,
+      "loss": 2.2963,
+      "step": 15275
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.068980213089802e-05,
+      "loss": 2.29,
+      "step": 15300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.067458143074581e-05,
+      "loss": 2.2412,
+      "step": 15325
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.065936073059361e-05,
+      "loss": 2.3399,
+      "step": 15350
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0644140030441405e-05,
+      "loss": 2.1743,
+      "step": 15375
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.0628919330289194e-05,
+      "loss": 2.2421,
+      "step": 15400
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.061369863013699e-05,
+      "loss": 2.337,
+      "step": 15425
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.059847792998478e-05,
+      "loss": 2.3268,
+      "step": 15450
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.0583257229832576e-05,
+      "loss": 2.2685,
+      "step": 15475
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.0568036529680366e-05,
+      "loss": 2.2797,
+      "step": 15500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.055281582952816e-05,
+      "loss": 2.27,
+      "step": 15525
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.053759512937595e-05,
+      "loss": 2.2245,
+      "step": 15550
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.052237442922375e-05,
+      "loss": 2.25,
+      "step": 15575
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.050715372907154e-05,
+      "loss": 2.1833,
+      "step": 15600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0491933028919333e-05,
+      "loss": 2.2166,
+      "step": 15625
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0476712328767126e-05,
+      "loss": 2.3001,
+      "step": 15650
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.046149162861492e-05,
+      "loss": 2.2848,
+      "step": 15675
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.0446270928462712e-05,
+      "loss": 2.1487,
+      "step": 15700
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.0431050228310505e-05,
+      "loss": 2.1833,
+      "step": 15725
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.0415829528158298e-05,
+      "loss": 2.1392,
+      "step": 15750
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.0400608828006094e-05,
+      "loss": 2.2441,
+      "step": 15775
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.0385388127853887e-05,
+      "loss": 2.2952,
+      "step": 15800
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.037016742770168e-05,
+      "loss": 2.1949,
+      "step": 15825
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.035494672754947e-05,
+      "loss": 2.2957,
+      "step": 15850
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.0339726027397262e-05,
+      "loss": 2.1833,
+      "step": 15875
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.0324505327245055e-05,
+      "loss": 2.2777,
+      "step": 15900
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.0309284627092847e-05,
+      "loss": 2.3056,
+      "step": 15925
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.029406392694064e-05,
+      "loss": 2.2749,
+      "step": 15950
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.0278843226788436e-05,
+      "loss": 2.3683,
+      "step": 15975
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.026362252663623e-05,
+      "loss": 2.3215,
+      "step": 16000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0248401826484022e-05,
+      "loss": 2.1942,
+      "step": 16025
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.0233181126331815e-05,
+      "loss": 2.17,
+      "step": 16050
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.0217960426179608e-05,
+      "loss": 2.2755,
+      "step": 16075
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.02027397260274e-05,
+      "loss": 2.2263,
+      "step": 16100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.018751902587519e-05,
+      "loss": 2.2477,
+      "step": 16125
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.0172298325722983e-05,
+      "loss": 2.3542,
+      "step": 16150
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.0157077625570783e-05,
+      "loss": 2.2229,
+      "step": 16175
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.0141856925418572e-05,
+      "loss": 2.1218,
+      "step": 16200
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.0126636225266365e-05,
+      "loss": 2.2416,
+      "step": 16225
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.0112024353120244e-05,
+      "loss": 2.2834,
+      "step": 16250
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.009680365296804e-05,
+      "loss": 2.2788,
+      "step": 16275
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.0081582952815833e-05,
+      "loss": 2.1984,
+      "step": 16300
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.0066362252663626e-05,
+      "loss": 2.2963,
+      "step": 16325
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.005114155251142e-05,
+      "loss": 2.2688,
+      "step": 16350
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.0035920852359212e-05,
+      "loss": 2.2477,
+      "step": 16375
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.0020700152207e-05,
+      "loss": 2.2984,
+      "step": 16400
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.0005479452054794e-05,
+      "loss": 2.1656,
+      "step": 16425
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.9990258751902587e-05,
+      "loss": 2.2591,
+      "step": 16450
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.9975038051750383e-05,
+      "loss": 2.129,
+      "step": 16475
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.9959817351598176e-05,
+      "loss": 2.2228,
+      "step": 16500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.994459665144597e-05,
+      "loss": 2.2753,
+      "step": 16525
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.9929375951293762e-05,
+      "loss": 2.1261,
+      "step": 16550
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.9914155251141555e-05,
+      "loss": 2.194,
+      "step": 16575
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.9898934550989348e-05,
+      "loss": 2.2243,
+      "step": 16600
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.988371385083714e-05,
+      "loss": 2.1936,
+      "step": 16625
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.9868493150684933e-05,
+      "loss": 2.1148,
+      "step": 16650
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.985327245053273e-05,
+      "loss": 2.215,
+      "step": 16675
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.9838051750380522e-05,
+      "loss": 2.192,
+      "step": 16700
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.9822831050228315e-05,
+      "loss": 2.1374,
+      "step": 16725
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.9807610350076105e-05,
+      "loss": 2.202,
+      "step": 16750
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.9792389649923897e-05,
+      "loss": 2.3239,
+      "step": 16775
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.977716894977169e-05,
+      "loss": 2.2086,
+      "step": 16800
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.9761948249619483e-05,
+      "loss": 2.2127,
+      "step": 16825
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.9746727549467276e-05,
+      "loss": 2.1751,
+      "step": 16850
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.9731506849315072e-05,
+      "loss": 2.2261,
+      "step": 16875
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.9716286149162865e-05,
+      "loss": 2.3385,
+      "step": 16900
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.9701065449010658e-05,
+      "loss": 2.2147,
+      "step": 16925
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.968584474885845e-05,
+      "loss": 2.2131,
+      "step": 16950
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.9670624048706244e-05,
+      "loss": 2.1689,
+      "step": 16975
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.9655403348554036e-05,
+      "loss": 2.1745,
+      "step": 17000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.9640182648401826e-05,
+      "loss": 2.1453,
+      "step": 17025
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.962496194824962e-05,
+      "loss": 2.2432,
+      "step": 17050
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.9609741248097418e-05,
+      "loss": 2.2764,
+      "step": 17075
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.9594520547945208e-05,
+      "loss": 2.2671,
+      "step": 17100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.9579299847793e-05,
+      "loss": 2.2056,
+      "step": 17125
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.9564079147640793e-05,
+      "loss": 2.2265,
+      "step": 17150
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.9549467275494676e-05,
+      "loss": 2.2748,
+      "step": 17175
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.953424657534247e-05,
+      "loss": 2.1414,
+      "step": 17200
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.9519025875190262e-05,
+      "loss": 2.218,
+      "step": 17225
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.9503805175038055e-05,
+      "loss": 2.2895,
+      "step": 17250
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.9488584474885848e-05,
+      "loss": 2.2456,
+      "step": 17275
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.9473363774733637e-05,
+      "loss": 2.1579,
+      "step": 17300
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.945814307458143e-05,
+      "loss": 2.1798,
+      "step": 17325
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.944292237442923e-05,
+      "loss": 2.2155,
+      "step": 17350
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.942770167427702e-05,
+      "loss": 2.2606,
+      "step": 17375
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.9412480974124812e-05,
+      "loss": 2.1556,
+      "step": 17400
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.9397260273972605e-05,
+      "loss": 2.2074,
+      "step": 17425
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.9382039573820398e-05,
+      "loss": 2.2237,
+      "step": 17450
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.936681887366819e-05,
+      "loss": 2.1979,
+      "step": 17475
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.9351598173515983e-05,
+      "loss": 2.2213,
+      "step": 17500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.9336377473363776e-05,
+      "loss": 2.2229,
+      "step": 17525
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.9321156773211572e-05,
+      "loss": 2.2467,
+      "step": 17550
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.9305936073059365e-05,
+      "loss": 2.3175,
+      "step": 17575
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.9290715372907158e-05,
+      "loss": 2.105,
+      "step": 17600
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.927549467275495e-05,
+      "loss": 2.2018,
+      "step": 17625
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.926027397260274e-05,
+      "loss": 2.0804,
+      "step": 17650
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.9245053272450533e-05,
+      "loss": 2.2365,
+      "step": 17675
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.9229832572298326e-05,
+      "loss": 2.1185,
+      "step": 17700
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.921461187214612e-05,
+      "loss": 2.1344,
+      "step": 17725
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.9199391171993915e-05,
+      "loss": 2.2028,
+      "step": 17750
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.9184170471841708e-05,
+      "loss": 2.2562,
+      "step": 17775
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.91689497716895e-05,
+      "loss": 2.2123,
+      "step": 17800
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.9153729071537294e-05,
+      "loss": 2.2453,
+      "step": 17825
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.9138508371385086e-05,
+      "loss": 2.2053,
+      "step": 17850
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.912328767123288e-05,
+      "loss": 2.1562,
+      "step": 17875
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.9108066971080672e-05,
+      "loss": 2.0943,
+      "step": 17900
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.909284627092846e-05,
+      "loss": 2.1485,
+      "step": 17925
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.907762557077626e-05,
+      "loss": 2.2019,
+      "step": 17950
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.9062404870624054e-05,
+      "loss": 2.1521,
+      "step": 17975
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.9047184170471843e-05,
+      "loss": 2.2207,
+      "step": 18000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.9031963470319636e-05,
+      "loss": 2.1452,
+      "step": 18025
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.901674277016743e-05,
+      "loss": 2.1539,
+      "step": 18050
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.9001522070015222e-05,
+      "loss": 2.1437,
+      "step": 18075
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.8986301369863015e-05,
+      "loss": 2.1719,
+      "step": 18100
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.8971080669710808e-05,
+      "loss": 2.1572,
+      "step": 18125
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.8955859969558604e-05,
+      "loss": 2.2169,
+      "step": 18150
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.8940639269406397e-05,
+      "loss": 2.2541,
+      "step": 18175
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.892541856925419e-05,
+      "loss": 2.2111,
+      "step": 18200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.8910197869101982e-05,
+      "loss": 2.2755,
+      "step": 18225
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.8894977168949775e-05,
+      "loss": 2.131,
+      "step": 18250
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.8879756468797565e-05,
+      "loss": 2.1832,
+      "step": 18275
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.8864535768645357e-05,
+      "loss": 2.2637,
+      "step": 18300
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.884931506849315e-05,
+      "loss": 2.3163,
+      "step": 18325
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.8834094368340947e-05,
+      "loss": 2.2512,
+      "step": 18350
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.881887366818874e-05,
+      "loss": 2.2196,
+      "step": 18375
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.8803652968036532e-05,
+      "loss": 2.1694,
+      "step": 18400
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.8788432267884325e-05,
+      "loss": 2.1062,
+      "step": 18425
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.8773211567732118e-05,
+      "loss": 2.2445,
+      "step": 18450
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.875799086757991e-05,
+      "loss": 2.2507,
+      "step": 18475
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.8742770167427704e-05,
+      "loss": 2.2063,
+      "step": 18500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.8727549467275496e-05,
+      "loss": 2.3017,
+      "step": 18525
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.8712328767123293e-05,
+      "loss": 2.1666,
+      "step": 18550
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.8697108066971085e-05,
+      "loss": 2.0839,
+      "step": 18575
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.868188736681888e-05,
+      "loss": 2.2795,
+      "step": 18600
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.8666666666666668e-05,
+      "loss": 2.2939,
+      "step": 18625
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.865144596651446e-05,
+      "loss": 2.2818,
+      "step": 18650
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.8636225266362253e-05,
+      "loss": 2.2006,
+      "step": 18675
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.8621004566210046e-05,
+      "loss": 2.055,
+      "step": 18700
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.860578386605784e-05,
+      "loss": 2.217,
+      "step": 18725
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.8590563165905635e-05,
+      "loss": 2.2419,
+      "step": 18750
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.8575342465753428e-05,
+      "loss": 2.2466,
+      "step": 18775
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.856012176560122e-05,
+      "loss": 2.2655,
+      "step": 18800
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.8544901065449014e-05,
+      "loss": 2.1704,
+      "step": 18825
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.8529680365296807e-05,
+      "loss": 2.2642,
+      "step": 18850
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.85144596651446e-05,
+      "loss": 2.1253,
+      "step": 18875
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.849923896499239e-05,
+      "loss": 2.1265,
+      "step": 18900
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.8484018264840182e-05,
+      "loss": 2.2404,
+      "step": 18925
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.846879756468798e-05,
+      "loss": 2.2131,
+      "step": 18950
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.845357686453577e-05,
+      "loss": 2.2115,
+      "step": 18975
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.8438356164383564e-05,
+      "loss": 2.1654,
+      "step": 19000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.8423135464231357e-05,
+      "loss": 2.1543,
+      "step": 19025
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.840791476407915e-05,
+      "loss": 2.1774,
+      "step": 19050
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.8392694063926942e-05,
+      "loss": 2.348,
+      "step": 19075
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.8377473363774735e-05,
+      "loss": 2.1481,
+      "step": 19100
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.8362252663622528e-05,
+      "loss": 2.2803,
+      "step": 19125
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.8347031963470324e-05,
+      "loss": 2.1804,
+      "step": 19150
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.8331811263318117e-05,
+      "loss": 2.1929,
+      "step": 19175
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.831659056316591e-05,
+      "loss": 2.1959,
+      "step": 19200
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.8301369863013703e-05,
+      "loss": 2.2637,
+      "step": 19225
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.8286149162861492e-05,
+      "loss": 2.2302,
+      "step": 19250
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.8270928462709285e-05,
+      "loss": 2.2437,
+      "step": 19275
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.8255707762557078e-05,
+      "loss": 2.2423,
+      "step": 19300
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.824048706240487e-05,
+      "loss": 2.2753,
+      "step": 19325
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.8225266362252667e-05,
+      "loss": 2.1957,
+      "step": 19350
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.821004566210046e-05,
+      "loss": 2.2732,
+      "step": 19375
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.8194824961948253e-05,
+      "loss": 2.2094,
+      "step": 19400
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.8179604261796045e-05,
+      "loss": 2.106,
+      "step": 19425
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.8164383561643838e-05,
+      "loss": 2.2468,
+      "step": 19450
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.814916286149163e-05,
+      "loss": 2.1399,
+      "step": 19475
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.8133942161339424e-05,
+      "loss": 2.2894,
+      "step": 19500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.8118721461187213e-05,
+      "loss": 2.2403,
+      "step": 19525
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.8103500761035013e-05,
+      "loss": 2.1764,
+      "step": 19550
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.8088280060882806e-05,
+      "loss": 2.2491,
+      "step": 19575
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.8073059360730595e-05,
+      "loss": 2.2154,
+      "step": 19600
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.8057838660578388e-05,
+      "loss": 2.1643,
+      "step": 19625
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.804261796042618e-05,
+      "loss": 2.1644,
+      "step": 19650
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.8027397260273974e-05,
+      "loss": 2.1413,
+      "step": 19675
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.8012176560121767e-05,
+      "loss": 2.1563,
+      "step": 19700
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.799695585996956e-05,
+      "loss": 2.2266,
+      "step": 19725
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.7981735159817356e-05,
+      "loss": 2.1458,
+      "step": 19750
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.796651445966515e-05,
+      "loss": 2.1858,
+      "step": 19775
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.795129375951294e-05,
+      "loss": 2.1965,
+      "step": 19800
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.7936073059360734e-05,
+      "loss": 2.1685,
+      "step": 19825
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.7920852359208527e-05,
+      "loss": 2.1751,
+      "step": 19850
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.7905631659056317e-05,
+      "loss": 2.0926,
+      "step": 19875
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.789041095890411e-05,
+      "loss": 2.2655,
+      "step": 19900
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.7875190258751902e-05,
+      "loss": 2.1804,
+      "step": 19925
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.78599695585997e-05,
+      "loss": 2.1363,
+      "step": 19950
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.784474885844749e-05,
+      "loss": 2.2423,
+      "step": 19975
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.7829528158295284e-05,
+      "loss": 2.0739,
+      "step": 20000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.7814307458143077e-05,
+      "loss": 2.2207,
+      "step": 20025
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.779908675799087e-05,
+      "loss": 2.1476,
+      "step": 20050
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.7783866057838663e-05,
+      "loss": 2.3169,
+      "step": 20075
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.7768645357686455e-05,
+      "loss": 2.1508,
+      "step": 20100
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.775342465753425e-05,
+      "loss": 2.1237,
+      "step": 20125
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.7738203957382045e-05,
+      "loss": 2.2374,
+      "step": 20150
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.7722983257229837e-05,
+      "loss": 2.2625,
+      "step": 20175
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.770776255707763e-05,
+      "loss": 2.1413,
+      "step": 20200
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.769254185692542e-05,
+      "loss": 2.2093,
+      "step": 20225
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.7677321156773213e-05,
+      "loss": 2.1778,
+      "step": 20250
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.7662100456621005e-05,
+      "loss": 2.1181,
+      "step": 20275
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.7646879756468798e-05,
+      "loss": 2.2382,
+      "step": 20300
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.763165905631659e-05,
+      "loss": 2.2121,
+      "step": 20325
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.7616438356164387e-05,
+      "loss": 2.1685,
+      "step": 20350
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.760121765601218e-05,
+      "loss": 2.1297,
+      "step": 20375
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.7585996955859973e-05,
+      "loss": 2.0985,
+      "step": 20400
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.7570776255707766e-05,
+      "loss": 2.2109,
+      "step": 20425
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.755555555555556e-05,
+      "loss": 2.1285,
+      "step": 20450
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.754033485540335e-05,
+      "loss": 2.1805,
+      "step": 20475
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.752511415525114e-05,
+      "loss": 2.2301,
+      "step": 20500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.7509893455098934e-05,
+      "loss": 2.0494,
+      "step": 20525
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.7494672754946733e-05,
+      "loss": 2.2462,
+      "step": 20550
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.7479452054794523e-05,
+      "loss": 2.174,
+      "step": 20575
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.7464231354642316e-05,
+      "loss": 2.2102,
+      "step": 20600
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.744901065449011e-05,
+      "loss": 2.1673,
+      "step": 20625
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.74337899543379e-05,
+      "loss": 2.1535,
+      "step": 20650
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.7418569254185694e-05,
+      "loss": 2.1412,
+      "step": 20675
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.7403348554033487e-05,
+      "loss": 2.1825,
+      "step": 20700
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.738812785388128e-05,
+      "loss": 2.1337,
+      "step": 20725
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.7372907153729076e-05,
+      "loss": 2.1748,
+      "step": 20750
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.735768645357687e-05,
+      "loss": 2.2003,
+      "step": 20775
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.7342465753424662e-05,
+      "loss": 2.1943,
+      "step": 20800
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.7327245053272455e-05,
+      "loss": 2.188,
+      "step": 20825
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.7312024353120244e-05,
+      "loss": 2.1538,
+      "step": 20850
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.7296803652968037e-05,
+      "loss": 2.1763,
+      "step": 20875
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.728158295281583e-05,
+      "loss": 2.202,
+      "step": 20900
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.7266362252663623e-05,
+      "loss": 2.0576,
+      "step": 20925
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.725114155251142e-05,
+      "loss": 2.0509,
+      "step": 20950
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.723592085235921e-05,
+      "loss": 2.1629,
+      "step": 20975
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.7220700152207005e-05,
+      "loss": 2.2119,
+      "step": 21000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.7205479452054797e-05,
+      "loss": 2.1668,
+      "step": 21025
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.719025875190259e-05,
+      "loss": 2.1514,
+      "step": 21050
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.7175038051750383e-05,
+      "loss": 2.1272,
+      "step": 21075
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.7159817351598176e-05,
+      "loss": 2.1029,
+      "step": 21100
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.7144596651445965e-05,
+      "loss": 2.112,
+      "step": 21125
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.7129375951293765e-05,
+      "loss": 2.1119,
+      "step": 21150
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.7114155251141558e-05,
+      "loss": 2.0902,
+      "step": 21175
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.7098934550989347e-05,
+      "loss": 2.1778,
+      "step": 21200
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.708371385083714e-05,
+      "loss": 2.0972,
+      "step": 21225
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.7068493150684933e-05,
+      "loss": 2.1768,
+      "step": 21250
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.7053272450532726e-05,
+      "loss": 2.178,
+      "step": 21275
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.703805175038052e-05,
+      "loss": 2.1795,
+      "step": 21300
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.702283105022831e-05,
+      "loss": 2.0706,
+      "step": 21325
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.7007610350076108e-05,
+      "loss": 2.241,
+      "step": 21350
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.69923896499239e-05,
+      "loss": 2.102,
+      "step": 21375
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.6977168949771693e-05,
+      "loss": 2.2229,
+      "step": 21400
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.6961948249619486e-05,
+      "loss": 2.083,
+      "step": 21425
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.694672754946728e-05,
+      "loss": 2.0917,
+      "step": 21450
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.693150684931507e-05,
+      "loss": 2.256,
+      "step": 21475
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.691628614916286e-05,
+      "loss": 2.1306,
+      "step": 21500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.6901065449010654e-05,
+      "loss": 2.1752,
+      "step": 21525
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.688584474885845e-05,
+      "loss": 2.1803,
+      "step": 21550
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.6870624048706243e-05,
+      "loss": 2.1966,
+      "step": 21575
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.6855403348554036e-05,
+      "loss": 2.193,
+      "step": 21600
