diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5334 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9997456980588284,
+  "eval_steps": 500,
+  "global_step": 17694,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.90960451977401e-08,
+      "loss": 0.0044,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.581920903954802e-07,
+      "loss": 0.0051,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3728813559322033e-07,
+      "loss": 0.007,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.163841807909604e-07,
+      "loss": 0.0038,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9548022598870054e-07,
+      "loss": 0.0052,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7457627118644066e-07,
+      "loss": 0.0037,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.536723163841808e-07,
+      "loss": 0.0045,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.327683615819209e-07,
+      "loss": 0.0057,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.118644067796611e-07,
+      "loss": 0.0063,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.909604519774011e-07,
+      "loss": 0.0054,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.700564971751413e-07,
+      "loss": 0.0041,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.491525423728813e-07,
+      "loss": 0.0061,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0282485875706214e-06,
+      "loss": 0.0054,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1073446327683617e-06,
+      "loss": 0.0046,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1864406779661017e-06,
+      "loss": 0.0083,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2655367231638417e-06,
+      "loss": 0.0041,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.344632768361582e-06,
+      "loss": 0.0054,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4237288135593222e-06,
+      "loss": 0.0043,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5028248587570622e-06,
+      "loss": 0.0058,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5819209039548022e-06,
+      "loss": 0.0057,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6610169491525424e-06,
+      "loss": 0.0042,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.7401129943502826e-06,
+      "loss": 0.0036,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8192090395480224e-06,
+      "loss": 0.0045,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8983050847457627e-06,
+      "loss": 0.0056,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977401129943503e-06,
+      "loss": 0.0054,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.056497175141243e-06,
+      "loss": 0.0057,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1355932203389833e-06,
+      "loss": 0.0064,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.2146892655367233e-06,
+      "loss": 0.0045,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.293785310734463e-06,
+      "loss": 0.0048,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.3728813559322034e-06,
+      "loss": 0.0049,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.4519774011299434e-06,
+      "loss": 0.0031,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.5310734463276834e-06,
+      "loss": 0.0039,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.610169491525424e-06,
+      "loss": 0.0063,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.689265536723164e-06,
+      "loss": 0.0051,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.768361581920904e-06,
+      "loss": 0.0036,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8474576271186443e-06,
+      "loss": 0.0057,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.926553672316384e-06,
+      "loss": 0.0047,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.0056497175141243e-06,
+      "loss": 0.0067,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.0847457627118643e-06,
+      "loss": 0.0057,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.1638418079096044e-06,
+      "loss": 0.0051,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.242937853107345e-06,
+      "loss": 0.0036,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.322033898305085e-06,
+      "loss": 0.0029,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.401129943502825e-06,
+      "loss": 0.0049,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.4802259887005653e-06,
+      "loss": 0.0048,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.559322033898305e-06,
+      "loss": 0.0045,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.638418079096045e-06,
+      "loss": 0.0043,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.7175141242937853e-06,
+      "loss": 0.0048,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.7966101694915253e-06,
+      "loss": 0.0035,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.875706214689265e-06,
+      "loss": 0.0045,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.954802259887006e-06,
+      "loss": 0.0042,
+      "step": 1000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.033898305084746e-06,
+      "loss": 0.0064,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.112994350282486e-06,
+      "loss": 0.0056,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.192090395480226e-06,
+      "loss": 0.0045,
+      "step": 1060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.271186440677967e-06,
+      "loss": 0.0038,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.350282485875706e-06,
+      "loss": 0.0054,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.429378531073447e-06,
+      "loss": 0.0045,
+      "step": 1120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.508474576271186e-06,
+      "loss": 0.0037,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.587570621468926e-06,
+      "loss": 0.0049,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.666666666666666e-06,
+      "loss": 0.0056,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.745762711864407e-06,
+      "loss": 0.0043,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.824858757062146e-06,
+      "loss": 0.0077,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.903954802259887e-06,
+      "loss": 0.0046,
+      "step": 1240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.983050847457627e-06,
+      "loss": 0.0072,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.062146892655367e-06,
+      "loss": 0.005,
+      "step": 1280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.141242937853107e-06,
+      "loss": 0.003,
+      "step": 1300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.220338983050848e-06,
+      "loss": 0.0037,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.299435028248587e-06,