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.684018264840183e-05,
+      "loss": 2.1957,
+      "step": 21625
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.6824961948249622e-05,
+      "loss": 2.2211,
+      "step": 21650
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.6809741248097415e-05,
+      "loss": 2.1052,
+      "step": 21675
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.6794520547945207e-05,
+      "loss": 2.1593,
+      "step": 21700
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.6779299847793e-05,
+      "loss": 2.0678,
+      "step": 21725
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.6764079147640796e-05,
+      "loss": 2.2496,
+      "step": 21750
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.674885844748859e-05,
+      "loss": 2.1239,
+      "step": 21775
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.6733637747336382e-05,
+      "loss": 2.1516,
+      "step": 21800
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.671841704718417e-05,
+      "loss": 2.0752,
+      "step": 21825
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.6703196347031964e-05,
+      "loss": 2.2453,
+      "step": 21850
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.6687975646879757e-05,
+      "loss": 2.3072,
+      "step": 21875
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.667275494672755e-05,
+      "loss": 2.1156,
+      "step": 21900
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.6657534246575343e-05,
+      "loss": 2.1962,
+      "step": 21925
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.664231354642314e-05,
+      "loss": 2.0963,
+      "step": 21950
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.6627092846270932e-05,
+      "loss": 2.191,
+      "step": 21975
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.6611872146118725e-05,
+      "loss": 2.2095,
+      "step": 22000
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.6596651445966518e-05,
+      "loss": 2.1171,
+      "step": 22025
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.658143074581431e-05,
+      "loss": 2.0649,
+      "step": 22050
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.6566210045662103e-05,
+      "loss": 2.1147,
+      "step": 22075
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.6550989345509893e-05,
+      "loss": 2.2051,
+      "step": 22100
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.6535768645357686e-05,
+      "loss": 2.2622,
+      "step": 22125
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.6520547945205485e-05,
+      "loss": 2.1057,
+      "step": 22150
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.6505327245053275e-05,
+      "loss": 2.2136,
+      "step": 22175
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.6490106544901068e-05,
+      "loss": 2.1146,
+      "step": 22200
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.647488584474886e-05,
+      "loss": 2.1073,
+      "step": 22225
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.6459665144596653e-05,
+      "loss": 2.2008,
+      "step": 22250
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.6444444444444446e-05,
+      "loss": 2.1423,
+      "step": 22275
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.642922374429224e-05,
+      "loss": 2.1471,
+      "step": 22300
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.6414003044140032e-05,
+      "loss": 2.2099,
+      "step": 22325
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.6398782343987828e-05,
+      "loss": 2.1029,
+      "step": 22350
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.638356164383562e-05,
+      "loss": 2.1722,
+      "step": 22375
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.6368340943683414e-05,
+      "loss": 2.2043,
+      "step": 22400
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.6353120243531207e-05,
+      "loss": 2.144,
+      "step": 22425
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.6337899543378996e-05,
+      "loss": 2.1838,
+      "step": 22450
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.632267884322679e-05,
+      "loss": 2.2831,
+      "step": 22475
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.630745814307458e-05,
+      "loss": 2.1283,
+      "step": 22500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.6292237442922375e-05,
+      "loss": 2.1583,
+      "step": 22525
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.627701674277017e-05,
+      "loss": 2.1836,
+      "step": 22550
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.6261796042617964e-05,
+      "loss": 2.1956,
+      "step": 22575
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.6246575342465756e-05,
+      "loss": 2.1679,
+      "step": 22600
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.623135464231355e-05,
+      "loss": 2.1463,
+      "step": 22625
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.6216133942161342e-05,
+      "loss": 2.2765,
+      "step": 22650
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.6200913242009135e-05,
+      "loss": 2.094,
+      "step": 22675
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6185692541856928e-05,
+      "loss": 2.0926,
+      "step": 22700
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6170471841704717e-05,
+      "loss": 2.0845,
+      "step": 22725
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6155251141552517e-05,
+      "loss": 2.2262,
+      "step": 22750
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.614003044140031e-05,
+      "loss": 2.1352,
+      "step": 22775
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.61248097412481e-05,
+      "loss": 2.1186,
+      "step": 22800
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6109589041095892e-05,
+      "loss": 2.1048,
+      "step": 22825
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6094368340943685e-05,
+      "loss": 2.1266,
+      "step": 22850
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6079147640791478e-05,
+      "loss": 2.2109,
+      "step": 22875
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.606392694063927e-05,
+      "loss": 2.1962,
+      "step": 22900
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6048706240487063e-05,
+      "loss": 2.2317,
+      "step": 22925
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.603348554033486e-05,
+      "loss": 2.1549,
+      "step": 22950
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6018264840182652e-05,
+      "loss": 2.2,
+      "step": 22975
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6003044140030445e-05,
+      "loss": 2.1342,
+      "step": 23000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.5987823439878238e-05,
+      "loss": 2.0854,
+      "step": 23025
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.597260273972603e-05,
+      "loss": 2.0935,
+      "step": 23050
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.595738203957382e-05,
+      "loss": 2.0792,
+      "step": 23075
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.5942161339421613e-05,
+      "loss": 2.1502,
+      "step": 23100
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.5926940639269406e-05,
+      "loss": 2.1767,
+      "step": 23125
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.5911719939117202e-05,
+      "loss": 2.2009,
+      "step": 23150
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.5896499238964995e-05,
+      "loss": 2.1147,
+      "step": 23175
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.5881278538812788e-05,
+      "loss": 2.1359,
+      "step": 23200
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.586605783866058e-05,
+      "loss": 2.0548,
+      "step": 23225
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.5850837138508374e-05,
+      "loss": 2.0974,
+      "step": 23250
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.5835616438356166e-05,
+      "loss": 2.0914,
+      "step": 23275
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.582039573820396e-05,
+      "loss": 2.0525,
+      "step": 23300
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.5805175038051752e-05,
+      "loss": 2.1788,
+      "step": 23325
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.578995433789955e-05,
+      "loss": 2.0718,
+      "step": 23350
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.577473363774734e-05,
+      "loss": 2.191,
+      "step": 23375
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.5759512937595134e-05,
+      "loss": 2.0822,
+      "step": 23400
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.5744292237442924e-05,
+      "loss": 2.1245,
+      "step": 23425
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.5729071537290716e-05,
+      "loss": 2.1894,
+      "step": 23450
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.571385083713851e-05,
+      "loss": 2.1632,
+      "step": 23475
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.5698630136986302e-05,
+      "loss": 2.2474,
+      "step": 23500
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.5683409436834095e-05,
+      "loss": 2.1482,
+      "step": 23525
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.566818873668189e-05,
+      "loss": 2.164,
+      "step": 23550
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.5652968036529684e-05,
+      "loss": 2.0679,
+      "step": 23575
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.5637747336377477e-05,
+      "loss": 2.0875,
+      "step": 23600
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.562252663622527e-05,
+      "loss": 2.1351,
+      "step": 23625
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.5607305936073062e-05,
+      "loss": 2.137,
+      "step": 23650
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5592085235920855e-05,
+      "loss": 2.1051,
+      "step": 23675
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5576864535768645e-05,
+      "loss": 2.0808,
+      "step": 23700
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5561643835616438e-05,
+      "loss": 2.1877,
+      "step": 23725
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5546423135464237e-05,
+      "loss": 2.16,
+      "step": 23750
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5531202435312027e-05,
+      "loss": 2.1346,
+      "step": 23775
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.551598173515982e-05,
+      "loss": 2.0598,
+      "step": 23800
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5500761035007612e-05,
+      "loss": 2.1666,
+      "step": 23825
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5485540334855405e-05,
+      "loss": 2.1896,
+      "step": 23850
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5470319634703198e-05,
+      "loss": 2.1093,
+      "step": 23875
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.545509893455099e-05,
+      "loss": 2.0752,
+      "step": 23900
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5439878234398784e-05,
+      "loss": 2.1268,
+      "step": 23925
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.542465753424658e-05,
+      "loss": 2.1263,
+      "step": 23950
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5409436834094373e-05,
+      "loss": 2.1407,
+      "step": 23975
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.539482496194825e-05,
+      "loss": 2.1309,
+      "step": 24000
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5379604261796042e-05,
+      "loss": 2.093,
+      "step": 24025
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5364383561643838e-05,
+      "loss": 2.1654,
+      "step": 24050
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.534916286149163e-05,
+      "loss": 2.1689,
+      "step": 24075
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5333942161339424e-05,
+      "loss": 2.1654,
+      "step": 24100
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5318721461187216e-05,
+      "loss": 2.2091,
+      "step": 24125
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.53041095890411e-05,
+      "loss": 2.0029,
+      "step": 24150
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5288888888888892e-05,
+      "loss": 2.1621,
+      "step": 24175
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5273668188736685e-05,
+      "loss": 2.1024,
+      "step": 24200
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5258447488584478e-05,
+      "loss": 2.0642,
+      "step": 24225
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5243226788432267e-05,
+      "loss": 2.1373,
+      "step": 24250
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.522800608828006e-05,
+      "loss": 2.1484,
+      "step": 24275
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5212785388127853e-05,
+      "loss": 2.1251,
+      "step": 24300
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.519756468797565e-05,
+      "loss": 2.1474,
+      "step": 24325
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5182343987823442e-05,
+      "loss": 2.173,
+      "step": 24350
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5167123287671235e-05,
+      "loss": 2.0536,
+      "step": 24375
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5151902587519028e-05,
+      "loss": 2.1659,
+      "step": 24400
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.513668188736682e-05,
+      "loss": 2.1325,
+      "step": 24425
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5121461187214613e-05,
+      "loss": 2.1678,
+      "step": 24450
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5106240487062406e-05,
+      "loss": 2.1031,
+      "step": 24475
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.50910197869102e-05,
+      "loss": 2.1283,
+      "step": 24500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5075799086757995e-05,
+      "loss": 2.0485,
+      "step": 24525
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5060578386605788e-05,
+      "loss": 2.1137,
+      "step": 24550
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.504535768645358e-05,
+      "loss": 2.043,
+      "step": 24575
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.503013698630137e-05,
+      "loss": 2.128,
+      "step": 24600
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.5014916286149163e-05,
+      "loss": 2.1445,
+      "step": 24625
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.4999695585996956e-05,
+      "loss": 2.1807,
+      "step": 24650
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.498447488584475e-05,
+      "loss": 2.1145,
+      "step": 24675
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4969254185692542e-05,
+      "loss": 2.1025,
+      "step": 24700
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4954033485540338e-05,
+      "loss": 2.2375,
+      "step": 24725
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.493881278538813e-05,
+      "loss": 2.0273,
+      "step": 24750
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4923592085235924e-05,
+      "loss": 2.202,
+      "step": 24775
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4908371385083717e-05,
+      "loss": 2.0888,
+      "step": 24800
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.48937595129376e-05,
+      "loss": 2.1503,
+      "step": 24825
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4878538812785392e-05,
+      "loss": 1.9954,
+      "step": 24850
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4863318112633185e-05,
+      "loss": 2.1307,
+      "step": 24875
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4848097412480975e-05,
+      "loss": 2.0709,
+      "step": 24900
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4832876712328767e-05,
+      "loss": 2.0874,
+      "step": 24925
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.481765601217656e-05,
+      "loss": 2.0786,
+      "step": 24950
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4802435312024353e-05,
+      "loss": 2.1518,
+      "step": 24975
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.478721461187215e-05,
+      "loss": 2.1586,
+      "step": 25000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4771993911719942e-05,
+      "loss": 2.0567,
+      "step": 25025
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4756773211567735e-05,
+      "loss": 2.1463,
+      "step": 25050
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4741552511415528e-05,
+      "loss": 2.1937,
+      "step": 25075
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.472633181126332e-05,
+      "loss": 2.0548,
+      "step": 25100
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4711111111111114e-05,
+      "loss": 2.2335,
+      "step": 25125
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4695890410958906e-05,
+      "loss": 2.0733,
+      "step": 25150
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4680669710806696e-05,
+      "loss": 2.057,
+      "step": 25175
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4665449010654495e-05,
+      "loss": 2.0907,
+      "step": 25200
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.465083713850837e-05,
+      "loss": 2.124,
+      "step": 25225
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4635616438356164e-05,
+      "loss": 2.1252,
+      "step": 25250
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.462039573820396e-05,
+      "loss": 2.135,
+      "step": 25275
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4605175038051753e-05,
+      "loss": 2.0997,
+      "step": 25300
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4589954337899546e-05,
+      "loss": 2.1866,
+      "step": 25325
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.457473363774734e-05,
+      "loss": 2.1311,
+      "step": 25350
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4559512937595132e-05,
+      "loss": 2.1598,
+      "step": 25375
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4544292237442925e-05,
+      "loss": 1.9998,
+      "step": 25400
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4529071537290718e-05,
+      "loss": 2.1786,
+      "step": 25425
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4513850837138507e-05,
+      "loss": 2.1348,
+      "step": 25450
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4498630136986307e-05,
+      "loss": 2.1772,
+      "step": 25475
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.44834094368341e-05,
+      "loss": 2.1718,
+      "step": 25500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.446818873668189e-05,
+      "loss": 2.0588,
+      "step": 25525
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4452968036529682e-05,
+      "loss": 2.1257,
+      "step": 25550
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4437747336377475e-05,
+      "loss": 2.0751,
+      "step": 25575
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4422526636225268e-05,
+      "loss": 2.2192,
+      "step": 25600
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.440730593607306e-05,
+      "loss": 2.125,
+      "step": 25625
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.4392085235920853e-05,
+      "loss": 2.0945,
+      "step": 25650
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.437686453576865e-05,
+      "loss": 2.2376,
+      "step": 25675
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.4361643835616442e-05,
+      "loss": 2.1176,
+      "step": 25700
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.4346423135464235e-05,
+      "loss": 2.0887,
+      "step": 25725
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.4331202435312028e-05,
+      "loss": 2.1796,
+      "step": 25750
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.431598173515982e-05,
+      "loss": 2.1368,
+      "step": 25775
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.430076103500761e-05,
+      "loss": 2.2211,
+      "step": 25800
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.4285540334855403e-05,
+      "loss": 2.147,
+      "step": 25825
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.4270319634703196e-05,
+      "loss": 2.2212,
+      "step": 25850
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.4255098934550992e-05,
+      "loss": 2.0163,
+      "step": 25875
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.4239878234398785e-05,
+      "loss": 2.1679,
+      "step": 25900
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.4224657534246578e-05,
+      "loss": 2.1631,
+      "step": 25925
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.420943683409437e-05,
+      "loss": 2.2289,
+      "step": 25950
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.4194216133942164e-05,
+      "loss": 2.1687,
+      "step": 25975
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.4178995433789956e-05,
+      "loss": 2.1701,
+      "step": 26000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.416377473363775e-05,
+      "loss": 2.1613,
+      "step": 26025
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.4148554033485542e-05,
+      "loss": 2.0793,
+      "step": 26050
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.413333333333334e-05,
+      "loss": 2.2479,
+      "step": 26075
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.411811263318113e-05,
+      "loss": 2.1886,
+      "step": 26100
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.4102891933028924e-05,
+      "loss": 2.2579,
+      "step": 26125
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.4087671232876713e-05,
+      "loss": 2.1238,
+      "step": 26150
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.4072450532724506e-05,
+      "loss": 2.121,
+      "step": 26175
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.40572298325723e-05,
+      "loss": 2.0184,
+      "step": 26200
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.4042009132420092e-05,
+      "loss": 2.0781,
+      "step": 26225
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.4026788432267885e-05,
+      "loss": 2.1665,
+      "step": 26250
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.401156773211568e-05,
+      "loss": 2.078,
+      "step": 26275
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3996347031963474e-05,
+      "loss": 2.0936,
+      "step": 26300
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3981126331811267e-05,
+      "loss": 2.1457,
+      "step": 26325
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.396590563165906e-05,
+      "loss": 2.218,
+      "step": 26350
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3950684931506852e-05,
+      "loss": 2.1591,
+      "step": 26375
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3935464231354645e-05,
+      "loss": 2.0221,
+      "step": 26400
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3920243531202435e-05,
+      "loss": 2.0926,
+      "step": 26425
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3905022831050228e-05,
+      "loss": 2.0845,
+      "step": 26450
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3889802130898027e-05,
+      "loss": 2.1087,
+      "step": 26475
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3874581430745817e-05,
+      "loss": 2.1333,
+      "step": 26500
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.385936073059361e-05,
+      "loss": 2.0067,
+      "step": 26525
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3844140030441402e-05,
+      "loss": 2.1004,
+      "step": 26550
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3828919330289195e-05,
+      "loss": 2.1132,
+      "step": 26575
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3813698630136988e-05,
+      "loss": 2.1223,
+      "step": 26600
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.379847792998478e-05,
+      "loss": 2.054,
+      "step": 26625
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.3783257229832574e-05,
+      "loss": 2.0263,
+      "step": 26650
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.376803652968037e-05,
+      "loss": 2.102,
+      "step": 26675
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.3752815829528163e-05,
+      "loss": 2.0161,
+      "step": 26700
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.3737595129375956e-05,
+      "loss": 2.1884,
+      "step": 26725
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.372237442922375e-05,
+      "loss": 2.0326,
+      "step": 26750
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.3707153729071538e-05,
+      "loss": 2.0053,
+      "step": 26775
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.369193302891933e-05,
+      "loss": 2.101,
+      "step": 26800
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.3676712328767124e-05,
+      "loss": 2.1225,
+      "step": 26825
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.3661491628614916e-05,
+      "loss": 2.165,
+      "step": 26850
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.3646270928462713e-05,
+      "loss": 2.0627,
+      "step": 26875
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.3631050228310505e-05,
+      "loss": 2.1096,
+      "step": 26900
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.3615829528158298e-05,
+      "loss": 2.1333,
+      "step": 26925
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.360060882800609e-05,
+      "loss": 2.0718,
+      "step": 26950
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.3585388127853884e-05,
+      "loss": 2.0667,
+      "step": 26975
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.3570167427701677e-05,
+      "loss": 2.0372,
+      "step": 27000
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.355494672754947e-05,
+      "loss": 2.1661,
+      "step": 27025
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.353972602739726e-05,
+      "loss": 2.1163,
+      "step": 27050
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.3524505327245052e-05,
+      "loss": 2.0357,
+      "step": 27075
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.350928462709285e-05,
+      "loss": 2.0945,
+      "step": 27100
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.349406392694064e-05,
+      "loss": 2.2122,
+      "step": 27125
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.3478843226788434e-05,
+      "loss": 2.2343,
+      "step": 27150
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.3463622526636227e-05,
+      "loss": 2.083,
+      "step": 27175
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.344840182648402e-05,
+      "loss": 2.1022,
+      "step": 27200
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.3433181126331812e-05,
+      "loss": 2.0508,
+      "step": 27225
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.3417960426179605e-05,
+      "loss": 2.0839,
+      "step": 27250
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.3402739726027398e-05,
+      "loss": 2.0814,
+      "step": 27275
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.3387519025875194e-05,
+      "loss": 2.0378,
+      "step": 27300
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.3372298325722987e-05,
+      "loss": 2.0934,
+      "step": 27325
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.335707762557078e-05,
+      "loss": 2.0811,
+      "step": 27350
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.3341856925418573e-05,
+      "loss": 2.0548,
+      "step": 27375
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.3326636225266362e-05,
+      "loss": 2.0106,
+      "step": 27400
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.3311415525114155e-05,
+      "loss": 2.0514,
+      "step": 27425
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.3296194824961948e-05,
+      "loss": 2.0765,
+      "step": 27450
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.328097412480974e-05,
+      "loss": 2.2477,
+      "step": 27475
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.3265753424657537e-05,
+      "loss": 2.0045,
+      "step": 27500
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.325053272450533e-05,
+      "loss": 2.1184,
+      "step": 27525
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.3235312024353123e-05,
+      "loss": 2.143,
+      "step": 27550
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.3220091324200915e-05,
+      "loss": 2.0317,
+      "step": 27575
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.320487062404871e-05,
+      "loss": 2.0431,
+      "step": 27600
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.31896499238965e-05,
+      "loss": 2.0331,
+      "step": 27625
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.3174429223744294e-05,
+      "loss": 1.9947,
+      "step": 27650
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.3159208523592083e-05,
+      "loss": 2.143,
+      "step": 27675
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.3143987823439883e-05,
+      "loss": 2.2113,
+      "step": 27700
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.3128767123287676e-05,
+      "loss": 2.0961,
+      "step": 27725
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.3113546423135465e-05,
+      "loss": 2.0829,
+      "step": 27750
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.3098325722983258e-05,
+      "loss": 2.1045,
+      "step": 27775
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.308310502283105e-05,
+      "loss": 2.2277,
+      "step": 27800
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.3067884322678844e-05,
+      "loss": 1.9936,
+      "step": 27825
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.3052663622526637e-05,
+      "loss": 2.1255,
+      "step": 27850
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.303744292237443e-05,
+      "loss": 2.1072,
+      "step": 27875
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.3022222222222226e-05,
+      "loss": 2.0673,
+      "step": 27900
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.300700152207002e-05,
+      "loss": 2.1319,
+      "step": 27925
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.299178082191781e-05,
+      "loss": 2.1209,
+      "step": 27950