+      "loss": 0.0059,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.378531073446328e-06,
+      "loss": 0.0065,
+      "step": 1360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.457627118644068e-06,
+      "loss": 0.0074,
+      "step": 1380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.536723163841808e-06,
+      "loss": 0.0066,
+      "step": 1400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.615819209039548e-06,
+      "loss": 0.0044,
+      "step": 1420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.694915254237289e-06,
+      "loss": 0.006,
+      "step": 1440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.774011299435028e-06,
+      "loss": 0.0051,
+      "step": 1460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.853107344632768e-06,
+      "loss": 0.0048,
+      "step": 1480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.932203389830508e-06,
+      "loss": 0.0067,
+      "step": 1500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.011299435028249e-06,
+      "loss": 0.0039,
+      "step": 1520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.090395480225988e-06,
+      "loss": 0.0055,
+      "step": 1540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.169491525423729e-06,
+      "loss": 0.0073,
+      "step": 1560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.248587570621469e-06,
+      "loss": 0.0063,
+      "step": 1580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.327683615819209e-06,
+      "loss": 0.0064,
+      "step": 1600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.406779661016949e-06,
+      "loss": 0.0058,
+      "step": 1620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.48587570621469e-06,
+      "loss": 0.0066,
+      "step": 1640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.564971751412429e-06,
+      "loss": 0.005,
+      "step": 1660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 6.64406779661017e-06,
+      "loss": 0.0071,
+      "step": 1680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.72316384180791e-06,
+      "loss": 0.009,
+      "step": 1700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.80225988700565e-06,
+      "loss": 0.0047,
+      "step": 1720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 6.88135593220339e-06,
+      "loss": 0.004,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.9604519774011305e-06,
+      "loss": 0.0058,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.999993188648143e-06,
+      "loss": 0.0054,
+      "step": 1780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999938697992356e-06,
+      "loss": 0.007,
+      "step": 1800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999829717529136e-06,
+      "loss": 0.0062,
+      "step": 1820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.999666248955175e-06,
+      "loss": 0.0042,
+      "step": 1840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.999448294815481e-06,
+      "loss": 0.0052,
+      "step": 1860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.999175858503339e-06,
+      "loss": 0.0054,
+      "step": 1880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.99884894426025e-06,
+      "loss": 0.0055,
+      "step": 1900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.998467557175874e-06,
+      "loss": 0.0052,
+      "step": 1920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.998031703187944e-06,
+      "loss": 0.0092,
+      "step": 1940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.997541389082182e-06,
+      "loss": 0.0061,
+      "step": 1960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.996996622492182e-06,
+      "loss": 0.006,
+      "step": 1980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.996397411899301e-06,
+      "loss": 0.0083,
+      "step": 2000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.995743766632524e-06,
+      "loss": 0.0053,
+      "step": 2020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.995035696868314e-06,
+      "loss": 0.006,
+      "step": 2040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.99427321363046e-06,
+      "loss": 0.0059,
+      "step": 2060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.993456328789906e-06,
+      "loss": 0.0066,
+      "step": 2080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.992585055064556e-06,
+      "loss": 0.0061,
+      "step": 2100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.991659406019087e-06,
+      "loss": 0.006,
+      "step": 2120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 6.990679396064737e-06,
+      "loss": 0.0041,
+      "step": 2140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.989645040459071e-06,
+      "loss": 0.0047,
+      "step": 2160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.988556355305755e-06,
+      "loss": 0.0053,
+      "step": 2180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.987413357554298e-06,
+      "loss": 0.0084,
+      "step": 2200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.986216064999791e-06,
+      "loss": 0.0091,
+      "step": 2220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9849644962826295e-06,
+      "loss": 0.0074,
+      "step": 2240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9836586708882216e-06,
+      "loss": 0.008,
+      "step": 2260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9822986091466896e-06,
+      "loss": 0.0042,
+      "step": 2280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980884332232547e-06,
+      "loss": 0.0049,
+      "step": 2300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979415862164372e-06,
+      "loss": 0.0068,
+      "step": 2320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.977893221804466e-06,
+      "loss": 0.0051,
+      "step": 2340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.9763164348584956e-06,
+      "loss": 0.0047,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.974685525875124e-06,
+      "loss": 0.007,
+      "step": 2380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9730005202456296e-06,
+      "loss": 0.0099,
+      "step": 2400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.97126144420351e-06,
+      "loss": 0.0067,
+      "step": 2420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.969468324824072e-06,
+      "loss": 0.0047,
+      "step": 2440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.967621190024015e-06,
+      "loss": 0.0048,
+      "step": 2460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.9657200685609895e-06,
+      "loss": 0.0048,
+      "step": 2480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.963764990033155e-06,
+      "loss": 0.0039,
+      "step": 2500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.961755984878717e-06,
+      "loss": 0.0048,
+      "step": 2520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.959693084375453e-06,
+      "loss": 0.005,
+      "step": 2540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.9575763206402245e-06,
+      "loss": 0.0049,
+      "step": 2560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.955405726628483e-06,
+      "loss": 0.0051,
+      "step": 2580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9531813361337455e-06,
+      "loss": 0.0052,
+      "step": 2600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.95090318378708e-06,
+      "loss": 0.0052,
+      "step": 2620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.94857130505656e-06,
+      "loss": 0.0063,
+      "step": 2640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.946185736246713e-06,
+      "loss": 0.0068,
+      "step": 2660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.943746514497957e-06,
+      "loss": 0.0074,
+      "step": 2680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.9412536777860215e-06,
+      "loss": 0.0047,
+      "step": 2700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.938707264921354e-06,
+      "loss": 0.0073,
+      "step": 2720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.936107315548521e-06,
+      "loss": 0.0044,
+      "step": 2740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.933453870145586e-06,
+      "loss": 0.006,
+      "step": 2760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.930746970023479e-06,
+      "loss": 0.0042,
+      "step": 2780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.927986657325359e-06,
+      "loss": 0.0042,
+      "step": 2800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.925172975025951e-06,
+      "loss": 0.004,
+      "step": 2820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.922305966930881e-06,
+      "loss": 0.0057,
+      "step": 2840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.919385677675995e-06,
+      "loss": 0.0066,
+      "step": 2860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.916412152726659e-06,
+      "loss": 0.0057,
+      "step": 2880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.913385438377057e-06,
+      "loss": 0.0069,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.910305581749467e-06,
+      "loss": 0.0045,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.907172630793524e-06,
+      "loss": 0.0048,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.903986634285483e-06,
+      "loss": 0.006,
+      "step": 2960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.900747641827452e-06,
+      "loss": 0.0082,
+      "step": 2980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.897455703846622e-06,
+      "loss": 0.0067,
+      "step": 3000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 6.894110871594477e-06,
+      "loss": 0.0053,
+      "step": 3020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.890713197146011e-06,
+      "loss": 0.0046,
+      "step": 3040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.887262733398897e-06,
+      "loss": 0.0058,
+      "step": 3060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 6.883759534072677e-06,
+      "loss": 0.0064,
+      "step": 3080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.880203653707924e-06,
+      "loss": 0.0046,
+      "step": 3100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.87659514766539e-06,
+      "loss": 0.0036,
+      "step": 3120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 6.872934072125144e-06,
+      "loss": 0.0037,
+      "step": 3140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.8692204840856964e-06,
+      "loss": 0.0048,
+      "step": 3160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.8654544413631205e-06,
+      "loss": 0.0049,
+      "step": 3180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.8616360025901385e-06,
+      "loss": 0.0073,
+      "step": 3200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.8577652272152175e-06,
+      "loss": 0.0067,
+      "step": 3220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.853842175501646e-06,
+      "loss": 0.0049,
+      "step": 3240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.849866908526585e-06,
+      "loss": 0.0063,
+      "step": 3260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.845839488180131e-06,
+      "loss": 0.0046,
+      "step": 3280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.841759977164341e-06,
+      "loss": 0.0054,
+      "step": 3300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.837628438992261e-06,
+      "loss": 0.0063,
+      "step": 3320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.833444937986937e-06,
+      "loss": 0.0034,
+      "step": 3340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.829209539280415e-06,
+      "loss": 0.0045,
+      "step": 3360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.8249223088127236e-06,
+      "loss": 0.006,
+      "step": 3380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.8205833133308505e-06,
+      "loss": 0.0065,
+      "step": 3400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.816192620387699e-06,
+      "loss": 0.0079,
+      "step": 3420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.811750298341044e-06,
+      "loss": 0.0053,
+      "step": 3440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.80725641635246e-06,
+      "loss": 0.0053,
+      "step": 3460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.802711044386247e-06,
+      "loss": 0.0056,
+      "step": 3480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.798114253208343e-06,
+      "loss": 0.0065,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.79346611438522e-06,
+      "loss": 0.0055,
+      "step": 3520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.788766700282773e-06,
+      "loss": 0.0066,
+      "step": 3540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.784016084065186e-06,
+      "loss": 0.007,
+      "step": 3560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.779214339693803e-06,
+      "loss": 0.0074,
+      "step": 3580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.774361541925969e-06,
+      "loss": 0.0072,
+      "step": 3600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.769457766313867e-06,
+      "loss": 0.0046,
+      "step": 3620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764503089203349e-06,
+      "loss": 0.0058,
+      "step": 3640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759497587732735e-06,
+      "loss": 0.0051,
+      "step": 3660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754441339831623e-06,
+      "loss": 0.0033,
+      "step": 3680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.749334424219671e-06,