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.2977168949771694e-05,
+      "loss": 2.1048,
+      "step": 27975
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.2961948249619487e-05,
+      "loss": 2.0485,
+      "step": 28000
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.2946727549467277e-05,
+      "loss": 2.0914,
+      "step": 28025
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.293150684931507e-05,
+      "loss": 2.0312,
+      "step": 28050
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.2916286149162862e-05,
+      "loss": 2.1114,
+      "step": 28075
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.2901065449010655e-05,
+      "loss": 2.0711,
+      "step": 28100
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.2885844748858448e-05,
+      "loss": 2.182,
+      "step": 28125
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.287062404870624e-05,
+      "loss": 2.1511,
+      "step": 28150
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.2855403348554037e-05,
+      "loss": 2.0635,
+      "step": 28175
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.284018264840183e-05,
+      "loss": 2.0999,
+      "step": 28200
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.2824961948249623e-05,
+      "loss": 2.0751,
+      "step": 28225
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.2809741248097416e-05,
+      "loss": 2.0463,
+      "step": 28250
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.279452054794521e-05,
+      "loss": 2.0961,
+      "step": 28275
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.2779299847792998e-05,
+      "loss": 2.1299,
+      "step": 28300
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.276407914764079e-05,
+      "loss": 2.0544,
+      "step": 28325
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.2748858447488584e-05,
+      "loss": 2.0825,
+      "step": 28350
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.273363774733638e-05,
+      "loss": 2.1132,
+      "step": 28375
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.2718417047184173e-05,
+      "loss": 2.1715,
+      "step": 28400
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.2703196347031965e-05,
+      "loss": 2.0798,
+      "step": 28425
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.268797564687976e-05,
+      "loss": 2.0872,
+      "step": 28450
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.267275494672755e-05,
+      "loss": 2.0075,
+      "step": 28475
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.2657534246575344e-05,
+      "loss": 2.029,
+      "step": 28500
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.2642313546423137e-05,
+      "loss": 2.1095,
+      "step": 28525
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.262709284627093e-05,
+      "loss": 2.144,
+      "step": 28550
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.2611872146118726e-05,
+      "loss": 2.1777,
+      "step": 28575
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.259665144596652e-05,
+      "loss": 2.1214,
+      "step": 28600
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.258143074581431e-05,
+      "loss": 2.1533,
+      "step": 28625
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.25662100456621e-05,
+      "loss": 2.2198,
+      "step": 28650
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.2550989345509894e-05,
+      "loss": 2.1757,
+      "step": 28675
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.2535768645357687e-05,
+      "loss": 1.9718,
+      "step": 28700
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.252054794520548e-05,
+      "loss": 2.06,
+      "step": 28725
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.2505327245053272e-05,
+      "loss": 2.113,
+      "step": 28750
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.249010654490107e-05,
+      "loss": 1.9986,
+      "step": 28775
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.247488584474886e-05,
+      "loss": 2.0891,
+      "step": 28800
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.2459665144596654e-05,
+      "loss": 2.1643,
+      "step": 28825
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.2444444444444447e-05,
+      "loss": 2.1331,
+      "step": 28850
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.242922374429224e-05,
+      "loss": 2.1478,
+      "step": 28875
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.2414003044140033e-05,
+      "loss": 2.1363,
+      "step": 28900
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.2398782343987822e-05,
+      "loss": 2.092,
+      "step": 28925
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.2383561643835615e-05,
+      "loss": 2.1163,
+      "step": 28950
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.2368340943683415e-05,
+      "loss": 2.0726,
+      "step": 28975
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.2353120243531204e-05,
+      "loss": 2.1177,
+      "step": 29000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.2337899543378997e-05,
+      "loss": 2.1109,
+      "step": 29025
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.232267884322679e-05,
+      "loss": 1.9997,
+      "step": 29050
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.2307458143074583e-05,
+      "loss": 2.1331,
+      "step": 29075
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.2292237442922376e-05,
+      "loss": 2.1436,
+      "step": 29100
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.227701674277017e-05,
+      "loss": 1.897,
+      "step": 29125
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.226179604261796e-05,
+      "loss": 2.1544,
+      "step": 29150
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.2246575342465757e-05,
+      "loss": 2.0765,
+      "step": 29175
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.223135464231355e-05,
+      "loss": 2.0012,
+      "step": 29200
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.2216133942161343e-05,
+      "loss": 2.1186,
+      "step": 29225
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.2200913242009136e-05,
+      "loss": 2.0887,
+      "step": 29250
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.2185692541856925e-05,
+      "loss": 2.109,
+      "step": 29275
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.2170471841704718e-05,
+      "loss": 2.119,
+      "step": 29300
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.215525114155251e-05,
+      "loss": 2.0025,
+      "step": 29325
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.2140030441400304e-05,
+      "loss": 2.0606,
+      "step": 29350
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.21248097412481e-05,
+      "loss": 2.1352,
+      "step": 29375
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.2109589041095893e-05,
+      "loss": 2.1215,
+      "step": 29400
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.2094368340943686e-05,
+      "loss": 2.0059,
+      "step": 29425
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.207914764079148e-05,
+      "loss": 2.1267,
+      "step": 29450
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.206392694063927e-05,
+      "loss": 2.1622,
+      "step": 29475
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.2048706240487064e-05,
+      "loss": 2.1535,
+      "step": 29500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.2033485540334857e-05,
+      "loss": 2.0931,
+      "step": 29525
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2018264840182647e-05,
+      "loss": 2.0341,
+      "step": 29550
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.2003044140030446e-05,
+      "loss": 2.0635,
+      "step": 29575
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 2.198782343987824e-05,
+      "loss": 2.1759,
+      "step": 29600
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 2.197260273972603e-05,
+      "loss": 1.9937,
+      "step": 29625
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.195738203957382e-05,
+      "loss": 1.9897,
+      "step": 29650
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 2.1942161339421614e-05,
+      "loss": 2.0304,
+      "step": 29675
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 2.1926940639269407e-05,
+      "loss": 2.0477,
+      "step": 29700
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 2.19117199391172e-05,
+      "loss": 1.9738,
+      "step": 29725
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.1896499238964993e-05,
+      "loss": 1.9111,
+      "step": 29750
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 2.188127853881279e-05,
+      "loss": 1.9978,
+      "step": 29775
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 2.1866057838660582e-05,
+      "loss": 2.134,
+      "step": 29800
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.1850837138508375e-05,
+      "loss": 2.059,
+      "step": 29825
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.1835616438356168e-05,
+      "loss": 2.0287,
+      "step": 29850
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.182039573820396e-05,
+      "loss": 2.0847,
+      "step": 29875
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.180517503805175e-05,
+      "loss": 1.9959,
+      "step": 29900
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.1789954337899543e-05,
+      "loss": 2.0193,
+      "step": 29925
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.1774733637747335e-05,
+      "loss": 2.0612,
+      "step": 29950
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.1759512937595132e-05,
+      "loss": 2.0862,
+      "step": 29975
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.1744292237442925e-05,
+      "loss": 2.0891,
+      "step": 30000
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.1729071537290717e-05,
+      "loss": 2.0906,
+      "step": 30025
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.171385083713851e-05,
+      "loss": 1.9966,
+      "step": 30050
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.1698630136986303e-05,
+      "loss": 2.0964,
+      "step": 30075
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.1683409436834096e-05,
+      "loss": 2.0682,
+      "step": 30100
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.166818873668189e-05,
+      "loss": 2.0812,
+      "step": 30125
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.165296803652968e-05,
+      "loss": 2.0623,
+      "step": 30150
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.1637747336377478e-05,
+      "loss": 2.1535,
+      "step": 30175
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.162252663622527e-05,
+      "loss": 2.1764,
+      "step": 30200
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.1607305936073064e-05,
+      "loss": 1.9868,
+      "step": 30225
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.1592085235920853e-05,
+      "loss": 2.0232,
+      "step": 30250
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.1576864535768646e-05,
+      "loss": 2.102,
+      "step": 30275
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.156164383561644e-05,
+      "loss": 2.0973,
+      "step": 30300
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.154642313546423e-05,
+      "loss": 2.0408,
+      "step": 30325
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.1531202435312024e-05,
+      "loss": 2.1228,
+      "step": 30350
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.151598173515982e-05,
+      "loss": 2.0236,
+      "step": 30375
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.1500761035007613e-05,
+      "loss": 2.0122,
+      "step": 30400
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.1485540334855406e-05,
+      "loss": 2.0017,
+      "step": 30425
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.14703196347032e-05,
+      "loss": 2.0903,
+      "step": 30450
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.1455098934550992e-05,
+      "loss": 2.0511,
+      "step": 30475
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.1439878234398785e-05,
+      "loss": 2.0603,
+      "step": 30500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.1424657534246574e-05,
+      "loss": 2.0815,
+      "step": 30525
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.1409436834094367e-05,
+      "loss": 2.0313,
+      "step": 30550
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.1394216133942167e-05,
+      "loss": 1.9726,
+      "step": 30575
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.1378995433789956e-05,
+      "loss": 2.0626,
+      "step": 30600
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.136377473363775e-05,
+      "loss": 2.1151,
+      "step": 30625
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1348554033485542e-05,
+      "loss": 2.0992,
+      "step": 30650
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1333333333333335e-05,
+      "loss": 2.0547,
+      "step": 30675
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1318112633181127e-05,
+      "loss": 2.0273,
+      "step": 30700
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.130289193302892e-05,
+      "loss": 2.0666,
+      "step": 30725
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1287671232876713e-05,
+      "loss": 2.1278,
+      "step": 30750
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.127245053272451e-05,
+      "loss": 2.08,
+      "step": 30775
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.1257229832572302e-05,
+      "loss": 2.1754,
+      "step": 30800
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.1242009132420095e-05,
+      "loss": 2.0204,
+      "step": 30825
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.1226788432267888e-05,
+      "loss": 2.0768,
+      "step": 30850
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.1211567732115677e-05,
+      "loss": 2.1745,
+      "step": 30875
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.119634703196347e-05,
+      "loss": 2.0624,
+      "step": 30900
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.1181126331811263e-05,
+      "loss": 2.0544,
+      "step": 30925
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.1165905631659056e-05,
+      "loss": 1.9643,
+      "step": 30950
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.1150684931506852e-05,
+      "loss": 2.0692,
+      "step": 30975
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.1135464231354645e-05,
+      "loss": 2.0349,
+      "step": 31000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.1120243531202438e-05,
+      "loss": 2.0648,
+      "step": 31025
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.110502283105023e-05,
+      "loss": 2.0094,
+      "step": 31050
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.1089802130898023e-05,
+      "loss": 2.118,
+      "step": 31075
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.1074581430745816e-05,
+      "loss": 2.0851,
+      "step": 31100
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.105936073059361e-05,
+      "loss": 2.0231,
+      "step": 31125
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.10441400304414e-05,
+      "loss": 2.1099,
+      "step": 31150
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.1028919330289198e-05,
+      "loss": 2.1271,
+      "step": 31175
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.101369863013699e-05,
+      "loss": 2.0688,
+      "step": 31200
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.099847792998478e-05,
+      "loss": 2.0647,
+      "step": 31225
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.0983257229832573e-05,
+      "loss": 2.1555,
+      "step": 31250
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.0968036529680366e-05,
+      "loss": 2.0468,
+      "step": 31275
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.095281582952816e-05,
+      "loss": 2.1865,
+      "step": 31300
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.0937595129375952e-05,
+      "loss": 2.1422,
+      "step": 31325
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.0922374429223745e-05,
+      "loss": 2.1395,
+      "step": 31350
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.090715372907154e-05,
+      "loss": 2.09,
+      "step": 31375
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.0891933028919334e-05,
+      "loss": 2.0477,
+      "step": 31400
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.0876712328767127e-05,
+      "loss": 2.0753,
+      "step": 31425
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.086149162861492e-05,
+      "loss": 2.0948,
+      "step": 31450
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.0846270928462712e-05,
+      "loss": 2.0595,
+      "step": 31475
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.0831050228310502e-05,
+      "loss": 2.01,
+      "step": 31500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.0815829528158295e-05,
+      "loss": 2.1305,
+      "step": 31525
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.0800608828006087e-05,
+      "loss": 2.1159,
+      "step": 31550
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.0785388127853884e-05,
+      "loss": 2.0244,
+      "step": 31575
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.0770167427701676e-05,
+      "loss": 1.9978,
+      "step": 31600
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.075494672754947e-05,
+      "loss": 2.0019,
+      "step": 31625
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 2.0739726027397262e-05,
+      "loss": 2.1559,
+      "step": 31650
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.0724505327245055e-05,
+      "loss": 2.0498,
+      "step": 31675
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.0709284627092848e-05,
+      "loss": 2.0949,
+      "step": 31700
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 2.069406392694064e-05,
+      "loss": 2.0069,
+      "step": 31725
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.0678843226788434e-05,
+      "loss": 2.1079,
+      "step": 31750
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.066362252663623e-05,
+      "loss": 2.0984,
+      "step": 31775
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 2.0648401826484023e-05,
+      "loss": 2.0458,
+      "step": 31800
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.0633181126331815e-05,
+      "loss": 2.0529,
+      "step": 31825
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.0617960426179605e-05,
+      "loss": 2.1335,
+      "step": 31850
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 2.0602739726027398e-05,
+      "loss": 2.0454,
+      "step": 31875
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.058751902587519e-05,
+      "loss": 2.0767,
+      "step": 31900
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 2.0572298325722983e-05,
+      "loss": 2.0955,
+      "step": 31925
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.0557077625570776e-05,
+      "loss": 2.0321,
+      "step": 31950
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.0541856925418572e-05,
+      "loss": 1.9845,
+      "step": 31975
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 2.0526636225266365e-05,
+      "loss": 1.9996,
+      "step": 32000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.0511415525114158e-05,
+      "loss": 2.1175,
+      "step": 32025
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.049619482496195e-05,
+      "loss": 2.1869,
+      "step": 32050
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.0480974124809744e-05,
+      "loss": 2.0169,
+      "step": 32075
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.0465753424657537e-05,
+      "loss": 2.0489,
+      "step": 32100
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.0450532724505326e-05,
+      "loss": 2.1809,
+      "step": 32125
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.043531202435312e-05,
+      "loss": 2.1068,
+      "step": 32150
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.042009132420092e-05,
+      "loss": 2.1649,
+      "step": 32175
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.0404870624048708e-05,
+      "loss": 2.1867,
+      "step": 32200
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 2.03896499238965e-05,
+      "loss": 2.1665,
+      "step": 32225
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.0374429223744294e-05,
+      "loss": 2.1239,
+      "step": 32250
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.0359208523592087e-05,
+      "loss": 2.0871,
+      "step": 32275
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 2.034398782343988e-05,
+      "loss": 2.0232,
+      "step": 32300
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.0328767123287672e-05,
+      "loss": 2.028,
+      "step": 32325
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.0313546423135465e-05,
+      "loss": 2.0638,
+      "step": 32350
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.029832572298326e-05,
+      "loss": 2.0474,
+      "step": 32375
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 2.0283105022831054e-05,
+      "loss": 2.0768,
+      "step": 32400
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.0267884322678847e-05,
+      "loss": 2.0103,
+      "step": 32425
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 2.025266362252664e-05,
+      "loss": 2.0039,
+      "step": 32450
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.023744292237443e-05,
+      "loss": 2.0592,
+      "step": 32475
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.0222222222222222e-05,
+      "loss": 1.9841,
+      "step": 32500
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 2.0207001522070015e-05,
+      "loss": 2.0784,
+      "step": 32525
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.0191780821917808e-05,
+      "loss": 2.1935,
+      "step": 32550
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.0176560121765604e-05,
+      "loss": 2.0357,
+      "step": 32575
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.0161339421613397e-05,
+      "loss": 2.1061,
+      "step": 32600
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.014611872146119e-05,
+      "loss": 2.1107,
+      "step": 32625
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.0130898021308983e-05,
+      "loss": 2.1319,
+      "step": 32650
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.0115677321156775e-05,
+      "loss": 2.059,
+      "step": 32675
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 2.0100456621004568e-05,
+      "loss": 2.1054,
+      "step": 32700
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.008523592085236e-05,
+      "loss": 2.1039,
+      "step": 32725
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 2.007001522070015e-05,
+      "loss": 2.0244,
+      "step": 32750
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.005479452054795e-05,
+      "loss": 2.0507,
+      "step": 32775
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.0039573820395743e-05,
+      "loss": 2.0643,
+      "step": 32800
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.0024353120243532e-05,
+      "loss": 2.0597,
+      "step": 32825
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.0009132420091325e-05,
+      "loss": 2.0781,
+      "step": 32850
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.9993911719939118e-05,
+      "loss": 2.039,
+      "step": 32875
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.997869101978691e-05,
+      "loss": 2.0825,
+      "step": 32900
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.9963470319634707e-05,
+      "loss": 2.0215,
+      "step": 32925
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.9948249619482497e-05,
+      "loss": 1.9872,
+      "step": 32950
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.993302891933029e-05,
+      "loss": 1.9938,
+      "step": 32975
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.9917808219178082e-05,
+      "loss": 2.0155,
+      "step": 33000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.990258751902588e-05,
+      "loss": 1.9574,
+      "step": 33025
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.988736681887367e-05,
+      "loss": 1.9415,
+      "step": 33050
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.9872146118721464e-05,
+      "loss": 2.1311,
+      "step": 33075
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.9856925418569254e-05,
+      "loss": 2.0027,
+      "step": 33100
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.984170471841705e-05,
+      "loss": 2.0285,
+      "step": 33125
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.9826484018264843e-05,
+      "loss": 2.0183,
+      "step": 33150
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.9811263318112636e-05,
+      "loss": 2.0641,
+      "step": 33175
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.979604261796043e-05,
+      "loss": 2.0903,
+      "step": 33200
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.978082191780822e-05,
+      "loss": 1.9673,
+      "step": 33225
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.9765601217656014e-05,
+      "loss": 2.0883,
+      "step": 33250
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.9750380517503807e-05,
+      "loss": 2.0343,
+      "step": 33275
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.97351598173516e-05,
+      "loss": 2.0388,
+      "step": 33300
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.9719939117199393e-05,
+      "loss": 1.9403,
+      "step": 33325
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.9704718417047185e-05,
+      "loss": 2.1293,
+      "step": 33350
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.9689497716894978e-05,
+      "loss": 1.9008,
+      "step": 33375
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.967427701674277e-05,
+      "loss": 2.0903,
+      "step": 33400
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.9659056316590567e-05,
+      "loss": 2.1492,
+      "step": 33425
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.9643835616438357e-05,
+      "loss": 2.0181,
+      "step": 33450
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.962861491628615e-05,
+      "loss": 2.1131,
+      "step": 33475
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.9613394216133942e-05,
+      "loss": 1.9928,
+      "step": 33500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.959817351598174e-05,
+      "loss": 2.091,
+      "step": 33525
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.958295281582953e-05,
+      "loss": 2.0378,
+      "step": 33550
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.956773211567732e-05,
+      "loss": 2.0189,
+      "step": 33575
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.9552511415525114e-05,
+      "loss": 2.1016,
+      "step": 33600
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.953729071537291e-05,
+      "loss": 2.0928,
+      "step": 33625
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.9522070015220703e-05,
+      "loss": 2.1194,
+      "step": 33650
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.9506849315068496e-05,
+      "loss": 2.0628,
+      "step": 33675
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.949162861491629e-05,
+      "loss": 2.0574,
+      "step": 33700
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.947640791476408e-05,
+      "loss": 1.9877,
+      "step": 33725
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.9461187214611874e-05,
+      "loss": 1.959,
+      "step": 33750
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.9445966514459667e-05,
+      "loss": 2.0333,
+      "step": 33775
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.943074581430746e-05,
+      "loss": 2.0887,
+      "step": 33800
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.9415525114155253e-05,
+      "loss": 2.0249,
+      "step": 33825
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.9400304414003046e-05,
+      "loss": 2.0648,
+      "step": 33850
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.938508371385084e-05,
+      "loss": 2.1138,
+      "step": 33875
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.936986301369863e-05,
+      "loss": 2.0822,
+      "step": 33900
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.9354642313546424e-05,
+      "loss": 2.1693,
+      "step": 33925
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.9339421613394217e-05,
+      "loss": 2.0125,
+      "step": 33950
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.932420091324201e-05,
+      "loss": 2.0564,
+      "step": 33975
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.9308980213089803e-05,
+      "loss": 2.1638,
+      "step": 34000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.92937595129376e-05,
+      "loss": 2.0393,
+      "step": 34025
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.9278538812785392e-05,
+      "loss": 2.0248,
+      "step": 34050
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.926331811263318e-05,
+      "loss": 1.9676,
+      "step": 34075
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.9248097412480974e-05,
+      "loss": 2.0428,
+      "step": 34100
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.923287671232877e-05,
+      "loss": 2.0268,
+      "step": 34125
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.9217656012176563e-05,
+      "loss": 2.1396,
+      "step": 34150
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.9202435312024356e-05,
+      "loss": 2.0436,
+      "step": 34175
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.9187214611872145e-05,
+      "loss": 2.1294,
+      "step": 34200
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.917199391171994e-05,
+      "loss": 2.0051,
+      "step": 34225
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.9156773211567734e-05,
+      "loss": 2.0874,
+      "step": 34250
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.9141552511415527e-05,
+      "loss": 2.0328,
+      "step": 34275