+      "loss": 0.0057,
+      "step": 3700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.744176920405369e-06,
+      "loss": 0.004,
+      "step": 3720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.738968908684806e-06,
+      "loss": 0.0055,
+      "step": 3740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.733710470140421e-06,
+      "loss": 0.006,
+      "step": 3760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.728401686639732e-06,
+      "loss": 0.0063,
+      "step": 3780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.723042640834071e-06,
+      "loss": 0.0062,
+      "step": 3800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.7176334161572906e-06,
+      "loss": 0.0054,
+      "step": 3820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.7121740968244705e-06,
+      "loss": 0.0043,
+      "step": 3840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.706664767830601e-06,
+      "loss": 0.0057,
+      "step": 3860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.701105514949264e-06,
+      "loss": 0.0035,
+      "step": 3880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.695496424731294e-06,
+      "loss": 0.0049,
+      "step": 3900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.689837584503435e-06,
+      "loss": 0.0047,
+      "step": 3920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.6841290823669765e-06,
+      "loss": 0.0051,
+      "step": 3940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.678371007196384e-06,
+      "loss": 0.006,
+      "step": 3960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.672563448637914e-06,
+      "loss": 0.0045,
+      "step": 3980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.666706497108224e-06,
+      "loss": 0.0043,
+      "step": 4000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.660800243792955e-06,
+      "loss": 0.0042,
+      "step": 4020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.65484478064532e-06,
+      "loss": 0.0061,
+      "step": 4040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.648840200384668e-06,
+      "loss": 0.0047,
+      "step": 4060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6427865964950436e-06,
+      "loss": 0.0047,
+      "step": 4080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.636684063223728e-06,
+      "loss": 0.0051,
+      "step": 4100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.630532695579776e-06,
+      "loss": 0.0087,
+      "step": 4120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.624332589332532e-06,
+      "loss": 0.0049,
+      "step": 4140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.618083841010143e-06,
+      "loss": 0.006,
+      "step": 4160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.611786547898053e-06,
+      "loss": 0.0057,
+      "step": 4180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.6054408080374915e-06,
+      "loss": 0.008,
+      "step": 4200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.5990467202239436e-06,
+      "loss": 0.0054,
+      "step": 4220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.592604384005612e-06,
+      "loss": 0.0052,
+      "step": 4240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.586113899681874e-06,
+      "loss": 0.0066,
+      "step": 4260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.579575368301708e-06,
+      "loss": 0.0066,
+      "step": 4280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.5729888916621325e-06,
+      "loss": 0.0042,
+      "step": 4300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.566354572306613e-06,
+      "loss": 0.0046,
+      "step": 4320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.559672513523466e-06,
+      "loss": 0.0049,
+      "step": 4340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.552942819344257e-06,
+      "loss": 0.0059,
+      "step": 4360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.546165594542176e-06,
+      "loss": 0.0042,
+      "step": 4380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.539340944630405e-06,
+      "loss": 0.0039,
+      "step": 4400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.532468975860482e-06,
+      "loss": 0.0056,
+      "step": 4420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.525549795220636e-06,
+      "loss": 0.0056,
+      "step": 4440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.518583510434131e-06,
+      "loss": 0.0058,
+      "step": 4460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.51157022995759e-06,
+      "loss": 0.0052,
+      "step": 4480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.5045100629792925e-06,
+      "loss": 0.0044,
+      "step": 4500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.497403119417493e-06,
+      "loss": 0.0059,
+      "step": 4520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4902495099186945e-06,
+      "loss": 0.004,
+      "step": 4540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.483049345855935e-06,
+      "loss": 0.0044,
+      "step": 4560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.475802739327051e-06,
+      "loss": 0.0033,
+      "step": 4580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.468509803152932e-06,
+      "loss": 0.0058,
+      "step": 4600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.46117065087576e-06,
+      "loss": 0.004,
+      "step": 4620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.4537853967572524e-06,
+      "loss": 0.0047,
+      "step": 4640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.446354155776871e-06,
+      "loss": 0.0059,
+      "step": 4660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.43887704363004e-06,
+      "loss": 0.0046,
+      "step": 4680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.43135417672634e-06,
+      "loss": 0.0046,
+      "step": 4700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.423785672187702e-06,
+      "loss": 0.0058,
+      "step": 4720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.416171647846575e-06,
+      "loss": 0.003,
+      "step": 4740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.408512222244099e-06,
+      "loss": 0.0047,
+      "step": 4760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.400807514628255e-06,
+      "loss": 0.0042,
+      "step": 4780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 6.393057644952012e-06,
+      "loss": 0.0069,
+      "step": 4800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.385262733871454e-06,
+      "loss": 0.0063,
+      "step": 4820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.3774229027439105e-06,
+      "loss": 0.006,
+      "step": 4840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 6.369538273626057e-06,
+      "loss": 0.0044,