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.912633181126332e-05,
+      "loss": 2.0851,
+      "step": 34300
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 2.051,
+      "step": 34325
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.9095890410958906e-05,
+      "loss": 2.1141,
+      "step": 34350
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.90806697108067e-05,
+      "loss": 2.0351,
+      "step": 34375
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.906544901065449e-05,
+      "loss": 1.9976,
+      "step": 34400
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.9050228310502284e-05,
+      "loss": 2.0309,
+      "step": 34425
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.9035007610350077e-05,
+      "loss": 2.0315,
+      "step": 34450
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.901978691019787e-05,
+      "loss": 2.1068,
+      "step": 34475
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.9004566210045663e-05,
+      "loss": 1.9746,
+      "step": 34500
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.898934550989346e-05,
+      "loss": 2.0455,
+      "step": 34525
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.897412480974125e-05,
+      "loss": 2.1544,
+      "step": 34550
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.895890410958904e-05,
+      "loss": 2.0103,
+      "step": 34575
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.8943683409436834e-05,
+      "loss": 2.0323,
+      "step": 34600
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.892846270928463e-05,
+      "loss": 2.138,
+      "step": 34625
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.8913242009132423e-05,
+      "loss": 2.0743,
+      "step": 34650
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.8898021308980216e-05,
+      "loss": 2.097,
+      "step": 34675
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.8882800608828006e-05,
+      "loss": 2.1001,
+      "step": 34700
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.8867579908675802e-05,
+      "loss": 2.1197,
+      "step": 34725
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.8852359208523595e-05,
+      "loss": 2.0936,
+      "step": 34750
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.8837138508371387e-05,
+      "loss": 2.0947,
+      "step": 34775
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.882191780821918e-05,
+      "loss": 1.9545,
+      "step": 34800
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.8806697108066973e-05,
+      "loss": 1.9313,
+      "step": 34825
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.8791476407914766e-05,
+      "loss": 1.9543,
+      "step": 34850
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.877625570776256e-05,
+      "loss": 1.8876,
+      "step": 34875
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.876103500761035e-05,
+      "loss": 2.0014,
+      "step": 34900
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.8745814307458145e-05,
+      "loss": 2.0244,
+      "step": 34925
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.8730593607305937e-05,
+      "loss": 2.033,
+      "step": 34950
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.871537290715373e-05,
+      "loss": 2.0249,
+      "step": 34975
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.8700761035007613e-05,
+      "loss": 2.052,
+      "step": 35000
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.8685540334855406e-05,
+      "loss": 1.9771,
+      "step": 35025
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.86703196347032e-05,
+      "loss": 2.036,
+      "step": 35050
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.865509893455099e-05,
+      "loss": 2.0739,
+      "step": 35075
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.8639878234398784e-05,
+      "loss": 1.9556,
+      "step": 35100
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.8624657534246577e-05,
+      "loss": 2.0215,
+      "step": 35125
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.860943683409437e-05,
+      "loss": 1.9275,
+      "step": 35150
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.8594216133942163e-05,
+      "loss": 2.0273,
+      "step": 35175
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.8578995433789956e-05,
+      "loss": 2.0919,
+      "step": 35200
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.856377473363775e-05,
+      "loss": 2.0538,
+      "step": 35225
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.854855403348554e-05,
+      "loss": 2.0061,
+      "step": 35250
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 2.0814,
+      "step": 35275
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.851811263318113e-05,
+      "loss": 2.1047,
+      "step": 35300
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.850289193302892e-05,
+      "loss": 1.9969,
+      "step": 35325
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.8487671232876713e-05,
+      "loss": 2.0814,
+      "step": 35350
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.8472450532724506e-05,
+      "loss": 1.9949,
+      "step": 35375
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.8457229832572302e-05,
+      "loss": 2.037,
+      "step": 35400
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.8442009132420095e-05,
+      "loss": 2.0312,
+      "step": 35425
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.8426788432267884e-05,
+      "loss": 1.987,
+      "step": 35450
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.8411567732115677e-05,
+      "loss": 2.1242,
+      "step": 35475
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.839634703196347e-05,
+      "loss": 2.0262,
+      "step": 35500
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.8381126331811266e-05,
+      "loss": 2.0393,
+      "step": 35525
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.836590563165906e-05,
+      "loss": 2.0485,
+      "step": 35550
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.8350684931506852e-05,
+      "loss": 2.0249,
+      "step": 35575
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.833546423135464e-05,
+      "loss": 1.9595,
+      "step": 35600
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.8320243531202437e-05,
+      "loss": 2.0546,
+      "step": 35625
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.830502283105023e-05,
+      "loss": 2.0052,
+      "step": 35650
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.8289802130898023e-05,
+      "loss": 2.0484,
+      "step": 35675
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.8274581430745816e-05,
+      "loss": 2.0976,
+      "step": 35700
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.825936073059361e-05,
+      "loss": 2.0822,
+      "step": 35725
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.82441400304414e-05,
+      "loss": 2.0432,
+      "step": 35750
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.8228919330289195e-05,
+      "loss": 1.9908,
+      "step": 35775
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.8213698630136987e-05,
+      "loss": 2.1414,
+      "step": 35800
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.819847792998478e-05,
+      "loss": 1.9891,
+      "step": 35825
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.8183257229832573e-05,
+      "loss": 2.054,
+      "step": 35850
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.8168036529680366e-05,
+      "loss": 2.1285,
+      "step": 35875
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.815281582952816e-05,
+      "loss": 2.0289,
+      "step": 35900
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.8137595129375955e-05,
+      "loss": 2.0027,
+      "step": 35925
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.8122374429223744e-05,
+      "loss": 2.0613,
+      "step": 35950
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.8107153729071537e-05,
+      "loss": 2.0867,
+      "step": 35975
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.809193302891933e-05,
+      "loss": 2.0042,
+      "step": 36000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.8076712328767126e-05,
+      "loss": 1.9788,
+      "step": 36025
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.806149162861492e-05,
+      "loss": 2.1272,
+      "step": 36050
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.804627092846271e-05,
+      "loss": 2.1209,
+      "step": 36075
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.80310502283105e-05,
+      "loss": 2.0478,
+      "step": 36100
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.8015829528158298e-05,
+      "loss": 2.0026,
+      "step": 36125
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.800060882800609e-05,
+      "loss": 1.9632,
+      "step": 36150
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.7985388127853883e-05,
+      "loss": 1.9789,
+      "step": 36175
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.7970167427701676e-05,
+      "loss": 2.006,
+      "step": 36200
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.795494672754947e-05,
+      "loss": 1.9146,
+      "step": 36225
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.7939726027397262e-05,
+      "loss": 2.0145,
+      "step": 36250
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.7924505327245055e-05,
+      "loss": 1.9779,
+      "step": 36275
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.7909284627092848e-05,
+      "loss": 2.0325,
+      "step": 36300
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.789406392694064e-05,
+      "loss": 2.1302,
+      "step": 36325
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.7878843226788433e-05,
+      "loss": 2.035,
+      "step": 36350
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.7863622526636226e-05,
+      "loss": 2.0235,
+      "step": 36375
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.784840182648402e-05,
+      "loss": 2.0755,
+      "step": 36400
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.7833181126331812e-05,
+      "loss": 2.1273,
+      "step": 36425
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.7817960426179605e-05,
+      "loss": 1.8809,
+      "step": 36450
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.7802739726027397e-05,
+      "loss": 2.0696,
+      "step": 36475
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.778751902587519e-05,
+      "loss": 2.0482,
+      "step": 36500
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.7772298325722986e-05,
+      "loss": 1.9742,
+      "step": 36525
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.775707762557078e-05,
+      "loss": 2.0024,
+      "step": 36550
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.774185692541857e-05,
+      "loss": 2.13,
+      "step": 36575
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.772663622526636e-05,
+      "loss": 2.055,
+      "step": 36600
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.7711415525114158e-05,
+      "loss": 2.0081,
+      "step": 36625
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.769619482496195e-05,
+      "loss": 2.083,
+      "step": 36650
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.7680974124809744e-05,
+      "loss": 1.9954,
+      "step": 36675
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.7665753424657533e-05,
+      "loss": 1.9231,
+      "step": 36700
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.765053272450533e-05,
+      "loss": 2.0633,
+      "step": 36725
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.7635312024353122e-05,
+      "loss": 2.0269,
+      "step": 36750
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.7620091324200915e-05,
+      "loss": 2.0619,
+      "step": 36775
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.7604870624048708e-05,
+      "loss": 2.1641,
+      "step": 36800
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.75896499238965e-05,
+      "loss": 2.0819,
+      "step": 36825
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.7574429223744293e-05,
+      "loss": 2.0132,
+      "step": 36850
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.7559208523592086e-05,
+      "loss": 2.0833,
+      "step": 36875
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.754398782343988e-05,
+      "loss": 1.9813,
+      "step": 36900
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.7528767123287672e-05,
+      "loss": 1.9838,
+      "step": 36925
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.7513546423135465e-05,
+      "loss": 1.9377,
+      "step": 36950
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.7498325722983258e-05,
+      "loss": 2.0033,
+      "step": 36975
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.748310502283105e-05,
+      "loss": 2.0323,
+      "step": 37000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.7467884322678847e-05,
+      "loss": 1.8743,
+      "step": 37025
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.7452663622526636e-05,
+      "loss": 2.021,
+      "step": 37050
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.743744292237443e-05,
+      "loss": 1.995,
+      "step": 37075
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.7422222222222222e-05,
+      "loss": 2.0027,
+      "step": 37100
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.7407001522070018e-05,
+      "loss": 1.9754,
+      "step": 37125
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.739178082191781e-05,
+      "loss": 2.042,
+      "step": 37150
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.7376560121765604e-05,
+      "loss": 1.9945,
+      "step": 37175
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.7361339421613393e-05,
+      "loss": 1.9659,
+      "step": 37200
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.734611872146119e-05,
+      "loss": 2.083,
+      "step": 37225
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.7330898021308982e-05,
+      "loss": 1.9871,
+      "step": 37250
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.7315677321156775e-05,
+      "loss": 1.9931,
+      "step": 37275
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.7300456621004568e-05,
+      "loss": 2.0554,
+      "step": 37300
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.728523592085236e-05,
+      "loss": 2.118,
+      "step": 37325
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.7270015220700154e-05,
+      "loss": 2.0515,
+      "step": 37350
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.7254794520547946e-05,
+      "loss": 2.0505,
+      "step": 37375
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.723957382039574e-05,
+      "loss": 2.1095,
+      "step": 37400
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.7224353120243532e-05,
+      "loss": 2.0882,
+      "step": 37425
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.7209132420091325e-05,
+      "loss": 1.9593,
+      "step": 37450
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.7193911719939118e-05,
+      "loss": 2.0295,
+      "step": 37475
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.717869101978691e-05,
+      "loss": 2.0128,
+      "step": 37500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.7163470319634707e-05,
+      "loss": 2.0363,
+      "step": 37525
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.7148249619482496e-05,
+      "loss": 2.0494,
+      "step": 37550
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.713302891933029e-05,
+      "loss": 2.0278,
+      "step": 37575
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.7117808219178082e-05,
+      "loss": 1.9859,
+      "step": 37600
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.7102587519025878e-05,
+      "loss": 2.0568,
+      "step": 37625
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.708736681887367e-05,
+      "loss": 1.9878,
+      "step": 37650
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.707214611872146e-05,
+      "loss": 2.0664,
+      "step": 37675
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.7056925418569253e-05,
+      "loss": 1.9954,
+      "step": 37700
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.704170471841705e-05,
+      "loss": 1.9442,
+      "step": 37725
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.7026484018264842e-05,
+      "loss": 1.9689,
+      "step": 37750
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.7011263318112635e-05,
+      "loss": 1.9743,
+      "step": 37775
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.6996042617960428e-05,
+      "loss": 2.0726,
+      "step": 37800
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.698082191780822e-05,
+      "loss": 1.9607,
+      "step": 37825
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.6965601217656014e-05,
+      "loss": 2.0583,
+      "step": 37850
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.6950380517503807e-05,
+      "loss": 2.0146,
+      "step": 37875
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.69351598173516e-05,
+      "loss": 2.01,
+      "step": 37900
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 1.6919939117199392e-05,
+      "loss": 2.0699,
+      "step": 37925
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.6904718417047185e-05,
+      "loss": 2.0734,
+      "step": 37950
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 1.6889497716894978e-05,
+      "loss": 2.0048,
+      "step": 37975
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 1.687427701674277e-05,
+      "loss": 1.9044,
+      "step": 38000
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.6859056316590564e-05,
+      "loss": 2.0226,
+      "step": 38025
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.6843835616438357e-05,
+      "loss": 1.9519,
+      "step": 38050
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 1.682861491628615e-05,
+      "loss": 2.03,
+      "step": 38075
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.6813394216133942e-05,
+      "loss": 2.0279,
+      "step": 38100
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.679817351598174e-05,
+      "loss": 1.9242,
+      "step": 38125
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 1.678295281582953e-05,
+      "loss": 2.0287,
+      "step": 38150
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.676773211567732e-05,
+      "loss": 1.9837,
+      "step": 38175
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 1.6752511415525114e-05,
+      "loss": 2.1662,
+      "step": 38200
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.673729071537291e-05,
+      "loss": 2.0756,
+      "step": 38225
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.6722070015220703e-05,
+      "loss": 2.0991,
+      "step": 38250
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 1.6706849315068495e-05,
+      "loss": 2.0585,
+      "step": 38275
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 1.6691628614916285e-05,
+      "loss": 2.0222,
+      "step": 38300
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.667640791476408e-05,
+      "loss": 2.1262,
+      "step": 38325
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.6661187214611874e-05,
+      "loss": 2.0557,
+      "step": 38350
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 1.6645966514459667e-05,
+      "loss": 2.0179,
+      "step": 38375
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.663074581430746e-05,
+      "loss": 2.0048,
+      "step": 38400
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.6615525114155252e-05,
+      "loss": 1.9947,
+      "step": 38425
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 1.6600304414003045e-05,
+      "loss": 1.9683,
+      "step": 38450
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.6585083713850838e-05,
+      "loss": 1.9387,
+      "step": 38475
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 1.656986301369863e-05,
+      "loss": 2.0681,
+      "step": 38500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.6554642313546424e-05,
+      "loss": 2.0681,
+      "step": 38525
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.6539421613394217e-05,
+      "loss": 2.0351,
+      "step": 38550
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 1.652420091324201e-05,
+      "loss": 2.0481,
+      "step": 38575
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 1.6508980213089802e-05,
+      "loss": 1.9093,
+      "step": 38600
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.64937595129376e-05,
+      "loss": 1.9843,
+      "step": 38625
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.6478538812785388e-05,
+      "loss": 1.958,
+      "step": 38650
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 1.646331811263318e-05,
+      "loss": 2.0893,
+      "step": 38675
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.6448097412480974e-05,
+      "loss": 1.9919,
+      "step": 38700
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 1.643287671232877e-05,
+      "loss": 1.906,
+      "step": 38725
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.6417656012176563e-05,
+      "loss": 2.062,
+      "step": 38750
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.6402435312024356e-05,
+      "loss": 2.1367,
+      "step": 38775
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 1.6387214611872145e-05,
+      "loss": 2.022,
+      "step": 38800
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.637199391171994e-05,
+      "loss": 1.9896,
+      "step": 38825
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.6356773211567734e-05,
+      "loss": 1.9913,
+      "step": 38850
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 1.6341552511415527e-05,
+      "loss": 2.0554,
+      "step": 38875
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 1.632633181126332e-05,
+      "loss": 2.0311,
+      "step": 38900
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.6311111111111113e-05,
+      "loss": 2.0395,
+      "step": 38925
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.6295890410958906e-05,
+      "loss": 2.0641,
+      "step": 38950
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 1.62806697108067e-05,
+      "loss": 1.9859,
+      "step": 38975
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.626544901065449e-05,
+      "loss": 1.9978,
+      "step": 39000
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 1.6250228310502284e-05,
+      "loss": 2.042,
+      "step": 39025
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.6235007610350077e-05,
+      "loss": 1.9892,
+      "step": 39050
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.621978691019787e-05,
+      "loss": 2.0111,
+      "step": 39075
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 1.6204566210045663e-05,
+      "loss": 2.0407,
+      "step": 39100
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.618934550989346e-05,
+      "loss": 2.0049,
+      "step": 39125
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 1.6174124809741248e-05,
+      "loss": 1.9931,
+      "step": 39150
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.615890410958904e-05,
+      "loss": 2.0479,
+      "step": 39175
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 1.6143683409436834e-05,
+      "loss": 2.1207,
+      "step": 39200
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.612846270928463e-05,
+      "loss": 2.0321,
+      "step": 39225
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 1.6113242009132423e-05,
+      "loss": 2.0861,
+      "step": 39250
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.6098021308980212e-05,
+      "loss": 2.0689,
+      "step": 39275
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.6082800608828005e-05,
+      "loss": 2.0498,
+      "step": 39300
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 1.60675799086758e-05,
+      "loss": 2.0761,
+      "step": 39325
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.6052359208523594e-05,
+      "loss": 1.9528,
+      "step": 39350
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.6037138508371387e-05,
+      "loss": 2.0626,
+      "step": 39375
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 1.602191780821918e-05,
+      "loss": 2.0268,
+      "step": 39400
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.6006697108066973e-05,
+      "loss": 1.9873,
+      "step": 39425
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 1.5991476407914766e-05,
+      "loss": 1.9849,
+      "step": 39450
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.597625570776256e-05,
+      "loss": 1.9636,
+      "step": 39475
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 1.596103500761035e-05,
+      "loss": 2.0264,
+      "step": 39500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.5945814307458144e-05,
+      "loss": 2.0667,
+      "step": 39525
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 1.5930593607305937e-05,
+      "loss": 2.0723,
+      "step": 39550
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.591537290715373e-05,
+      "loss": 1.923,
+      "step": 39575
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.5900152207001523e-05,
+      "loss": 2.0104,
+      "step": 39600
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 1.5884931506849316e-05,
+      "loss": 1.9761,
+      "step": 39625
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.586971080669711e-05,
+      "loss": 1.9907,
+      "step": 39650
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 1.58544901065449e-05,
+      "loss": 1.988,
+      "step": 39675
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.5839878234398784e-05,
+      "loss": 1.8632,
+      "step": 39700
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.5824657534246577e-05,
+      "loss": 2.0041,
+      "step": 39725
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 1.580943683409437e-05,
+      "loss": 1.9654,
+      "step": 39750
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 1.5794216133942163e-05,
+      "loss": 2.0408,
+      "step": 39775
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.5778995433789956e-05,
+      "loss": 1.991,
+      "step": 39800
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.576377473363775e-05,
+      "loss": 2.0496,
+      "step": 39825
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 1.574855403348554e-05,
+      "loss": 1.9504,
+      "step": 39850
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 1.982,
+      "step": 39875
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.5718112633181127e-05,
+      "loss": 1.9645,
+      "step": 39900
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 1.570289193302892e-05,
+      "loss": 1.9775,
+      "step": 39925
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.5687671232876713e-05,
+      "loss": 2.0309,
+      "step": 39950
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 1.5672450532724505e-05,
+      "loss": 1.94,
+      "step": 39975
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.56572298325723e-05,
+      "loss": 1.9667,
+      "step": 40000
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.5642009132420094e-05,
+      "loss": 1.9854,
+      "step": 40025
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 1.5626788432267884e-05,
+      "loss": 2.031,
+      "step": 40050
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 1.5611567732115677e-05,
+      "loss": 2.0135,
+      "step": 40075
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.5596347031963473e-05,
+      "loss": 2.0855,
+      "step": 40100
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.5581126331811266e-05,
+      "loss": 1.8975,
+      "step": 40125
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 1.556590563165906e-05,
+      "loss": 2.0706,
+      "step": 40150
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.5550684931506848e-05,
+      "loss": 2.043,
+      "step": 40175
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 1.5535464231354644e-05,
+      "loss": 1.9185,
+      "step": 40200
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.5520243531202437e-05,
+      "loss": 2.03,
+      "step": 40225
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.5505631659056317e-05,
+      "loss": 2.0385,
+      "step": 40250
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 1.5490410958904113e-05,
+      "loss": 2.0019,
+      "step": 40275
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.5475190258751906e-05,
+      "loss": 1.9028,
+      "step": 40300
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.5459969558599695e-05,
+      "loss": 2.0015,
+      "step": 40325
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 1.5444748858447488e-05,
+      "loss": 1.94,
+      "step": 40350
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 1.5429528158295284e-05,
+      "loss": 2.0226,
+      "step": 40375
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.5414307458143077e-05,
+      "loss": 2.0047,
+      "step": 40400
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.539908675799087e-05,
+      "loss": 2.0061,
+      "step": 40425
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 1.5383866057838663e-05,
+      "loss": 2.0525,
+      "step": 40450
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.5368645357686456e-05,
+      "loss": 1.9416,
+      "step": 40475
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 1.535342465753425e-05,
+      "loss": 1.8657,
+      "step": 40500
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.533820395738204e-05,
+      "loss": 1.9785,
+      "step": 40525
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.5322983257229834e-05,
+      "loss": 1.9967,
+      "step": 40550
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 1.5307762557077627e-05,
+      "loss": 2.1364,
+      "step": 40575
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.529254185692542e-05,
+      "loss": 2.0049,
+      "step": 40600
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.5277321156773213e-05,
+      "loss": 1.9425,
+      "step": 40625
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 1.5262100456621006e-05,
+      "loss": 1.9966,
+      "step": 40650
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 1.52468797564688e-05,
+      "loss": 2.0006,
+      "step": 40675
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.5231659056316593e-05,
+      "loss": 2.0976,
+      "step": 40700
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.5216438356164384e-05,
+      "loss": 2.0786,
+      "step": 40725
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 1.5201217656012177e-05,
+      "loss": 2.013,
+      "step": 40750
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.5185996955859971e-05,
+      "loss": 2.027,
+      "step": 40775
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 1.5170776255707764e-05,
+      "loss": 2.0121,
+      "step": 40800
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.5155555555555557e-05,
+      "loss": 2.0365,
+      "step": 40825
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.514033485540335e-05,
+      "loss": 1.9879,
+      "step": 40850
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 1.5125114155251144e-05,
+      "loss": 1.9805,
+      "step": 40875
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.5109893455098936e-05,
+      "loss": 1.9679,
+      "step": 40900
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 1.5094672754946728e-05,
+      "loss": 2.0051,
+      "step": 40925
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.5079452054794521e-05,
+      "loss": 1.9057,
+      "step": 40950
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 1.5064231354642316e-05,
+      "loss": 2.0421,
+      "step": 40975
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.5049010654490109e-05,
+      "loss": 1.9747,
+      "step": 41000
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 1.5033789954337901e-05,
+      "loss": 1.9291,
+      "step": 41025
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.5018569254185693e-05,
+      "loss": 1.9702,
+      "step": 41050
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.5003348554033487e-05,
+      "loss": 1.9764,
+      "step": 41075
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 1.498812785388128e-05,
+      "loss": 2.0485,
+      "step": 41100
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.4972907153729073e-05,
+      "loss": 2.0413,
+      "step": 41125
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.4957686453576866e-05,
+      "loss": 1.9439,
+      "step": 41150
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 1.494246575342466e-05,
+      "loss": 1.9991,
+      "step": 41175
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.4927245053272451e-05,
+      "loss": 1.9862,
+      "step": 41200
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 1.4912024353120244e-05,
+      "loss": 2.0129,
+      "step": 41225
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.4896803652968037e-05,
+      "loss": 2.0102,
+      "step": 41250
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 1.4881582952815832e-05,
+      "loss": 1.9498,
+      "step": 41275
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.4866362252663624e-05,
+      "loss": 2.0143,
+      "step": 41300
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 1.4851141552511417e-05,
+      "loss": 2.1354,
+      "step": 41325
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.4835920852359208e-05,
+      "loss": 1.9088,
+      "step": 41350
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.4820700152207003e-05,
+      "loss": 1.9757,
+      "step": 41375
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 1.4805479452054796e-05,
+      "loss": 2.0642,
+      "step": 41400
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.4790258751902589e-05,
+      "loss": 1.9588,
+      "step": 41425
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 1.4775038051750381e-05,
+      "loss": 2.0616,
+      "step": 41450
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.4759817351598176e-05,
+      "loss": 2.0629,
+      "step": 41475
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.4744596651445969e-05,
+      "loss": 2.0032,
+      "step": 41500
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 1.472937595129376e-05,
+      "loss": 2.0258,
+      "step": 41525
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 1.4714155251141553e-05,
+      "loss": 2.0051,
+      "step": 41550
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.4698934550989347e-05,
+      "loss": 1.9466,
+      "step": 41575
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.468371385083714e-05,
+      "loss": 1.9509,
+      "step": 41600
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 1.4668493150684933e-05,
+      "loss": 1.9443,
+      "step": 41625
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.4653272450532726e-05,
+      "loss": 2.09,
+      "step": 41650
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.463805175038052e-05,
+      "loss": 1.9646,
+      "step": 41675
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 1.4622831050228312e-05,
+      "loss": 1.9612,
+      "step": 41700
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.4607610350076104e-05,
+      "loss": 2.0115,
+      "step": 41725
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 1.4592389649923897e-05,
+      "loss": 1.9786,
+      "step": 41750
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.4577168949771692e-05,
+      "loss": 1.9919,
+      "step": 41775
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.4561948249619485e-05,
+      "loss": 1.9885,
+      "step": 41800
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 1.4546727549467277e-05,
+      "loss": 2.0261,
+      "step": 41825
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 1.4531506849315069e-05,
+      "loss": 2.0076,
+      "step": 41850
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.4516286149162863e-05,
+      "loss": 2.1131,
+      "step": 41875
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.4501065449010656e-05,
+      "loss": 1.9584,
+      "step": 41900
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 1.4485844748858449e-05,
+      "loss": 2.0659,
+      "step": 41925
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.4470624048706242e-05,
+      "loss": 1.9934,
+      "step": 41950
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 1.4455403348554036e-05,
+      "loss": 2.0764,
+      "step": 41975
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.4440182648401827e-05,
+      "loss": 1.9509,
+      "step": 42000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.442496194824962e-05,
+      "loss": 1.952,
+      "step": 42025
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.4409741248097413e-05,
+      "loss": 1.9733,
+      "step": 42050
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.4394520547945208e-05,
+      "loss": 2.055,
+      "step": 42075
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.4379299847793e-05,
+      "loss": 2.0173,
+      "step": 42100
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 1.4364079147640793e-05,
+      "loss": 2.0792,
+      "step": 42125
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.4348858447488584e-05,
+      "loss": 2.0051,
+      "step": 42150
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.4334246575342467e-05,
+      "loss": 2.1026,
+      "step": 42175
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.431902587519026e-05,
+      "loss": 1.9983,
+      "step": 42200
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.4303805175038053e-05,
+      "loss": 2.0111,
+      "step": 42225
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.4288584474885844e-05,
+      "loss": 1.9746,
+      "step": 42250
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.427336377473364e-05,
+      "loss": 1.9585,
+      "step": 42275
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4258143074581431e-05,
+      "loss": 1.9328,
+      "step": 42300
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4242922374429224e-05,
+      "loss": 2.0076,
+      "step": 42325
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 1.4227701674277017e-05,
+      "loss": 1.9502,
+      "step": 42350
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.4212480974124812e-05,
+      "loss": 2.0057,
+      "step": 42375
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.4197260273972605e-05,
+      "loss": 2.0274,
+      "step": 42400
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 1.4182039573820396e-05,
+      "loss": 2.0714,
+      "step": 42425
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.4166818873668189e-05,
+      "loss": 2.0538,
+      "step": 42450
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.4151598173515983e-05,
+      "loss": 2.0079,
+      "step": 42475
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.4136377473363776e-05,
+      "loss": 2.012,
+      "step": 42500
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.4121156773211569e-05,
+      "loss": 2.0648,
+      "step": 42525
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.4105936073059362e-05,
+      "loss": 1.9938,
+      "step": 42550
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.4090715372907156e-05,
+      "loss": 2.0337,
+      "step": 42575
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.4075494672754947e-05,
+      "loss": 1.9989,
+      "step": 42600
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.406027397260274e-05,
+      "loss": 2.0122,
+      "step": 42625
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.4045053272450533e-05,
+      "loss": 2.0362,
+      "step": 42650
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.4029832572298327e-05,
+      "loss": 2.0769,
+      "step": 42675
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 1.401461187214612e-05,
+      "loss": 1.9409,
+      "step": 42700
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.3999391171993913e-05,
+      "loss": 2.0088,
+      "step": 42725
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.3984170471841704e-05,
+      "loss": 1.9523,
+      "step": 42750
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.3968949771689499e-05,
+      "loss": 2.0154,
+      "step": 42775
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.3953729071537292e-05,
+      "loss": 1.8923,
+      "step": 42800
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.3938508371385084e-05,
+      "loss": 1.9403,
+      "step": 42825
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.3923287671232877e-05,
+      "loss": 1.9734,
+      "step": 42850
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.3908066971080672e-05,
+      "loss": 1.9603,
+      "step": 42875
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.3892846270928465e-05,
+      "loss": 2.0133,
+      "step": 42900
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.3877625570776256e-05,
+      "loss": 1.9425,
+      "step": 42925
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.3862404870624049e-05,
+      "loss": 2.0419,
+      "step": 42950
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.3847184170471843e-05,
+      "loss": 2.0259,
+      "step": 42975
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.3831963470319636e-05,
+      "loss": 2.0865,
+      "step": 43000
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.3816742770167429e-05,
+      "loss": 1.939,
+      "step": 43025
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.380152207001522e-05,
+      "loss": 1.9798,
+      "step": 43050
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.3786301369863016e-05,
+      "loss": 1.9552,
+      "step": 43075
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.3771080669710807e-05,
+      "loss": 2.015,
+      "step": 43100
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.37558599695586e-05,
+      "loss": 2.0089,
+      "step": 43125
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.3740639269406393e-05,
+      "loss": 2.0603,
+      "step": 43150
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.3725418569254188e-05,
+      "loss": 2.0267,
+      "step": 43175
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.371019786910198e-05,
+      "loss": 1.9773,
+      "step": 43200
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.3694977168949772e-05,
+      "loss": 2.0409,
+      "step": 43225
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.3679756468797564e-05,
+      "loss": 2.0045,
+      "step": 43250
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.3664535768645359e-05,
+      "loss": 1.9808,
+      "step": 43275
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.3649315068493152e-05,
+      "loss": 1.9854,
+      "step": 43300
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.3634094368340945e-05,
+      "loss": 2.0598,
+      "step": 43325
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.3618873668188738e-05,
+      "loss": 1.9223,
+      "step": 43350
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.3603652968036532e-05,
+      "loss": 1.917,
+      "step": 43375
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.3588432267884323e-05,
+      "loss": 2.0412,
+      "step": 43400
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.3573211567732116e-05,
+      "loss": 1.9624,
+      "step": 43425
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.3557990867579909e-05,
+      "loss": 2.0322,
+      "step": 43450
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.3543378995433792e-05,
+      "loss": 1.9836,
+      "step": 43475
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.3528158295281583e-05,
+      "loss": 2.0068,
+      "step": 43500
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.3512937595129376e-05,
+      "loss": 2.0092,
+      "step": 43525
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.349771689497717e-05,
+      "loss": 2.0502,
+      "step": 43550
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.3482496194824963e-05,
+      "loss": 1.9953,
+      "step": 43575
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.3467275494672756e-05,
+      "loss": 1.9734,
+      "step": 43600
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.3452054794520549e-05,
+      "loss": 1.9861,
+      "step": 43625
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.3436834094368343e-05,
+      "loss": 1.9144,
+      "step": 43650
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.3421613394216134e-05,
+      "loss": 1.8742,
+      "step": 43675
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.3406392694063927e-05,
+      "loss": 2.1363,
+      "step": 43700
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.339117199391172e-05,
+      "loss": 1.9555,
+      "step": 43725
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.3375951293759515e-05,
+      "loss": 1.8268,
+      "step": 43750
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.3360730593607308e-05,
+      "loss": 1.9572,
+      "step": 43775
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.33455098934551e-05,
+      "loss": 2.0172,
+      "step": 43800
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.3330289193302892e-05,
+      "loss": 1.9264,
+      "step": 43825
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.3315068493150686e-05,
+      "loss": 1.9078,
+      "step": 43850
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.3299847792998479e-05,
+      "loss": 1.9868,
+      "step": 43875
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.3284627092846272e-05,
+      "loss": 2.0821,
+      "step": 43900
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 1.3269406392694065e-05,
+      "loss": 2.0153,
+      "step": 43925
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.3254185692541859e-05,
+      "loss": 1.9711,
+      "step": 43950
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.3238964992389652e-05,
+      "loss": 1.9948,
+      "step": 43975
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.3223744292237443e-05,
+      "loss": 2.0175,
+      "step": 44000
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.3208523592085236e-05,
+      "loss": 2.0479,
+      "step": 44025
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.319330289193303e-05,
+      "loss": 1.97,
+      "step": 44050
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.3178082191780823e-05,
+      "loss": 2.1167,
+      "step": 44075
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.3162861491628616e-05,
+      "loss": 1.9078,
+      "step": 44100
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.3147640791476409e-05,
+      "loss": 1.9969,
+      "step": 44125
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.3132420091324204e-05,
+      "loss": 2.076,
+      "step": 44150
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.3117199391171995e-05,
+      "loss": 2.0344,
+      "step": 44175
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.3101978691019788e-05,
+      "loss": 1.9706,
+      "step": 44200
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 1.308675799086758e-05,
+      "loss": 1.9537,
+      "step": 44225
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.3071537290715375e-05,
+      "loss": 1.9367,
+      "step": 44250
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.3056316590563168e-05,
+      "loss": 1.9292,
+      "step": 44275
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.3041095890410959e-05,
+      "loss": 1.9613,
+      "step": 44300
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.3025875190258752e-05,
+      "loss": 1.9972,
+      "step": 44325
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.3010654490106546e-05,
+      "loss": 2.0244,
+      "step": 44350
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.2995433789954339e-05,
+      "loss": 1.8689,
+      "step": 44375
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.2980213089802132e-05,
+      "loss": 2.0505,
+      "step": 44400
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.2964992389649925e-05,
+      "loss": 1.9141,
+      "step": 44425
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.294977168949772e-05,
+      "loss": 1.9494,
+      "step": 44450
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.293455098934551e-05,
+      "loss": 2.0679,
+      "step": 44475
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.2919330289193303e-05,
+      "loss": 2.0277,
+      "step": 44500
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.2904109589041096e-05,
+      "loss": 1.979,
+      "step": 44525
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 1.9098,
+      "step": 44550
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.2873668188736683e-05,
+      "loss": 1.944,
+      "step": 44575
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.2858447488584476e-05,
+      "loss": 2.0591,
+      "step": 44600
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.2843226788432267e-05,
+      "loss": 2.0085,
+      "step": 44625
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.2828006088280062e-05,
+      "loss": 1.9236,
+      "step": 44650
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.2812785388127855e-05,
+      "loss": 1.9142,
+      "step": 44675
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.2797564687975648e-05,
+      "loss": 1.9177,
+      "step": 44700
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.278234398782344e-05,
+      "loss": 1.9706,
+      "step": 44725
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.2767123287671235e-05,
+      "loss": 1.8988,
+      "step": 44750
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.2751902587519028e-05,
+      "loss": 1.9589,
+      "step": 44775
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.2736681887366819e-05,
+      "loss": 2.0487,
+      "step": 44800
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.2721461187214612e-05,
+      "loss": 2.066,
+      "step": 44825
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.2706240487062406e-05,
+      "loss": 2.0137,
+      "step": 44850
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 1.26910197869102e-05,
+      "loss": 2.0245,
+      "step": 44875
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.2675799086757992e-05,
+      "loss": 2.0623,
+      "step": 44900
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.2660578386605783e-05,
+      "loss": 1.9067,
+      "step": 44925
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.264535768645358e-05,
+      "loss": 2.0114,
+      "step": 44950
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.263013698630137e-05,
+      "loss": 1.9816,
+      "step": 44975
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.2614916286149163e-05,
+      "loss": 2.0646,
+      "step": 45000
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.2599695585996956e-05,
+      "loss": 1.9261,
+      "step": 45025
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.258447488584475e-05,
+      "loss": 2.0998,
+      "step": 45050
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.2569254185692544e-05,
+      "loss": 2.0812,
+      "step": 45075
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.2554033485540335e-05,
+      "loss": 1.9178,
+      "step": 45100
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.2538812785388128e-05,
+      "loss": 1.9155,
+      "step": 45125
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.2523592085235922e-05,
+      "loss": 1.9267,
+      "step": 45150
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 1.2508371385083715e-05,
+      "loss": 2.039,
+      "step": 45175
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.2493150684931508e-05,
+      "loss": 2.0541,
+      "step": 45200
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.24779299847793e-05,
+      "loss": 1.9726,
+      "step": 45225
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.2462709284627095e-05,
+      "loss": 1.9593,
+      "step": 45250
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.2447488584474886e-05,
+      "loss": 1.9423,
+      "step": 45275
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.243226788432268e-05,
+      "loss": 1.9559,
+      "step": 45300
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 1.2417047184170472e-05,
+      "loss": 1.9699,
+      "step": 45325
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.2401826484018267e-05,
+      "loss": 1.9827,
+      "step": 45350
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.238660578386606e-05,
+      "loss": 1.9052,
+      "step": 45375
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.2371385083713852e-05,
+      "loss": 1.962,
+      "step": 45400
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.2356164383561643e-05,
+      "loss": 1.9576,
+      "step": 45425
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.2340943683409438e-05,
+      "loss": 2.0912,
+      "step": 45450
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 1.232572298325723e-05,
+      "loss": 1.9244,
+      "step": 45475
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.2310502283105024e-05,
+      "loss": 1.9651,
+      "step": 45500
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.2295281582952816e-05,
+      "loss": 2.1149,
+      "step": 45525
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.2280060882800611e-05,
+      "loss": 1.9963,
+      "step": 45550
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.2264840182648404e-05,
+      "loss": 1.9967,
+      "step": 45575
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.2249619482496195e-05,
+      "loss": 1.9299,
+      "step": 45600
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 1.2234398782343988e-05,
+      "loss": 1.936,
+      "step": 45625
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.2219178082191782e-05,
+      "loss": 2.0121,
+      "step": 45650
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.2203957382039575e-05,
+      "loss": 1.9676,
+      "step": 45675
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.2188736681887368e-05,
+      "loss": 1.9345,
+      "step": 45700
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.217351598173516e-05,
+      "loss": 1.9886,
+      "step": 45725
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.2158295281582955e-05,
+      "loss": 1.9737,
+      "step": 45750
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 1.2143074581430747e-05,
+      "loss": 1.9502,
+      "step": 45775
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.212785388127854e-05,
+      "loss": 1.925,
+      "step": 45800
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.2112633181126332e-05,
+      "loss": 2.0506,
+      "step": 45825
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.2097412480974127e-05,
+      "loss": 1.9704,
+      "step": 45850
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.208219178082192e-05,
+      "loss": 1.9452,
+      "step": 45875
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.206697108066971e-05,
+      "loss": 2.022,
+      "step": 45900
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 1.2051750380517504e-05,
+      "loss": 1.9387,
+      "step": 45925
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.2036529680365298e-05,
+      "loss": 1.9663,
+      "step": 45950
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.2021308980213091e-05,
+      "loss": 2.0126,
+      "step": 45975
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.2006088280060884e-05,
+      "loss": 2.0184,
+      "step": 46000
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.1990867579908677e-05,
+      "loss": 1.893,
+      "step": 46025
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 1.1975646879756471e-05,
+      "loss": 2.0362,
+      "step": 46050
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.1960426179604262e-05,
+      "loss": 1.9939,
+      "step": 46075
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.1945205479452055e-05,
+      "loss": 2.0096,
+      "step": 46100
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.1929984779299848e-05,
+      "loss": 2.0146,
+      "step": 46125
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.1914764079147643e-05,
+      "loss": 1.9037,
+      "step": 46150
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.1899543378995435e-05,
+      "loss": 2.0156,
+      "step": 46175
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.1884322678843228e-05,
+      "loss": 1.9033,
+      "step": 46200
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.186910197869102e-05,
+      "loss": 2.0248,
+      "step": 46225
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.1853881278538814e-05,
+      "loss": 2.0205,
+      "step": 46250
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.1838660578386607e-05,
+      "loss": 1.991,
+      "step": 46275
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.18234398782344e-05,
+      "loss": 1.888,
+      "step": 46300
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.1808219178082192e-05,
+      "loss": 1.8724,
+      "step": 46325
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 1.1792998477929987e-05,
+      "loss": 1.8538,
+      "step": 46350
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.177777777777778e-05,
+      "loss": 1.9678,
+      "step": 46375
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.1762557077625571e-05,
+      "loss": 1.9731,
+      "step": 46400
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 1.1747336377473364e-05,
+      "loss": 1.9184,
+      "step": 46425
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.1732115677321158e-05,
+      "loss": 2.0107,
+      "step": 46450
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 1.1716894977168951e-05,
+      "loss": 1.8769,
+      "step": 46475
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.1701674277016744e-05,
+      "loss": 1.9563,
+      "step": 46500
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.1686453576864535e-05,
+      "loss": 1.9436,
+      "step": 46525
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.1671232876712331e-05,
+      "loss": 1.943,
+      "step": 46550
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 1.1656012176560123e-05,
+      "loss": 1.9551,
+      "step": 46575
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.1640791476407915e-05,
+      "loss": 2.0068,
+      "step": 46600
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.1625570776255708e-05,
+      "loss": 2.0303,
+      "step": 46625
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 1.1610350076103503e-05,
+      "loss": 1.9128,
+      "step": 46650
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.1595129375951296e-05,
+      "loss": 2.0005,
+      "step": 46675
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.1579908675799087e-05,
+      "loss": 1.9755,
+      "step": 46700
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 1.156468797564688e-05,
+      "loss": 1.9389,
+      "step": 46725
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.1549467275494674e-05,
+      "loss": 1.9299,
+      "step": 46750
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 1.1534246575342467e-05,
+      "loss": 2.0026,
+      "step": 46775
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.151902587519026e-05,
+      "loss": 1.9436,
+      "step": 46800
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.1503805175038053e-05,
+      "loss": 1.929,
+      "step": 46825
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.1488584474885847e-05,
+      "loss": 2.0119,
+      "step": 46850
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 1.1473363774733638e-05,
+      "loss": 1.9605,
+      "step": 46875
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.1458143074581431e-05,
+      "loss": 1.9182,
+      "step": 46900
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.1442922374429224e-05,
+      "loss": 2.0203,
+      "step": 46925