+      "step": 4860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.361608969272023e-06,
+      "loss": 0.0056,
+      "step": 4880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.353635113131475e-06,
+      "loss": 0.0031,
+      "step": 4900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.345616829347697e-06,
+      "loss": 0.0062,
+      "step": 4920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3375542427556575e-06,
+      "loss": 0.0052,
+      "step": 4940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3294474788800685e-06,
+      "loss": 0.0046,
+      "step": 4960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3212966639334254e-06,
+      "loss": 0.006,
+      "step": 4980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.313101924814048e-06,
+      "loss": 0.0053,
+      "step": 5000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.304863389104101e-06,
+      "loss": 0.0048,
+      "step": 5020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.29658118506761e-06,
+      "loss": 0.0062,
+      "step": 5040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.288255441648463e-06,
+      "loss": 0.0047,
+      "step": 5060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.279886288468401e-06,
+      "loss": 0.0063,
+      "step": 5080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.271473855825007e-06,
+      "loss": 0.0048,
+      "step": 5100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.26301827468967e-06,
+      "loss": 0.0044,
+      "step": 5120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.25451967670555e-06,
+      "loss": 0.0052,
+      "step": 5140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.245978194185526e-06,
+      "loss": 0.0045,
+      "step": 5160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.237393960110137e-06,
+      "loss": 0.0051,
+      "step": 5180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.228767108125514e-06,
+      "loss": 0.0062,
+      "step": 5200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.220097772541296e-06,
+      "loss": 0.0065,
+      "step": 5220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.211386088328541e-06,
+      "loss": 0.0029,
+      "step": 5240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.202632191117622e-06,
+      "loss": 0.0048,
+      "step": 5260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.193836217196118e-06,
+      "loss": 0.0043,
+      "step": 5280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.184998303506693e-06,
+      "loss": 0.0045,
+      "step": 5300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.176118587644961e-06,
+      "loss": 0.0053,
+      "step": 5320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.167197207857345e-06,
+      "loss": 0.0038,
+      "step": 5340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.1582343030389265e-06,
+      "loss": 0.007,
+      "step": 5360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.1492300127312795e-06,
+      "loss": 0.005,
+      "step": 5380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.1401844771203e-06,
+      "loss": 0.0053,
+      "step": 5400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.1310978370340255e-06,
+      "loss": 0.0048,
+      "step": 5420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.121970233940439e-06,
+      "loss": 0.0056,
+      "step": 5440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.112801809945266e-06,
+      "loss": 0.0051,
+      "step": 5460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.1035927077897695e-06,
+      "loss": 0.0049,
+      "step": 5480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.094343070848516e-06,
+      "loss": 0.0072,
+      "step": 5500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.085053043127153e-06,
+      "loss": 0.0063,
+      "step": 5520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.075722769260162e-06,
+      "loss": 0.0039,
+      "step": 5540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.066352394508611e-06,
+      "loss": 0.004,
+      "step": 5560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.056942064757886e-06,
+      "loss": 0.0042,
+      "step": 5580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.047491926515429e-06,
+      "loss": 0.0064,
+      "step": 5600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.038002126908445e-06,
+      "loss": 0.004,
+      "step": 5620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.028472813681627e-06,
+      "loss": 0.0054,
+      "step": 5640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.018904135194841e-06,
+      "loss": 0.0047,
+      "step": 5660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.009296240420823e-06,
+      "loss": 0.006,
+      "step": 5680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.999649278942862e-06,
+      "loss": 0.0061,
+      "step": 5700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.989963400952467e-06,
+      "loss": 0.0048,
+      "step": 5720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9802387572470285e-06,
+      "loss": 0.0052,
+      "step": 5740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.970475499227474e-06,
+      "loss": 0.0066,
+      "step": 5760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9606737788959075e-06,
+      "loss": 0.0047,
+      "step": 5780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.950833748853247e-06,
+      "loss": 0.0062,
+      "step": 5800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9409555622968434e-06,
+      "loss": 0.0073,
+      "step": 5820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.931039373018102e-06,
+      "loss": 0.0058,
+      "step": 5840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9210853354000806e-06,
+      "loss": 0.0066,
+      "step": 5860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.911093604415092e-06,
+      "loss": 0.0035,
+      "step": 5880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.901064335622288e-06,
+      "loss": 0.0035,
+      "step": 5900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.89099768516524e-06,
+      "loss": 0.0036,
+      "step": 5920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.880893809769506e-06,
+      "loss": 0.0043,
+      "step": 5940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.87075286674019e-06,
+      "loss": 0.004,
+      "step": 5960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.860575013959496e-06,
+      "loss": 0.003,
+      "step": 5980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.850360409884268e-06,
+      "loss": 0.0031,
+      "step": 6000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.840109213543521e-06,
+      "loss": 0.0048,