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 1.1427701674277019e-05,
+      "loss": 1.9565,
+      "step": 46950
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.1412480974124811e-05,
+      "loss": 2.0476,
+      "step": 46975
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 1.1397260273972604e-05,
+      "loss": 1.9167,
+      "step": 47000
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.1382039573820395e-05,
+      "loss": 1.8304,
+      "step": 47025
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.136681887366819e-05,
+      "loss": 2.0416,
+      "step": 47050
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 1.1351598173515983e-05,
+      "loss": 2.0166,
+      "step": 47075
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.1336377473363776e-05,
+      "loss": 1.9663,
+      "step": 47100
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.1321156773211568e-05,
+      "loss": 2.0136,
+      "step": 47125
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.1305936073059363e-05,
+      "loss": 1.936,
+      "step": 47150
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.1290715372907156e-05,
+      "loss": 1.9158,
+      "step": 47175
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.1275494672754947e-05,
+      "loss": 1.9207,
+      "step": 47200
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.126027397260274e-05,
+      "loss": 2.0042,
+      "step": 47225
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 1.1245053272450534e-05,
+      "loss": 2.0697,
+      "step": 47250
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.1229832572298327e-05,
+      "loss": 1.9097,
+      "step": 47275
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 1.121461187214612e-05,
+      "loss": 1.9998,
+      "step": 47300
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.1199391171993911e-05,
+      "loss": 1.9464,
+      "step": 47325
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.1184170471841707e-05,
+      "loss": 1.9361,
+      "step": 47350
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 1.1168949771689499e-05,
+      "loss": 2.0586,
+      "step": 47375
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.1153729071537291e-05,
+      "loss": 1.9005,
+      "step": 47400
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.1138508371385084e-05,
+      "loss": 1.969,
+      "step": 47425
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.1123287671232879e-05,
+      "loss": 2.015,
+      "step": 47450
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.1108066971080672e-05,
+      "loss": 2.03,
+      "step": 47475
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.1092846270928463e-05,
+      "loss": 1.9863,
+      "step": 47500
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.1077625570776256e-05,
+      "loss": 1.9524,
+      "step": 47525
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 1.106240487062405e-05,
+      "loss": 1.9094,
+      "step": 47550
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.1047184170471843e-05,
+      "loss": 1.9733,
+      "step": 47575
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 1.1031963470319636e-05,
+      "loss": 1.9335,
+      "step": 47600
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.1016742770167429e-05,
+      "loss": 1.9351,
+      "step": 47625
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.1001522070015223e-05,
+      "loss": 1.8879,
+      "step": 47650
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 1.0986301369863014e-05,
+      "loss": 1.956,
+      "step": 47675
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.0971080669710807e-05,
+      "loss": 1.9061,
+      "step": 47700
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.09558599695586e-05,
+      "loss": 2.0037,
+      "step": 47725
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.0940639269406395e-05,
+      "loss": 2.0163,
+      "step": 47750
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.0925418569254187e-05,
+      "loss": 2.1089,
+      "step": 47775
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.091019786910198e-05,
+      "loss": 1.9014,
+      "step": 47800
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 1.0894977168949771e-05,
+      "loss": 2.033,
+      "step": 47825
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.0879756468797566e-05,
+      "loss": 1.9819,
+      "step": 47850
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.0864535768645359e-05,
+      "loss": 1.9444,
+      "step": 47875
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 1.0849315068493152e-05,
+      "loss": 1.9933,
+      "step": 47900
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.0834094368340944e-05,
+      "loss": 2.0139,
+      "step": 47925
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.0818873668188739e-05,
+      "loss": 2.0485,
+      "step": 47950
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 1.0803652968036532e-05,
+      "loss": 1.9591,
+      "step": 47975
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.0788432267884323e-05,
+      "loss": 1.954,
+      "step": 48000
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.0773211567732116e-05,
+      "loss": 1.9854,
+      "step": 48025
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.075799086757991e-05,
+      "loss": 1.8831,
+      "step": 48050
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.0742770167427703e-05,
+      "loss": 2.074,
+      "step": 48075
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.0727549467275496e-05,
+      "loss": 1.9053,
+      "step": 48100
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.0712328767123287e-05,
+      "loss": 2.0317,
+      "step": 48125
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.0697108066971083e-05,
+      "loss": 1.9007,
+      "step": 48150
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0681887366818874e-05,
+      "loss": 1.9001,
+      "step": 48175
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.936,
+      "step": 48200
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.065144596651446e-05,
+      "loss": 1.9031,
+      "step": 48225
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 1.0636225266362255e-05,
+      "loss": 2.0159,
+      "step": 48250
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0621004566210048e-05,
+      "loss": 1.9913,
+      "step": 48275
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.0605783866057839e-05,
+      "loss": 2.0991,
+      "step": 48300
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.0590563165905632e-05,
+      "loss": 2.0389,
+      "step": 48325
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.0575342465753426e-05,
+      "loss": 1.9573,
+      "step": 48350
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.0560121765601219e-05,
+      "loss": 1.9287,
+      "step": 48375
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.0544901065449012e-05,
+      "loss": 1.9629,
+      "step": 48400
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.0529680365296805e-05,
+      "loss": 2.0366,
+      "step": 48425
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.0514459665144599e-05,
+      "loss": 2.0515,
+      "step": 48450
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.049923896499239e-05,
+      "loss": 1.949,
+      "step": 48475
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 1.0484018264840183e-05,
+      "loss": 2.0248,
+      "step": 48500
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.0468797564687976e-05,
+      "loss": 1.9764,
+      "step": 48525
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 1.045357686453577e-05,
+      "loss": 1.8845,
+      "step": 48550
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.0438356164383563e-05,
+      "loss": 2.0516,
+      "step": 48575
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.0423135464231356e-05,
+      "loss": 1.9194,
+      "step": 48600
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.0407914764079147e-05,
+      "loss": 2.0027,
+      "step": 48625
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.0392694063926942e-05,
+      "loss": 1.9999,
+      "step": 48650
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.0377473363774735e-05,
+      "loss": 1.9374,
+      "step": 48675
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.0362252663622528e-05,
+      "loss": 2.0,
+      "step": 48700
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.034703196347032e-05,
+      "loss": 2.0698,
+      "step": 48725
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.0331811263318115e-05,
+      "loss": 1.9073,
+      "step": 48750
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 1.0316590563165908e-05,
+      "loss": 1.879,
+      "step": 48775
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.0301369863013699e-05,
+      "loss": 2.0149,
+      "step": 48800
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.0286149162861492e-05,
+      "loss": 1.9765,
+      "step": 48825
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 1.0270928462709286e-05,
+      "loss": 2.0193,
+      "step": 48850
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.0255707762557079e-05,
+      "loss": 2.0319,
+      "step": 48875
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.0240487062404872e-05,
+      "loss": 2.0198,
+      "step": 48900
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.0225266362252663e-05,
+      "loss": 1.9011,
+      "step": 48925
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.021004566210046e-05,
+      "loss": 1.943,
+      "step": 48950
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.019482496194825e-05,
+      "loss": 1.9647,
+      "step": 48975
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.0179604261796043e-05,
+      "loss": 1.9477,
+      "step": 49000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.0164383561643836e-05,
+      "loss": 1.9735,
+      "step": 49025
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.014916286149163e-05,
+      "loss": 1.9634,
+      "step": 49050
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.0133942161339423e-05,
+      "loss": 2.0279,
+      "step": 49075
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0118721461187215e-05,
+      "loss": 1.9043,
+      "step": 49100
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.0103500761035007e-05,
+      "loss": 1.9345,
+      "step": 49125
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 1.0088280060882802e-05,
+      "loss": 1.9,
+      "step": 49150
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.0073059360730595e-05,
+      "loss": 1.9883,
+      "step": 49175
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.0057838660578388e-05,
+      "loss": 1.9939,
+      "step": 49200
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.004261796042618e-05,
+      "loss": 1.8871,
+      "step": 49225
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.0027397260273975e-05,
+      "loss": 2.009,
+      "step": 49250
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.0012176560121766e-05,
+      "loss": 2.0039,
+      "step": 49275
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 9.996955859969559e-06,
+      "loss": 2.0019,
+      "step": 49300
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 9.981735159817354e-06,
+      "loss": 1.9455,
+      "step": 49325
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 9.966514459665145e-06,
+      "loss": 1.9116,
+      "step": 49350
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 9.95129375951294e-06,
+      "loss": 1.9623,
+      "step": 49375
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 9.936073059360732e-06,
+      "loss": 2.0,
+      "step": 49400
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 9.920852359208525e-06,
+      "loss": 1.9419,
+      "step": 49425
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 9.905631659056318e-06,
+      "loss": 1.9811,
+      "step": 49450
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 9.89041095890411e-06,
+      "loss": 1.9864,
+      "step": 49475
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 9.875190258751903e-06,
+      "loss": 1.8739,
+      "step": 49500
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 9.859969558599696e-06,
+      "loss": 2.0509,
+      "step": 49525
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 9.844748858447489e-06,
+      "loss": 2.0082,
+      "step": 49550
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 9.829528158295284e-06,
+      "loss": 1.961,
+      "step": 49575
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 9.814307458143075e-06,
+      "loss": 1.9278,
+      "step": 49600
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.79908675799087e-06,
+      "loss": 1.9934,
+      "step": 49625
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 9.78386605783866e-06,
+      "loss": 1.8641,
+      "step": 49650
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 9.768645357686455e-06,
+      "loss": 1.9328,
+      "step": 49675
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 9.753424657534248e-06,
+      "loss": 1.991,
+      "step": 49700
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.73820395738204e-06,
+      "loss": 1.9905,
+      "step": 49725
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 9.722983257229834e-06,
+      "loss": 2.0744,
+      "step": 49750
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 9.707762557077626e-06,
+      "loss": 1.8815,
+      "step": 49775
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 9.69254185692542e-06,
+      "loss": 1.8839,
+      "step": 49800
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.677321156773212e-06,
+      "loss": 1.9512,
+      "step": 49825
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 9.662100456621005e-06,
+      "loss": 1.9877,
+      "step": 49850
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 9.6468797564688e-06,
+      "loss": 2.0149,
+      "step": 49875
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 9.63165905631659e-06,
+      "loss": 1.9746,
+      "step": 49900
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 9.616438356164385e-06,
+      "loss": 1.9746,
+      "step": 49925
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.601217656012178e-06,
+      "loss": 1.9566,
+      "step": 49950
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 9.58599695585997e-06,
+      "loss": 1.889,
+      "step": 49975
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 9.570776255707764e-06,
+      "loss": 1.978,
+      "step": 50000
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 9.555555555555556e-06,
+      "loss": 1.9767,
+      "step": 50025
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.54033485540335e-06,
+      "loss": 1.9758,
+      "step": 50050
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 9.525114155251142e-06,
+      "loss": 2.0169,
+      "step": 50075
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 9.509893455098935e-06,
+      "loss": 1.9795,
+      "step": 50100
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 9.49467275494673e-06,
+      "loss": 1.9105,
+      "step": 50125
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.47945205479452e-06,
+      "loss": 1.9358,
+      "step": 50150
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 9.464231354642315e-06,
+      "loss": 1.9197,
+      "step": 50175
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 9.449010654490108e-06,
+      "loss": 1.9679,
+      "step": 50200
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 9.433789954337901e-06,
+      "loss": 2.0478,
+      "step": 50225
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.418569254185694e-06,
+      "loss": 1.9561,
+      "step": 50250
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 9.403348554033487e-06,
+      "loss": 2.0571,
+      "step": 50275
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 9.38812785388128e-06,
+      "loss": 1.8852,
+      "step": 50300
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 9.372907153729072e-06,
+      "loss": 1.8964,
+      "step": 50325
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.357686453576865e-06,
+      "loss": 1.9007,
+      "step": 50350
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 9.342465753424658e-06,
+      "loss": 1.9047,
+      "step": 50375
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 9.32724505327245e-06,
+      "loss": 1.8826,
+      "step": 50400
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 9.312024353120245e-06,
+      "loss": 2.0846,
+      "step": 50425
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 9.296803652968036e-06,
+      "loss": 1.8743,
+      "step": 50450
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.281582952815831e-06,
+      "loss": 1.8567,
+      "step": 50475
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 9.266362252663624e-06,
+      "loss": 1.9336,
+      "step": 50500
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 9.251141552511417e-06,
+      "loss": 1.9018,
+      "step": 50525
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 9.23592085235921e-06,
+      "loss": 1.9833,
+      "step": 50550
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.220700152207002e-06,
+      "loss": 1.9243,
+      "step": 50575
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 9.205479452054795e-06,
+      "loss": 1.999,
+      "step": 50600
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 9.190258751902588e-06,
+      "loss": 2.07,
+      "step": 50625
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 9.17564687975647e-06,
+      "loss": 2.0052,
+      "step": 50650
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.160426179604262e-06,
+      "loss": 1.9705,
+      "step": 50675
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 9.145205479452055e-06,
+      "loss": 1.9387,
+      "step": 50700
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 9.12998477929985e-06,
+      "loss": 2.0054,
+      "step": 50725
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 9.11476407914764e-06,
+      "loss": 1.947,
+      "step": 50750
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.099543378995435e-06,
+      "loss": 1.9877,
+      "step": 50775
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 9.084322678843226e-06,
+      "loss": 1.9343,
+      "step": 50800
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 9.06910197869102e-06,
+      "loss": 1.977,
+      "step": 50825
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 9.053881278538814e-06,
+      "loss": 2.0358,
+      "step": 50850
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.038660578386606e-06,
+      "loss": 1.9384,
+      "step": 50875
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 9.0234398782344e-06,
+      "loss": 1.9109,
+      "step": 50900
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 9.008219178082192e-06,
+      "loss": 1.9731,
+      "step": 50925
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 8.992998477929985e-06,
+      "loss": 1.9435,
+      "step": 50950
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 8.977777777777778e-06,
+      "loss": 1.8879,
+      "step": 50975
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 8.96255707762557e-06,
+      "loss": 1.9415,
+      "step": 51000
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 8.947336377473365e-06,
+      "loss": 1.9765,
+      "step": 51025
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 8.932115677321156e-06,
+      "loss": 1.9545,
+      "step": 51050
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 8.916894977168951e-06,
+      "loss": 1.8787,
+      "step": 51075
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 8.901674277016744e-06,
+      "loss": 1.9561,
+      "step": 51100
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 8.886453576864537e-06,
+      "loss": 2.0651,
+      "step": 51125
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 8.87123287671233e-06,
+      "loss": 1.8878,
+      "step": 51150
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 8.856012176560122e-06,
+      "loss": 1.9929,
+      "step": 51175
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 8.840791476407915e-06,
+      "loss": 1.9261,
+      "step": 51200
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 8.825570776255708e-06,
+      "loss": 2.0192,
+      "step": 51225
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 8.8103500761035e-06,
+      "loss": 1.9563,
+      "step": 51250
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 8.795129375951295e-06,
+      "loss": 1.9667,
+      "step": 51275
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 8.779908675799086e-06,
+      "loss": 1.9906,
+      "step": 51300
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 8.764687975646881e-06,
+      "loss": 1.9746,
+      "step": 51325
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 8.749467275494674e-06,
+      "loss": 1.9885,
+      "step": 51350
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 8.734246575342467e-06,
+      "loss": 1.9304,
+      "step": 51375
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 8.71902587519026e-06,
+      "loss": 1.9789,
+      "step": 51400
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 8.703805175038052e-06,
+      "loss": 1.9598,
+      "step": 51425
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 8.688584474885845e-06,
+      "loss": 1.9788,
+      "step": 51450
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 8.673363774733638e-06,
+      "loss": 1.9393,
+      "step": 51475
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 8.658143074581431e-06,
+      "loss": 1.933,
+      "step": 51500
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 8.642922374429224e-06,
+      "loss": 1.9745,
+      "step": 51525
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 8.627701674277017e-06,
+      "loss": 1.9297,
+      "step": 51550
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 8.612480974124811e-06,
+      "loss": 1.8979,
+      "step": 51575
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 8.597260273972602e-06,
+      "loss": 1.9345,
+      "step": 51600
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 8.582039573820397e-06,
+      "loss": 1.9887,
+      "step": 51625
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 8.56681887366819e-06,
+      "loss": 2.0398,
+      "step": 51650
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 8.551598173515982e-06,
+      "loss": 1.9723,
+      "step": 51675
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 8.536377473363775e-06,
+      "loss": 1.9411,
+      "step": 51700
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 8.521156773211568e-06,
+      "loss": 1.91,
+      "step": 51725
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 8.505936073059361e-06,
+      "loss": 1.9958,
+      "step": 51750
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 8.490715372907154e-06,
+      "loss": 1.9466,
+      "step": 51775
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 8.475494672754947e-06,
+      "loss": 1.8703,
+      "step": 51800
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 8.460273972602741e-06,
+      "loss": 1.9106,
+      "step": 51825
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 8.445053272450532e-06,
+      "loss": 1.9092,
+      "step": 51850
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 8.429832572298327e-06,
+      "loss": 1.9051,
+      "step": 51875
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 8.41461187214612e-06,
+      "loss": 1.9716,
+      "step": 51900
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 8.399391171993913e-06,
+      "loss": 1.8972,
+      "step": 51925
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 8.384170471841705e-06,
+      "loss": 2.0439,
+      "step": 51950
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 8.368949771689498e-06,
+      "loss": 1.9814,
+      "step": 51975
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 8.353729071537291e-06,
+      "loss": 1.9067,
+      "step": 52000
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 8.338508371385084e-06,
+      "loss": 1.9205,
+      "step": 52025
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 8.323287671232877e-06,
+      "loss": 1.9346,
+      "step": 52050
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 8.308066971080671e-06,
+      "loss": 1.9421,
+      "step": 52075
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 8.292846270928462e-06,
+      "loss": 1.8881,
+      "step": 52100
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 8.277625570776257e-06,
+      "loss": 2.0199,
+      "step": 52125
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 8.26240487062405e-06,
+      "loss": 1.8355,
+      "step": 52150
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 8.247184170471843e-06,
+      "loss": 1.9857,
+      "step": 52175
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 8.231963470319635e-06,
+      "loss": 1.9821,
+      "step": 52200
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 8.216742770167428e-06,
+      "loss": 1.9458,
+      "step": 52225
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 8.201522070015221e-06,
+      "loss": 1.9139,
+      "step": 52250
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 8.186301369863014e-06,
+      "loss": 1.9435,
+      "step": 52275
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 8.171080669710807e-06,
+      "loss": 2.0494,
+      "step": 52300
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 8.1558599695586e-06,
+      "loss": 2.0148,
+      "step": 52325
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 8.140639269406393e-06,
+      "loss": 1.9855,
+      "step": 52350
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 8.125418569254187e-06,
+      "loss": 1.8551,
+      "step": 52375
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 8.110197869101978e-06,
+      "loss": 2.0901,
+      "step": 52400
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 8.094977168949773e-06,
+      "loss": 1.9605,
+      "step": 52425
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 8.079756468797566e-06,
+      "loss": 2.009,
+      "step": 52450
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 8.064535768645358e-06,
+      "loss": 1.9351,
+      "step": 52475
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 8.049315068493151e-06,
+      "loss": 1.9546,
+      "step": 52500
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 8.034094368340944e-06,
+      "loss": 1.9336,
+      "step": 52525
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 8.018873668188737e-06,
+      "loss": 1.9764,
+      "step": 52550
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 8.00365296803653e-06,
+      "loss": 1.9384,
+      "step": 52575
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 7.988432267884323e-06,
+      "loss": 1.8792,
+      "step": 52600
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 7.973211567732117e-06,
+      "loss": 1.8954,
+      "step": 52625
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 7.957990867579908e-06,
+      "loss": 1.887,
+      "step": 52650
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 7.942770167427703e-06,
+      "loss": 1.8549,
+      "step": 52675
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 7.927549467275496e-06,
+      "loss": 2.0322,
+      "step": 52700
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 7.912328767123288e-06,
+      "loss": 1.931,
+      "step": 52725
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 7.897108066971081e-06,
+      "loss": 1.9956,
+      "step": 52750
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 7.881887366818874e-06,
+      "loss": 1.9696,
+      "step": 52775
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 7.866666666666667e-06,
+      "loss": 1.9958,
+      "step": 52800
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 7.85144596651446e-06,
+      "loss": 1.9175,
+      "step": 52825
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 7.836225266362253e-06,
+      "loss": 1.8408,
+      "step": 52850
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 7.821004566210047e-06,
+      "loss": 1.8441,
+      "step": 52875
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 7.805783866057838e-06,
+      "loss": 1.9698,
+      "step": 52900
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 7.790563165905633e-06,
+      "loss": 1.8653,
+      "step": 52925
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 7.775342465753424e-06,
+      "loss": 2.0227,
+      "step": 52950
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 7.760121765601219e-06,
+      "loss": 1.9209,
+      "step": 52975
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 7.744901065449011e-06,
+      "loss": 1.9751,
+      "step": 53000
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 7.729680365296804e-06,
+      "loss": 1.9758,
+      "step": 53025
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 7.714459665144597e-06,
+      "loss": 2.0452,
+      "step": 53050
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 7.69923896499239e-06,
+      "loss": 1.9115,
+      "step": 53075
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 7.684018264840183e-06,
+      "loss": 1.961,
+      "step": 53100
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 7.668797564687976e-06,
+      "loss": 1.9015,
+      "step": 53125
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 7.653576864535768e-06,
+      "loss": 1.9459,
+      "step": 53150
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 7.638356164383563e-06,
+      "loss": 1.9937,
+      "step": 53175
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 7.623135464231355e-06,
+      "loss": 1.9711,
+      "step": 53200
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 7.607914764079149e-06,
+      "loss": 1.9129,
+      "step": 53225