+      "step": 6020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.829821584535971e-06,
+      "loss": 0.0043,
+      "step": 6040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.819497683027542e-06,
+      "loss": 0.0037,
+      "step": 6060
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.809137669748881e-06,
+      "loss": 0.0041,
+      "step": 6080
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.798741705992847e-06,
+      "loss": 0.0036,
+      "step": 6100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.78830995361201e-06,
+      "loss": 0.0036,
+      "step": 6120
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.777842575016122e-06,
+      "loss": 0.0033,
+      "step": 6140
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.767339733169594e-06,
+      "loss": 0.0028,
+      "step": 6160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.756801591588952e-06,
+      "loss": 0.0031,
+      "step": 6180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.746228314340302e-06,
+      "loss": 0.0036,
+      "step": 6200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.735620066036768e-06,
+      "loss": 0.0022,
+      "step": 6220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.72497701183593e-06,
+      "loss": 0.0035,
+      "step": 6240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.714299317437254e-06,
+      "loss": 0.0041,
+      "step": 6260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.703587149079512e-06,
+      "loss": 0.0029,
+      "step": 6280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.692840673538193e-06,
+      "loss": 0.0029,
+      "step": 6300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.68206005812291e-06,
+      "loss": 0.0029,
+      "step": 6320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.671245470674791e-06,
+      "loss": 0.0041,
+      "step": 6340
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.660397079563865e-06,
+      "loss": 0.0045,
+      "step": 6360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.649515053686447e-06,
+      "loss": 0.0036,
+      "step": 6380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.638599562462502e-06,
+      "loss": 0.0044,
+      "step": 6400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.62765077583301e-06,
+      "loss": 0.0059,
+      "step": 6420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.616668864257323e-06,
+      "loss": 0.0045,
+      "step": 6440
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.6056539987105045e-06,
+      "loss": 0.0032,
+      "step": 6460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.5946063506806724e-06,
+      "loss": 0.0033,
+      "step": 6480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.583526092166331e-06,
+      "loss": 0.0038,
+      "step": 6500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.572413395673688e-06,
+      "loss": 0.0027,
+      "step": 6520
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.561268434213971e-06,
+      "loss": 0.0027,
+      "step": 6540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.550091381300732e-06,
+      "loss": 0.0038,
+      "step": 6560
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.538882410947153e-06,
+      "loss": 0.002,
+      "step": 6580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.5276416976633295e-06,
+      "loss": 0.0024,
+      "step": 6600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.516369416453556e-06,
+      "loss": 0.0039,
+      "step": 6620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.505065742813599e-06,
+      "loss": 0.0037,
+      "step": 6640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.4937308527279724e-06,
+      "loss": 0.0034,
+      "step": 6660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.482364922667188e-06,
+      "loss": 0.0042,
+      "step": 6680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.470968129585012e-06,
+      "loss": 0.0028,
+      "step": 6700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.459540650915716e-06,
+      "loss": 0.0017,
+      "step": 6720
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.448082664571303e-06,
+      "loss": 0.0031,
+      "step": 6740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.4365943489387454e-06,
+      "loss": 0.003,
+      "step": 6760
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.425075882877206e-06,
+      "loss": 0.0019,
+      "step": 6780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.413527445715257e-06,
+      "loss": 0.0055,
+      "step": 6800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.401949217248077e-06,
+      "loss": 0.0034,
+      "step": 6820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.390341377734666e-06,
+      "loss": 0.0038,
+      "step": 6840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.378704107895028e-06,
+      "loss": 0.0044,
+      "step": 6860
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.367037588907361e-06,
+      "loss": 0.0037,
+      "step": 6880
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.355342002405242e-06,
+      "loss": 0.0022,
+      "step": 6900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.343617530474788e-06,
+      "loss": 0.003,
+      "step": 6920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.33186435565183e-06,
+      "loss": 0.0022,
+      "step": 6940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.320082660919069e-06,
+      "loss": 0.0029,
+      "step": 6960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.3082726297032264e-06,
+      "loss": 0.0033,
+      "step": 6980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.296434445872186e-06,
+      "loss": 0.0039,
+      "step": 7000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.284568293732133e-06,
+      "loss": 0.0045,
+      "step": 7020
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.272674358024693e-06,
+      "loss": 0.0033,
+      "step": 7040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.260752823924037e-06,
+      "loss": 0.003,
+      "step": 7060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.248803877034018e-06,
+      "loss": 0.0043,
+      "step": 7080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.236827703385269e-06,
+      "loss": 0.0037,
+      "step": 7100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.224824489432313e-06,
+      "loss": 0.0023,
+      "step": 7120
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.212794422050654e-06,
+      "loss": 0.0031,
+      "step": 7140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.200737688533877e-06,
+      "loss": 0.0051,
+      "step": 7160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.18865447659072e-06,
+      "loss": 0.0022,
+      "step": 7180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.176544974342165e-06,