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 7.592694063926941e-06,
+      "loss": 1.9291,
+      "step": 53250
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 7.577473363774734e-06,
+      "loss": 2.0245,
+      "step": 53275
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 7.562252663622527e-06,
+      "loss": 1.9558,
+      "step": 53300
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 7.54703196347032e-06,
+      "loss": 1.9083,
+      "step": 53325
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 7.531811263318113e-06,
+      "loss": 1.9875,
+      "step": 53350
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 7.5165905631659066e-06,
+      "loss": 1.9424,
+      "step": 53375
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 7.5013698630136986e-06,
+      "loss": 1.9449,
+      "step": 53400
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 7.486149162861492e-06,
+      "loss": 1.8812,
+      "step": 53425
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 7.470928462709285e-06,
+      "loss": 1.9874,
+      "step": 53450
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 7.455707762557079e-06,
+      "loss": 2.0413,
+      "step": 53475
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 7.440487062404871e-06,
+      "loss": 1.9994,
+      "step": 53500
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 7.4252663622526645e-06,
+      "loss": 1.9185,
+      "step": 53525
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 7.4100456621004565e-06,
+      "loss": 1.8973,
+      "step": 53550
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 7.39482496194825e-06,
+      "loss": 1.9159,
+      "step": 53575
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 7.379604261796043e-06,
+      "loss": 1.9243,
+      "step": 53600
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 7.364383561643837e-06,
+      "loss": 1.9837,
+      "step": 53625
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 7.349162861491629e-06,
+      "loss": 1.9124,
+      "step": 53650
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 7.333942161339422e-06,
+      "loss": 1.9815,
+      "step": 53675
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 7.318721461187215e-06,
+      "loss": 1.911,
+      "step": 53700
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 7.303500761035008e-06,
+      "loss": 1.934,
+      "step": 53725
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 7.288280060882801e-06,
+      "loss": 1.9157,
+      "step": 53750
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 7.2730593607305946e-06,
+      "loss": 1.9445,
+      "step": 53775
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 7.2578386605783865e-06,
+      "loss": 1.9126,
+      "step": 53800
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 7.24261796042618e-06,
+      "loss": 1.9318,
+      "step": 53825
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 7.227397260273973e-06,
+      "loss": 1.9675,
+      "step": 53850
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 7.212176560121767e-06,
+      "loss": 2.0658,
+      "step": 53875
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 7.196955859969559e-06,
+      "loss": 1.9834,
+      "step": 53900
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 7.1817351598173524e-06,
+      "loss": 1.9381,
+      "step": 53925
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 7.1665144596651444e-06,
+      "loss": 1.9349,
+      "step": 53950
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 7.151293759512938e-06,
+      "loss": 1.981,
+      "step": 53975
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 7.136073059360731e-06,
+      "loss": 1.9228,
+      "step": 54000
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 7.120852359208525e-06,
+      "loss": 1.9982,
+      "step": 54025
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 7.105631659056317e-06,
+      "loss": 1.9066,
+      "step": 54050
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 7.09041095890411e-06,
+      "loss": 1.8969,
+      "step": 54075
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 7.075190258751903e-06,
+      "loss": 1.9243,
+      "step": 54100
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 7.059969558599696e-06,
+      "loss": 2.0025,
+      "step": 54125
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 7.044748858447489e-06,
+      "loss": 1.9632,
+      "step": 54150
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 7.0295281582952825e-06,
+      "loss": 1.9642,
+      "step": 54175
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 7.0143074581430745e-06,
+      "loss": 1.8765,
+      "step": 54200
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 6.999086757990868e-06,
+      "loss": 1.8405,
+      "step": 54225
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 6.983866057838661e-06,
+      "loss": 1.9473,
+      "step": 54250
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 6.968645357686455e-06,
+      "loss": 1.9706,
+      "step": 54275
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 6.953424657534247e-06,
+      "loss": 1.9083,
+      "step": 54300
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 6.93820395738204e-06,
+      "loss": 1.9204,
+      "step": 54325
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 6.922983257229832e-06,
+      "loss": 1.9436,
+      "step": 54350
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 6.907762557077626e-06,
+      "loss": 2.0207,
+      "step": 54375
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 6.892541856925419e-06,
+      "loss": 1.961,
+      "step": 54400
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 6.877321156773213e-06,
+      "loss": 1.8648,
+      "step": 54425
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 6.862100456621005e-06,
+      "loss": 1.9194,
+      "step": 54450
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 6.846879756468798e-06,
+      "loss": 1.9261,
+      "step": 54475
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 6.831659056316591e-06,
+      "loss": 2.0225,
+      "step": 54500
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 6.816438356164384e-06,
+      "loss": 1.9711,
+      "step": 54525
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 6.801217656012177e-06,
+      "loss": 2.0096,
+      "step": 54550
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 6.7859969558599705e-06,
+      "loss": 1.9443,
+      "step": 54575
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 6.7707762557077625e-06,
+      "loss": 1.944,
+      "step": 54600
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 6.755555555555556e-06,
+      "loss": 2.0549,
+      "step": 54625
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 6.740334855403349e-06,
+      "loss": 1.9064,
+      "step": 54650
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 6.72572298325723e-06,
+      "loss": 1.9491,
+      "step": 54675
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 6.710502283105024e-06,
+      "loss": 1.9886,
+      "step": 54700
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 6.695281582952816e-06,
+      "loss": 1.9576,
+      "step": 54725
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 6.68006088280061e-06,
+      "loss": 1.921,
+      "step": 54750
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 6.6648401826484024e-06,
+      "loss": 1.9427,
+      "step": 54775
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 6.649619482496196e-06,
+      "loss": 1.9386,
+      "step": 54800
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 6.634398782343988e-06,
+      "loss": 1.9491,
+      "step": 54825
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 6.619178082191782e-06,
+      "loss": 1.9421,
+      "step": 54850
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 6.603957382039574e-06,
+      "loss": 1.9423,
+      "step": 54875
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 6.5887366818873675e-06,
+      "loss": 1.9382,
+      "step": 54900
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 6.57351598173516e-06,
+      "loss": 1.9891,
+      "step": 54925
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 6.558295281582954e-06,
+      "loss": 1.881,
+      "step": 54950
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 6.543074581430746e-06,
+      "loss": 1.9859,
+      "step": 54975
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 6.52785388127854e-06,
+      "loss": 1.8991,
+      "step": 55000
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 6.5126331811263325e-06,
+      "loss": 1.9316,
+      "step": 55025
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 6.497412480974125e-06,
+      "loss": 1.9139,
+      "step": 55050
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 6.482191780821918e-06,
+      "loss": 1.8429,
+      "step": 55075
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 6.466971080669712e-06,
+      "loss": 1.9872,
+      "step": 55100
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 6.451750380517504e-06,
+      "loss": 1.8164,
+      "step": 55125
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 6.4365296803652976e-06,
+      "loss": 1.9674,
+      "step": 55150
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 6.42130898021309e-06,
+      "loss": 1.9245,
+      "step": 55175
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 6.406088280060884e-06,
+      "loss": 1.9864,
+      "step": 55200
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 6.390867579908676e-06,
+      "loss": 1.9186,
+      "step": 55225
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 6.37564687975647e-06,
+      "loss": 1.9924,
+      "step": 55250
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 6.360426179604262e-06,
+      "loss": 1.9816,
+      "step": 55275
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 6.3452054794520555e-06,
+      "loss": 1.9403,
+      "step": 55300
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 6.3305936073059375e-06,
+      "loss": 1.9687,
+      "step": 55325
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 6.3153729071537295e-06,
+      "loss": 1.9345,
+      "step": 55350
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 6.300152207001523e-06,
+      "loss": 2.0326,
+      "step": 55375
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 6.284931506849315e-06,
+      "loss": 1.9578,
+      "step": 55400
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 6.269710806697109e-06,
+      "loss": 1.8891,
+      "step": 55425
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 6.254490106544902e-06,
+      "loss": 1.9388,
+      "step": 55450
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 6.239269406392695e-06,
+      "loss": 1.8948,
+      "step": 55475
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 6.224048706240487e-06,
+      "loss": 2.0185,
+      "step": 55500
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 6.208828006088281e-06,
+      "loss": 2.0193,
+      "step": 55525
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 6.193607305936074e-06,
+      "loss": 1.9999,
+      "step": 55550
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 6.178386605783867e-06,
+      "loss": 1.9439,
+      "step": 55575
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 6.16316590563166e-06,
+      "loss": 2.025,
+      "step": 55600
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 6.147945205479453e-06,
+      "loss": 1.9343,
+      "step": 55625
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 6.132724505327245e-06,
+      "loss": 1.982,
+      "step": 55650
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 6.117503805175039e-06,
+      "loss": 1.9344,
+      "step": 55675
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 6.102283105022832e-06,
+      "loss": 1.8833,
+      "step": 55700
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 6.087062404870625e-06,
+      "loss": 1.9786,
+      "step": 55725
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 6.0718417047184175e-06,
+      "loss": 1.8689,
+      "step": 55750
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 6.056621004566211e-06,
+      "loss": 1.9735,
+      "step": 55775
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 6.041400304414003e-06,
+      "loss": 1.9081,
+      "step": 55800
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 6.026179604261797e-06,
+      "loss": 1.9836,
+      "step": 55825
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 6.01095890410959e-06,
+      "loss": 1.9072,
+      "step": 55850
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 5.995738203957383e-06,
+      "loss": 1.9328,
+      "step": 55875
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 5.980517503805175e-06,
+      "loss": 1.8841,
+      "step": 55900
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 5.965296803652969e-06,
+      "loss": 1.8813,
+      "step": 55925
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 5.950076103500762e-06,
+      "loss": 1.8905,
+      "step": 55950
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 5.934855403348555e-06,
+      "loss": 1.9598,
+      "step": 55975
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 5.9196347031963476e-06,
+      "loss": 1.9439,
+      "step": 56000
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 5.904414003044141e-06,
+      "loss": 1.8998,
+      "step": 56025
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 5.889193302891933e-06,
+      "loss": 1.8348,
+      "step": 56050
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 5.873972602739727e-06,
+      "loss": 1.8576,
+      "step": 56075
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 5.85875190258752e-06,
+      "loss": 1.9066,
+      "step": 56100
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 5.843531202435313e-06,
+      "loss": 1.9717,
+      "step": 56125
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 5.8283105022831055e-06,
+      "loss": 1.9157,
+      "step": 56150
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 5.813089802130899e-06,
+      "loss": 1.9095,
+      "step": 56175
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 5.797869101978691e-06,
+      "loss": 1.9038,
+      "step": 56200
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 5.782648401826485e-06,
+      "loss": 1.9096,
+      "step": 56225
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 5.767427701674278e-06,
+      "loss": 1.9724,
+      "step": 56250
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 5.752207001522071e-06,
+      "loss": 1.9575,
+      "step": 56275
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 5.736986301369863e-06,
+      "loss": 1.9643,
+      "step": 56300
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 5.721765601217657e-06,
+      "loss": 1.9695,
+      "step": 56325
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 5.70654490106545e-06,
+      "loss": 1.9695,
+      "step": 56350
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 5.691324200913243e-06,
+      "loss": 1.8842,
+      "step": 56375
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 5.6761035007610355e-06,
+      "loss": 1.9068,
+      "step": 56400
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 5.660882800608829e-06,
+      "loss": 1.8677,
+      "step": 56425
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 5.645662100456621e-06,
+      "loss": 1.8634,
+      "step": 56450
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 5.630441400304415e-06,
+      "loss": 1.9824,
+      "step": 56475
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 5.615220700152208e-06,
+      "loss": 1.9314,
+      "step": 56500
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.9477,
+      "step": 56525
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 5.5847792998477934e-06,
+      "loss": 1.9666,
+      "step": 56550
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 5.569558599695587e-06,
+      "loss": 1.8937,
+      "step": 56575
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 5.554337899543379e-06,
+      "loss": 1.9245,
+      "step": 56600
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 5.539117199391173e-06,
+      "loss": 1.864,
+      "step": 56625
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 5.523896499238966e-06,
+      "loss": 1.9335,
+      "step": 56650
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 5.508675799086759e-06,
+      "loss": 1.9818,
+      "step": 56675
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 5.493455098934551e-06,
+      "loss": 1.8967,
+      "step": 56700
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 5.478234398782345e-06,
+      "loss": 1.8585,
+      "step": 56725
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 5.463013698630137e-06,
+      "loss": 1.9773,
+      "step": 56750
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 5.447792998477931e-06,
+      "loss": 1.9371,
+      "step": 56775
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 5.4325722983257235e-06,
+      "loss": 1.9483,
+      "step": 56800
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 5.417351598173517e-06,
+      "loss": 1.8756,
+      "step": 56825
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 5.402130898021309e-06,
+      "loss": 1.7767,
+      "step": 56850
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 5.386910197869103e-06,
+      "loss": 1.952,
+      "step": 56875
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 5.371689497716896e-06,
+      "loss": 1.9177,
+      "step": 56900
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 5.3564687975646886e-06,
+      "loss": 1.9437,
+      "step": 56925
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 5.341248097412481e-06,
+      "loss": 1.9826,
+      "step": 56950
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 5.326027397260275e-06,
+      "loss": 1.944,
+      "step": 56975
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 5.310806697108067e-06,
+      "loss": 1.9131,
+      "step": 57000
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 5.295585996955861e-06,
+      "loss": 1.917,
+      "step": 57025
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 5.280365296803654e-06,
+      "loss": 1.9166,
+      "step": 57050
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 5.265144596651447e-06,
+      "loss": 1.8458,
+      "step": 57075
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 5.249923896499239e-06,
+      "loss": 1.9775,
+      "step": 57100
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 5.234703196347033e-06,
+      "loss": 1.9948,
+      "step": 57125
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 5.219482496194825e-06,
+      "loss": 1.9398,
+      "step": 57150
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 5.204261796042619e-06,
+      "loss": 1.9874,
+      "step": 57175
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 5.1890410958904115e-06,
+      "loss": 1.9431,
+      "step": 57200
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 5.1738203957382035e-06,
+      "loss": 1.9826,
+      "step": 57225
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 5.158599695585997e-06,
+      "loss": 1.8329,
+      "step": 57250
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 5.14337899543379e-06,
+      "loss": 1.9459,
+      "step": 57275
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 5.128158295281584e-06,
+      "loss": 1.8475,
+      "step": 57300
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 5.112937595129376e-06,
+      "loss": 1.9633,
+      "step": 57325
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 5.097716894977169e-06,
+      "loss": 1.879,
+      "step": 57350
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 5.082496194824962e-06,
+      "loss": 1.9553,
+      "step": 57375
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 5.067275494672755e-06,
+      "loss": 1.9861,
+      "step": 57400
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 5.052054794520548e-06,
+      "loss": 1.9381,
+      "step": 57425
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 5.036834094368342e-06,
+      "loss": 1.8685,
+      "step": 57450
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 5.021613394216134e-06,
+      "loss": 1.8962,
+      "step": 57475
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 5.006392694063927e-06,
+      "loss": 1.879,
+      "step": 57500
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 4.99117199391172e-06,
+      "loss": 2.0326,
+      "step": 57525
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 4.975951293759513e-06,
+      "loss": 1.9831,
+      "step": 57550
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 4.960730593607307e-06,
+      "loss": 2.0538,
+      "step": 57575
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 4.9455098934550995e-06,
+      "loss": 1.9686,
+      "step": 57600
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 4.930289193302892e-06,
+      "loss": 1.9546,
+      "step": 57625
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 4.915068493150685e-06,
+      "loss": 2.0175,
+      "step": 57650
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 4.899847792998478e-06,
+      "loss": 1.9396,
+      "step": 57675
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 4.884627092846272e-06,
+      "loss": 1.8997,
+      "step": 57700
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 4.8694063926940645e-06,
+      "loss": 1.9828,
+      "step": 57725
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 4.854185692541857e-06,
+      "loss": 2.0147,
+      "step": 57750
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 4.83896499238965e-06,
+      "loss": 2.0053,
+      "step": 57775
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 4.823744292237443e-06,
+      "loss": 1.9466,
+      "step": 57800
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 4.808523592085237e-06,
+      "loss": 2.0544,
+      "step": 57825
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 4.793911719939118e-06,
+      "loss": 2.0409,
+      "step": 57850
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 4.778691019786911e-06,
+      "loss": 1.9398,
+      "step": 57875
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 4.763470319634704e-06,
+      "loss": 1.961,
+      "step": 57900
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 4.7482496194824965e-06,
+      "loss": 1.9101,
+      "step": 57925
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 4.73302891933029e-06,
+      "loss": 1.9418,
+      "step": 57950
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 4.717808219178083e-06,
+      "loss": 2.0025,
+      "step": 57975
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 4.702587519025876e-06,
+      "loss": 1.8889,
+      "step": 58000
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 4.687366818873669e-06,
+      "loss": 1.9201,
+      "step": 58025
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 4.6721461187214615e-06,
+      "loss": 1.8512,
+      "step": 58050
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 4.656925418569254e-06,
+      "loss": 1.9532,
+      "step": 58075
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 4.641704718417048e-06,
+      "loss": 1.9218,
+      "step": 58100
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 4.626484018264841e-06,
+      "loss": 1.8991,
+      "step": 58125
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 4.611263318112634e-06,
+      "loss": 1.8098,
+      "step": 58150
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 4.5960426179604265e-06,
+      "loss": 1.8689,
+      "step": 58175
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 4.580821917808219e-06,
+      "loss": 1.9573,
+      "step": 58200
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 4.565601217656013e-06,
+      "loss": 1.9308,
+      "step": 58225
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 4.550380517503806e-06,
+      "loss": 1.9812,
+      "step": 58250
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 4.535159817351599e-06,
+      "loss": 1.9183,
+      "step": 58275
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 4.519939117199392e-06,
+      "loss": 1.9293,
+      "step": 58300
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 4.5047184170471844e-06,
+      "loss": 1.8619,
+      "step": 58325
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 4.489497716894978e-06,
+      "loss": 1.9365,
+      "step": 58350
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 4.474277016742771e-06,
+      "loss": 1.961,
+      "step": 58375
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 4.459056316590564e-06,
+      "loss": 1.9869,
+      "step": 58400
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 4.443835616438357e-06,
+      "loss": 1.9984,
+      "step": 58425
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 4.4286149162861495e-06,
+      "loss": 1.9806,
+      "step": 58450
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 4.413394216133942e-06,
+      "loss": 1.9724,
+      "step": 58475
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 4.398173515981736e-06,
+      "loss": 1.9639,
+      "step": 58500
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 4.382952815829529e-06,
+      "loss": 2.0249,
+      "step": 58525
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 4.367732115677322e-06,
+      "loss": 1.9433,
+      "step": 58550
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 4.3525114155251145e-06,
+      "loss": 1.9502,
+      "step": 58575
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 4.337290715372907e-06,
+      "loss": 1.996,
+      "step": 58600
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 4.322070015220701e-06,
+      "loss": 1.9632,
+      "step": 58625
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 4.306849315068494e-06,
+      "loss": 1.9467,
+      "step": 58650
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 4.291628614916287e-06,
+      "loss": 1.915,
+      "step": 58675
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 4.2764079147640796e-06,
+      "loss": 1.9257,
+      "step": 58700
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 4.261187214611872e-06,
+      "loss": 1.9646,
+      "step": 58725
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 4.245966514459665e-06,
+      "loss": 1.9155,
+      "step": 58750
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 4.230745814307459e-06,
+      "loss": 1.8798,
+      "step": 58775
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 4.215525114155252e-06,
+      "loss": 1.9897,
+      "step": 58800
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 4.200304414003045e-06,
+      "loss": 1.9645,
+      "step": 58825
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 4.1850837138508375e-06,
+      "loss": 1.9438,
+      "step": 58850
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 4.16986301369863e-06,
+      "loss": 1.986,
+      "step": 58875
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 4.154642313546424e-06,
+      "loss": 1.9541,
+      "step": 58900
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 4.139421613394217e-06,
+      "loss": 1.9109,
+      "step": 58925
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 4.12420091324201e-06,
+      "loss": 1.9129,
+      "step": 58950
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 4.1089802130898025e-06,
+      "loss": 1.9034,
+      "step": 58975
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 4.093759512937595e-06,
+      "loss": 2.0131,
+      "step": 59000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 4.078538812785389e-06,
+      "loss": 1.8497,
+      "step": 59025
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 4.063318112633182e-06,
+      "loss": 1.9809,
+      "step": 59050
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 4.048097412480975e-06,
+      "loss": 1.9217,
+      "step": 59075
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 4.0328767123287676e-06,
+      "loss": 1.8929,
+      "step": 59100
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 4.01765601217656e-06,
+      "loss": 1.9193,
+      "step": 59125
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 4.002435312024353e-06,
+      "loss": 1.9315,
+      "step": 59150
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 3.987214611872147e-06,
+      "loss": 1.8903,
+      "step": 59175
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 3.97199391171994e-06,
+      "loss": 1.9688,
+      "step": 59200
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 3.956773211567733e-06,
+      "loss": 1.9036,
+      "step": 59225
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 3.9415525114155254e-06,
+      "loss": 1.9615,
+      "step": 59250
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 3.926331811263318e-06,
+      "loss": 1.8522,
+      "step": 59275
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 3.911111111111112e-06,
+      "loss": 1.9418,
+      "step": 59300
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 3.895890410958905e-06,
+      "loss": 1.8732,
+      "step": 59325
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 3.880669710806698e-06,
+      "loss": 1.9244,
+      "step": 59350
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 3.8654490106544905e-06,
+      "loss": 1.9065,
+      "step": 59375
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 3.850228310502283e-06,
+      "loss": 1.9452,
+      "step": 59400
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 3.835007610350077e-06,
+      "loss": 1.9752,
+      "step": 59425
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 3.81978691019787e-06,
+      "loss": 1.9594,
+      "step": 59450
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 3.8045662100456627e-06,
+      "loss": 1.9121,
+      "step": 59475
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 3.7893455098934555e-06,