+      "loss": 0.0028,
+      "step": 7200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.164409370318495e-06,
+      "loss": 0.0033,
+      "step": 7220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.152247853456372e-06,
+      "loss": 0.0025,
+      "step": 7240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.140060613095887e-06,
+      "loss": 0.0028,
+      "step": 7260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.127847838977612e-06,
+      "loss": 0.0035,
+      "step": 7280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.115609721239655e-06,
+      "loss": 0.0044,
+      "step": 7300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.1033464504146875e-06,
+      "loss": 0.0025,
+      "step": 7320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.091058217426989e-06,
+      "loss": 0.004,
+      "step": 7340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.078745213589466e-06,
+      "loss": 0.0044,
+      "step": 7360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.066407630600678e-06,
+      "loss": 0.0046,
+      "step": 7380
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.054045660541852e-06,
+      "loss": 0.0039,
+      "step": 7400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.041659495873894e-06,
+      "loss": 0.0031,
+      "step": 7420
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.029249329434389e-06,
+      "loss": 0.0031,
+      "step": 7440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.0168153544345995e-06,
+      "loss": 0.0034,
+      "step": 7460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.004357764456462e-06,
+      "loss": 0.0039,
+      "step": 7480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.991876753449567e-06,
+      "loss": 0.0026,
+      "step": 7500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.979372515728142e-06,
+      "loss": 0.0029,
+      "step": 7520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.966845245968026e-06,
+      "loss": 0.0052,
+      "step": 7540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.954295139203642e-06,
+      "loss": 0.0027,
+      "step": 7560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.941722390824955e-06,
+      "loss": 0.0041,
+      "step": 7580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.92912719657443e-06,
+      "loss": 0.0044,
+      "step": 7600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.9165097525439934e-06,
+      "loss": 0.0024,
+      "step": 7620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.903870255171967e-06,
+      "loss": 0.0033,
+      "step": 7640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.891208901240024e-06,
+      "loss": 0.0036,
+      "step": 7660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.878525887870109e-06,
+      "loss": 0.0035,
+      "step": 7680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.865821412521384e-06,
+      "loss": 0.0035,
+      "step": 7700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.853095672987148e-06,
+      "loss": 0.0023,
+      "step": 7720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.840348867391753e-06,
+      "loss": 0.0028,
+      "step": 7740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.8275811941875275e-06,
+      "loss": 0.0045,
+      "step": 7760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.814792852151685e-06,
+      "loss": 0.0021,
+      "step": 7780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.801984040383224e-06,
+      "loss": 0.0027,
+      "step": 7800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.789154958299833e-06,
+      "loss": 0.0039,
+      "step": 7820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.776305805634786e-06,
+      "loss": 0.0036,
+      "step": 7840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.76343678243383e-06,
+      "loss": 0.0044,
+      "step": 7860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.750548089052075e-06,
+      "loss": 0.0024,
+      "step": 7880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.737639926150867e-06,
+      "loss": 0.0015,
+      "step": 7900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.724712494694674e-06,
+      "loss": 0.0028,
+      "step": 7920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.711765995947949e-06,
+      "loss": 0.0037,
+      "step": 7940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.698800631472002e-06,
+      "loss": 0.0036,
+      "step": 7960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.685816603121858e-06,
+      "loss": 0.0033,
+      "step": 7980
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.6728141130431125e-06,
+      "loss": 0.0026,
+      "step": 8000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.659793363668795e-06,
+      "loss": 0.0047,
+      "step": 8020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.646754557716205e-06,
+      "loss": 0.0024,
+      "step": 8040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.633697898183761e-06,
+      "loss": 0.0036,
+      "step": 8060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.620623588347846e-06,
+      "loss": 0.0033,
+      "step": 8080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.607531831759626e-06,
+      "loss": 0.0033,
+      "step": 8100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.594422832241904e-06,
+      "loss": 0.0027,
+      "step": 8120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.581296793885925e-06,
+      "loss": 0.0036,
+      "step": 8140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.568811461351295e-06,
+      "loss": 0.0029,
+      "step": 8160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.55565278527992e-06,
+      "loss": 0.002,
+      "step": 8180
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.542477673972974e-06,
+      "loss": 0.0034,
+      "step": 8200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5292863325509856e-06,
+      "loss": 0.0044,
+      "step": 8220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.516078966387168e-06,
+      "loss": 0.0037,
+      "step": 8240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.502855781104219e-06,
+      "loss": 0.0045,
+      "step": 8260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4896169825711214e-06,
+      "loss": 0.0031,
+      "step": 8280
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4763627768999375e-06,
+      "loss": 0.0024,
+      "step": 8300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.4630933704426e-06,
+      "loss": 0.0037,
+      "step": 8320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.449808969787699e-06,
+      "loss": 0.0021,
+      "step": 8340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.436509781757268e-06,