+      "loss": 1.975,
+      "step": 59500
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 3.774124809741249e-06,
+      "loss": 1.8229,
+      "step": 59525
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 3.7589041095890416e-06,
+      "loss": 1.9183,
+      "step": 59550
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 3.7436834094368345e-06,
+      "loss": 1.824,
+      "step": 59575
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 3.7284627092846277e-06,
+      "loss": 1.9862,
+      "step": 59600
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 3.7132420091324206e-06,
+      "loss": 1.9525,
+      "step": 59625
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 3.6980213089802134e-06,
+      "loss": 1.9417,
+      "step": 59650
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 3.6828006088280067e-06,
+      "loss": 1.9509,
+      "step": 59675
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 3.6675799086757995e-06,
+      "loss": 1.9492,
+      "step": 59700
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 3.652359208523593e-06,
+      "loss": 1.947,
+      "step": 59725
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 3.6371385083713856e-06,
+      "loss": 1.8975,
+      "step": 59750
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 3.6219178082191785e-06,
+      "loss": 1.8844,
+      "step": 59775
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 3.6066971080669717e-06,
+      "loss": 1.9304,
+      "step": 59800
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 3.5914764079147646e-06,
+      "loss": 2.0124,
+      "step": 59825
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 3.5762557077625574e-06,
+      "loss": 1.876,
+      "step": 59850
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 3.5610350076103507e-06,
+      "loss": 1.893,
+      "step": 59875
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 3.5458143074581435e-06,
+      "loss": 1.9703,
+      "step": 59900
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 3.5305936073059368e-06,
+      "loss": 1.9334,
+      "step": 59925
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 3.5153729071537296e-06,
+      "loss": 1.9834,
+      "step": 59950
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 3.5001522070015225e-06,
+      "loss": 1.868,
+      "step": 59975
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 3.4849315068493157e-06,
+      "loss": 1.9806,
+      "step": 60000
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 3.4697108066971086e-06,
+      "loss": 1.9378,
+      "step": 60025
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 3.4544901065449014e-06,
+      "loss": 1.8521,
+      "step": 60050
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 3.4392694063926947e-06,
+      "loss": 1.9511,
+      "step": 60075
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 3.4240487062404875e-06,
+      "loss": 1.8839,
+      "step": 60100
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 3.4088280060882808e-06,
+      "loss": 1.954,
+      "step": 60125
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 3.3936073059360736e-06,
+      "loss": 1.9526,
+      "step": 60150
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 3.3783866057838664e-06,
+      "loss": 1.8578,
+      "step": 60175
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 3.3631659056316597e-06,
+      "loss": 1.9272,
+      "step": 60200
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 3.3479452054794526e-06,
+      "loss": 1.8606,
+      "step": 60225
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 3.3327245053272454e-06,
+      "loss": 1.8964,
+      "step": 60250
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 3.3175038051750387e-06,
+      "loss": 1.9205,
+      "step": 60275
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 3.3022831050228315e-06,
+      "loss": 1.8762,
+      "step": 60300
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 3.2870624048706248e-06,
+      "loss": 1.93,
+      "step": 60325
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 3.2718417047184176e-06,
+      "loss": 2.0007,
+      "step": 60350
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 3.2566210045662104e-06,
+      "loss": 1.935,
+      "step": 60375
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 3.2414003044140037e-06,
+      "loss": 1.9675,
+      "step": 60400
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 3.2261796042617965e-06,
+      "loss": 2.0031,
+      "step": 60425
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 3.2109589041095894e-06,
+      "loss": 1.9624,
+      "step": 60450
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 3.1957382039573826e-06,
+      "loss": 1.9126,
+      "step": 60475
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.1805175038051755e-06,
+      "loss": 1.9268,
+      "step": 60500
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 3.1652968036529687e-06,
+      "loss": 1.8126,
+      "step": 60525
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 3.1500761035007616e-06,
+      "loss": 1.9809,
+      "step": 60550
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 3.1348554033485544e-06,
+      "loss": 1.8864,
+      "step": 60575
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 3.1196347031963477e-06,
+      "loss": 1.9743,
+      "step": 60600
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3.1044140030441405e-06,
+      "loss": 2.0185,
+      "step": 60625
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 3.0891933028919334e-06,
+      "loss": 1.9321,
+      "step": 60650
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 3.0739726027397266e-06,
+      "loss": 1.9233,
+      "step": 60675
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 3.0587519025875195e-06,
+      "loss": 2.0277,
+      "step": 60700
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.0435312024353123e-06,
+      "loss": 1.98,
+      "step": 60725
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 3.0283105022831056e-06,
+      "loss": 1.9194,
+      "step": 60750
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 3.0130898021308984e-06,
+      "loss": 1.9477,
+      "step": 60775
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 2.9978691019786917e-06,
+      "loss": 1.881,
+      "step": 60800
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.9826484018264845e-06,
+      "loss": 1.9989,
+      "step": 60825
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 2.9680365296803653e-06,
+      "loss": 1.941,
+      "step": 60850
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 2.9528158295281586e-06,
+      "loss": 1.9003,
+      "step": 60875
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 2.9375951293759514e-06,
+      "loss": 1.8749,
+      "step": 60900
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.9223744292237442e-06,
+      "loss": 1.8517,
+      "step": 60925
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 2.9071537290715375e-06,
+      "loss": 1.9296,
+      "step": 60950
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 2.8919330289193303e-06,
+      "loss": 1.8825,
+      "step": 60975
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 2.876712328767123e-06,
+      "loss": 1.9291,
+      "step": 61000
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8614916286149164e-06,
+      "loss": 1.9779,
+      "step": 61025
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 2.8462709284627093e-06,
+      "loss": 1.9288,
+      "step": 61050
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 2.8310502283105025e-06,
+      "loss": 1.9261,
+      "step": 61075
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 2.8158295281582954e-06,
+      "loss": 1.939,
+      "step": 61100
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.8006088280060882e-06,
+      "loss": 2.0491,
+      "step": 61125
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 2.7853881278538815e-06,
+      "loss": 1.9455,
+      "step": 61150
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 2.7701674277016743e-06,
+      "loss": 1.913,
+      "step": 61175
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 2.754946727549467e-06,
+      "loss": 1.9188,
+      "step": 61200
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 1.9576,
+      "step": 61225
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.7245053272450533e-06,
+      "loss": 1.8853,
+      "step": 61250
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 2.7092846270928465e-06,
+      "loss": 1.9414,
+      "step": 61275
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 2.6940639269406394e-06,
+      "loss": 1.8567,
+      "step": 61300
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 2.6788432267884322e-06,
+      "loss": 1.9412,
+      "step": 61325
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.6636225266362255e-06,
+      "loss": 1.9092,
+      "step": 61350
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 2.6484018264840183e-06,
+      "loss": 1.9909,
+      "step": 61375
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 2.633181126331811e-06,
+      "loss": 1.9695,
+      "step": 61400
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 2.6179604261796044e-06,
+      "loss": 1.9589,
+      "step": 61425
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.6027397260273973e-06,
+      "loss": 1.9191,
+      "step": 61450
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 2.5875190258751905e-06,
+      "loss": 1.893,
+      "step": 61475
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 2.5722983257229834e-06,
+      "loss": 1.9381,
+      "step": 61500
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.557077625570776e-06,
+      "loss": 1.9433,
+      "step": 61525
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.5418569254185695e-06,
+      "loss": 1.8897,
+      "step": 61550
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 2.5266362252663623e-06,
+      "loss": 1.977,
+      "step": 61575
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 2.511415525114155e-06,
+      "loss": 1.8949,
+      "step": 61600
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 2.4961948249619484e-06,
+      "loss": 1.9551,
+      "step": 61625
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.4809741248097413e-06,
+      "loss": 1.8667,
+      "step": 61650
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 1.8624,
+      "step": 61675
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 2.4505327245053274e-06,
+      "loss": 1.9746,
+      "step": 61700
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 2.43531202435312e-06,
+      "loss": 1.9408,
+      "step": 61725
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 2.4200913242009135e-06,
+      "loss": 1.9824,
+      "step": 61750
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.4048706240487063e-06,
+      "loss": 1.8111,
+      "step": 61775
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 2.389649923896499e-06,
+      "loss": 2.0112,
+      "step": 61800
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 2.3744292237442924e-06,
+      "loss": 1.8705,
+      "step": 61825
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 2.3592085235920852e-06,
+      "loss": 1.9911,
+      "step": 61850
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.343987823439878e-06,
+      "loss": 1.9958,
+      "step": 61875
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 2.3287671232876713e-06,
+      "loss": 1.9827,
+      "step": 61900
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 2.313546423135464e-06,
+      "loss": 1.8441,
+      "step": 61925
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 2.2983257229832575e-06,
+      "loss": 1.9146,
+      "step": 61950
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.2831050228310503e-06,
+      "loss": 1.9028,
+      "step": 61975
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 2.267884322678843e-06,
+      "loss": 1.9987,
+      "step": 62000
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 2.2526636225266364e-06,
+      "loss": 1.8399,
+      "step": 62025
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 2.2374429223744292e-06,
+      "loss": 1.9116,
+      "step": 62050
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.9387,
+      "step": 62075
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 2.2070015220700153e-06,
+      "loss": 1.9983,
+      "step": 62100
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 1.9453,
+      "step": 62125
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 2.1765601217656014e-06,
+      "loss": 1.952,
+      "step": 62150
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.1613394216133943e-06,
+      "loss": 1.9109,
+      "step": 62175
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.146118721461187e-06,
+      "loss": 1.8966,
+      "step": 62200
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 2.1308980213089804e-06,
+      "loss": 1.9123,
+      "step": 62225
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.1156773211567732e-06,
+      "loss": 1.8517,
+      "step": 62250
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.100456621004566e-06,
+      "loss": 1.9353,
+      "step": 62275
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.0852359208523593e-06,
+      "loss": 1.8875,
+      "step": 62300
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 2.070015220700152e-06,
+      "loss": 1.9581,
+      "step": 62325
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.0547945205479454e-06,
+      "loss": 1.925,
+      "step": 62350
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 2.0395738203957383e-06,
+      "loss": 1.9363,
+      "step": 62375
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.024353120243531e-06,
+      "loss": 1.9514,
+      "step": 62400
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.0091324200913244e-06,
+      "loss": 1.8421,
+      "step": 62425
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 1.9939117199391172e-06,
+      "loss": 2.0036,
+      "step": 62450
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 1.97869101978691e-06,
+      "loss": 1.8492,
+      "step": 62475
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.9634703196347033e-06,
+      "loss": 1.8921,
+      "step": 62500
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.948249619482496e-06,
+      "loss": 1.9476,
+      "step": 62525
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 1.9330289193302894e-06,
+      "loss": 1.8792,
+      "step": 62550
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 1.8571,
+      "step": 62575
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 1.9025875190258753e-06,
+      "loss": 1.9357,
+      "step": 62600
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 1.8873668188736682e-06,
+      "loss": 1.8947,
+      "step": 62625
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 1.8721461187214612e-06,
+      "loss": 1.8989,
+      "step": 62650
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 1.8569254185692543e-06,
+      "loss": 1.997,
+      "step": 62675
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 1.8417047184170473e-06,
+      "loss": 1.9384,
+      "step": 62700
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 1.8264840182648401e-06,
+      "loss": 1.8858,
+      "step": 62725
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 1.8112633181126332e-06,
+      "loss": 1.8676,
+      "step": 62750
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 1.7960426179604263e-06,
+      "loss": 1.9975,
+      "step": 62775
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 1.7808219178082193e-06,
+      "loss": 1.9733,
+      "step": 62800
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 1.7656012176560121e-06,
+      "loss": 1.9497,
+      "step": 62825
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 1.7503805175038052e-06,
+      "loss": 1.8918,
+      "step": 62850
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 1.7351598173515982e-06,
+      "loss": 1.9333,
+      "step": 62875
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 1.7199391171993913e-06,
+      "loss": 1.9743,
+      "step": 62900
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 1.7047184170471841e-06,
+      "loss": 1.8864,
+      "step": 62925
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 1.6894977168949772e-06,
+      "loss": 1.9328,
+      "step": 62950
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 1.6742770167427702e-06,
+      "loss": 1.9335,
+      "step": 62975
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 1.6596651445966516e-06,
+      "loss": 1.9752,
+      "step": 63000
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 1.6444444444444447e-06,
+      "loss": 1.9317,
+      "step": 63025
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 1.6292237442922375e-06,
+      "loss": 1.9205,
+      "step": 63050
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 1.6140030441400306e-06,
+      "loss": 1.9616,
+      "step": 63075
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 1.5987823439878236e-06,
+      "loss": 2.047,
+      "step": 63100
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 1.5835616438356167e-06,
+      "loss": 1.8741,
+      "step": 63125
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 1.5683409436834095e-06,
+      "loss": 1.9207,
+      "step": 63150
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 1.5531202435312026e-06,
+      "loss": 1.8698,
+      "step": 63175
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 1.5378995433789956e-06,
+      "loss": 2.013,
+      "step": 63200
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.5226788432267887e-06,
+      "loss": 1.8767,
+      "step": 63225
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 1.5074581430745815e-06,
+      "loss": 1.9414,
+      "step": 63250
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.4922374429223746e-06,
+      "loss": 2.025,
+      "step": 63275
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 1.4770167427701676e-06,
+      "loss": 1.9642,
+      "step": 63300
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.4617960426179607e-06,
+      "loss": 1.9146,
+      "step": 63325
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.4465753424657535e-06,
+      "loss": 1.8681,
+      "step": 63350
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.4313546423135466e-06,
+      "loss": 1.8663,
+      "step": 63375
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.4161339421613396e-06,
+      "loss": 1.9674,
+      "step": 63400
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.4009132420091327e-06,
+      "loss": 1.9506,
+      "step": 63425
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.3856925418569255e-06,
+      "loss": 1.9306,
+      "step": 63450
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.3704718417047186e-06,
+      "loss": 1.9052,
+      "step": 63475
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.3552511415525116e-06,
+      "loss": 1.9244,
+      "step": 63500
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 1.3400304414003047e-06,
+      "loss": 1.9635,
+      "step": 63525
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.3248097412480975e-06,
+      "loss": 1.9551,
+      "step": 63550
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.3095890410958906e-06,
+      "loss": 1.9475,
+      "step": 63575
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.2943683409436836e-06,
+      "loss": 1.9417,
+      "step": 63600
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.2791476407914767e-06,
+      "loss": 1.9187,
+      "step": 63625
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.2639269406392695e-06,
+      "loss": 2.0175,
+      "step": 63650
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.2487062404870626e-06,
+      "loss": 1.9486,
+      "step": 63675
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.2334855403348556e-06,
+      "loss": 1.8935,
+      "step": 63700
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.2182648401826487e-06,
+      "loss": 1.9383,
+      "step": 63725
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.2030441400304415e-06,
+      "loss": 1.9645,
+      "step": 63750
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.1878234398782346e-06,
+      "loss": 1.9119,
+      "step": 63775
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.1726027397260276e-06,
+      "loss": 1.8621,
+      "step": 63800
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 1.1573820395738207e-06,
+      "loss": 1.7967,
+      "step": 63825
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.1421613394216135e-06,
+      "loss": 1.9894,
+      "step": 63850
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.1269406392694063e-06,
+      "loss": 1.8729,
+      "step": 63875
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 1.1117199391171994e-06,
+      "loss": 1.9467,
+      "step": 63900
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.0964992389649924e-06,
+      "loss": 1.9585,
+      "step": 63925
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.0812785388127855e-06,
+      "loss": 1.9316,
+      "step": 63950
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.0660578386605783e-06,
+      "loss": 1.8996,
+      "step": 63975
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 1.0508371385083714e-06,
+      "loss": 1.8568,
+      "step": 64000
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 1.0356164383561644e-06,
+      "loss": 1.924,
+      "step": 64025
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 1.0203957382039575e-06,
+      "loss": 1.8947,
+      "step": 64050
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 1.0051750380517503e-06,
+      "loss": 2.0517,
+      "step": 64075
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 9.899543378995434e-07,
+      "loss": 2.0314,
+      "step": 64100
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 9.747336377473364e-07,
+      "loss": 1.8407,
+      "step": 64125
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 9.595129375951295e-07,
+      "loss": 2.0457,
+      "step": 64150
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 9.442922374429224e-07,
+      "loss": 1.9252,
+      "step": 64175
+    },
+    {
+      "epoch": 19.54,
+      "learning_rate": 9.290715372907154e-07,
+      "loss": 1.9663,
+      "step": 64200
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 9.138508371385084e-07,
+      "loss": 2.0035,
+      "step": 64225
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 8.986301369863014e-07,
+      "loss": 1.8369,
+      "step": 64250
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 8.834094368340944e-07,
+      "loss": 1.89,
+      "step": 64275
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 8.681887366818874e-07,
+      "loss": 1.813,
+      "step": 64300
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 8.529680365296804e-07,
+      "loss": 1.9395,
+      "step": 64325
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 8.377473363774734e-07,
+      "loss": 1.9558,
+      "step": 64350
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 8.225266362252664e-07,
+      "loss": 1.9236,
+      "step": 64375
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 8.073059360730594e-07,
+      "loss": 1.8632,
+      "step": 64400
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 7.920852359208524e-07,
+      "loss": 1.8458,
+      "step": 64425
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 7.768645357686454e-07,
+      "loss": 1.8997,
+      "step": 64450
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 7.616438356164384e-07,
+      "loss": 1.9626,
+      "step": 64475
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 7.464231354642314e-07,
+      "loss": 1.8967,
+      "step": 64500
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 7.312024353120244e-07,
+      "loss": 1.961,
+      "step": 64525
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 7.159817351598174e-07,
+      "loss": 1.8946,
+      "step": 64550
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 7.007610350076104e-07,
+      "loss": 1.9476,
+      "step": 64575
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 6.855403348554034e-07,
+      "loss": 1.9353,
+      "step": 64600
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 6.703196347031964e-07,
+      "loss": 1.9945,
+      "step": 64625
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 6.550989345509894e-07,
+      "loss": 1.9124,
+      "step": 64650
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 6.398782343987824e-07,
+      "loss": 2.0121,
+      "step": 64675
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 6.246575342465755e-07,
+      "loss": 1.9495,
+      "step": 64700
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 6.094368340943684e-07,
+      "loss": 1.9243,
+      "step": 64725
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 5.942161339421615e-07,
+      "loss": 1.9793,
+      "step": 64750
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 5.789954337899544e-07,
+      "loss": 1.8846,
+      "step": 64775
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 5.637747336377475e-07,
+      "loss": 1.8872,
+      "step": 64800
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 5.485540334855404e-07,
+      "loss": 1.9736,
+      "step": 64825
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.8818,
+      "step": 64850
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 5.181126331811263e-07,
+      "loss": 1.9747,
+      "step": 64875
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 5.028919330289193e-07,
+      "loss": 1.9198,
+      "step": 64900
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 4.876712328767123e-07,
+      "loss": 1.8793,
+      "step": 64925
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 4.7245053272450534e-07,
+      "loss": 1.9494,
+      "step": 64950
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 4.5722983257229834e-07,
+      "loss": 1.9567,
+      "step": 64975
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 4.4200913242009134e-07,
+      "loss": 1.8885,
+      "step": 65000
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 4.2678843226788434e-07,
+      "loss": 1.946,
+      "step": 65025
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 4.1156773211567733e-07,
+      "loss": 1.9554,
+      "step": 65050
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 3.9634703196347033e-07,
+      "loss": 1.8447,
+      "step": 65075
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 3.8112633181126333e-07,
+      "loss": 1.9562,
+      "step": 65100
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 3.6590563165905633e-07,
+      "loss": 1.9531,
+      "step": 65125
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 3.5068493150684933e-07,
+      "loss": 1.9695,
+      "step": 65150
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 3.3546423135464233e-07,
+      "loss": 1.9478,
+      "step": 65175
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 3.2024353120243533e-07,
+      "loss": 2.0431,
+      "step": 65200
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 3.050228310502283e-07,
+      "loss": 1.8231,
+      "step": 65225
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 2.898021308980213e-07,
+      "loss": 1.8677,
+      "step": 65250
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 2.745814307458143e-07,
+      "loss": 2.0174,
+      "step": 65275
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 2.593607305936073e-07,
+      "loss": 1.9269,
+      "step": 65300
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.441400304414003e-07,
+      "loss": 1.8943,
+      "step": 65325
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 2.2891933028919335e-07,
+      "loss": 1.9661,
+      "step": 65350
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 2.1369863013698635e-07,
+      "loss": 1.9616,
+      "step": 65375
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 1.984779299847793e-07,
+      "loss": 1.8594,
+      "step": 65400
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 1.832572298325723e-07,
+      "loss": 1.8842,
+      "step": 65425
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 1.680365296803653e-07,
+      "loss": 2.032,
+      "step": 65450
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 1.5281582952815831e-07,
+      "loss": 1.8597,
+      "step": 65475
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 1.375951293759513e-07,
+      "loss": 1.9084,
+      "step": 65500
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.223744292237443e-07,
+      "loss": 1.8235,
+      "step": 65525
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.0715372907153731e-07,
+      "loss": 1.8212,
+      "step": 65550
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 9.193302891933031e-08,
+      "loss": 1.8892,
+      "step": 65575
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 7.67123287671233e-08,
+      "loss": 1.9391,
+      "step": 65600
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 6.149162861491629e-08,
+      "loss": 1.922,
+      "step": 65625
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 4.687975646879757e-08,
+      "loss": 1.8743,
+      "step": 65650
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 3.1659056316590565e-08,
+      "loss": 1.8869,
+      "step": 65675
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.6438356164383564e-08,
+      "loss": 2.0741,
+      "step": 65700
+    },
+    {
+      "epoch": 20.0,
+      "step": 65700,
+      "total_flos": 1.138795714927657e+18,
+      "train_loss": 2.126255558368096,
+      "train_runtime": 26711.3051,
+      "train_samples_per_second": 14.755,
+      "train_steps_per_second": 2.46
+    }
+  ],
+  "max_steps": 65700,
+  "num_train_epochs": 20,
+  "total_flos": 1.138795714927657e+18,
+  "trial_name": null,
+  "trial_params": null
+}