+      "loss": 0.0037,
+      "step": 8360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.42319601340356e-06,
+      "loss": 0.0035,
+      "step": 8380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.4098678720058245e-06,
+      "loss": 0.0045,
+      "step": 8400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.396525565067087e-06,
+      "loss": 0.0049,
+      "step": 8420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.383169300310908e-06,
+      "loss": 0.0027,
+      "step": 8440
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.369799285678158e-06,
+      "loss": 0.0041,
+      "step": 8460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.356415729323775e-06,
+      "loss": 0.0035,
+      "step": 8480
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.343018839613525e-06,
+      "loss": 0.0028,
+      "step": 8500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.329608825120757e-06,
+      "loss": 0.0033,
+      "step": 8520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.31618589462316e-06,
+      "loss": 0.0025,
+      "step": 8540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.302750257099508e-06,
+      "loss": 0.0023,
+      "step": 8560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.289302121726406e-06,
+      "loss": 0.0019,
+      "step": 8580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.27584169787504e-06,
+      "loss": 0.0035,
+      "step": 8600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.262369195107909e-06,
+      "loss": 0.0031,
+      "step": 8620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.248884823175566e-06,
+      "loss": 0.0019,
+      "step": 8640
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.235388792013355e-06,
+      "loss": 0.0022,
+      "step": 8660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.221881311738139e-06,
+      "loss": 0.0038,
+      "step": 8680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.208362592645029e-06,
+      "loss": 0.0033,
+      "step": 8700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.194832845204112e-06,
+      "loss": 0.0054,
+      "step": 8720
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.181292280057171e-06,
+      "loss": 0.0039,
+      "step": 8740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.16774110801441e-06,
+      "loss": 0.0026,
+      "step": 8760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.15417954005117e-06,
+      "loss": 0.0034,
+      "step": 8780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.140607787304638e-06,
+      "loss": 0.0025,
+      "step": 8800
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.127026061070575e-06,
+      "loss": 0.0038,
+      "step": 8820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.11343457280001e-06,
+      "loss": 0.0034,
+      "step": 8840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099833534095957e-06,
+      "loss": 0.0037,
+      "step": 8860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.08622315671012e-06,
+      "loss": 0.0036,
+      "step": 8880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.072603652539595e-06,
+      "loss": 0.0038,
+      "step": 8900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.058975233623567e-06,
+      "loss": 0.0024,
+      "step": 8920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.045338112140018e-06,
+      "loss": 0.0037,
+      "step": 8940
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.031692500402416e-06,
+      "loss": 0.0024,
+      "step": 8960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.018038610856411e-06,
+      "loss": 0.0031,
+      "step": 8980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.004376656076532e-06,
+      "loss": 0.0023,
+      "step": 9000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.990706848762869e-06,
+      "loss": 0.0048,
+      "step": 9020
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.977029401737772e-06,
+      "loss": 0.0021,
+      "step": 9040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.963344527942526e-06,
+      "loss": 0.0023,
+      "step": 9060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.949652440434046e-06,
+      "loss": 0.0017,
+      "step": 9080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.9359533523815564e-06,
+      "loss": 0.0027,
+      "step": 9100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.922247477063266e-06,
+      "loss": 0.0026,
+      "step": 9120
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.908535027863061e-06,
+      "loss": 0.0036,
+      "step": 9140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.894816218267169e-06,
+      "loss": 0.0029,
+      "step": 9160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.881091261860844e-06,
+      "loss": 0.0035,
+      "step": 9180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.867360372325038e-06,
+      "loss": 0.0033,
+      "step": 9200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8536237634330724e-06,
+      "loss": 0.0031,
+      "step": 9220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.839881649047315e-06,
+      "loss": 0.0027,
+      "step": 9240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8261342431158465e-06,
+      "loss": 0.0053,
+      "step": 9260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8123817596691307e-06,
+      "loss": 0.0037,
+      "step": 9280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7986244128166816e-06,
+      "loss": 0.0038,
+      "step": 9300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.7848624167437305e-06,
+      "loss": 0.003,
+      "step": 9320
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.771095985707891e-06,
+      "loss": 0.0035,
+      "step": 9340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.7573253340358276e-06,
+      "loss": 0.0023,
+      "step": 9360
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.7435506761199096e-06,
+      "loss": 0.0022,
+      "step": 9380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.7297722264148794e-06,
+      "loss": 0.0028,
+      "step": 9400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.715990199434519e-06,
+      "loss": 0.0031,
+      "step": 9420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.702204809748294e-06,
+      "loss": 0.0023,
+      "step": 9440
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.688416271978035e-06,
+      "loss": 0.0028,
+      "step": 9460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6746248007945733e-06,
+      "loss": 0.0031,
+      "step": 9480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6608306109144183e-06,
+      "loss": 0.0037,
+      "step": 9500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6470339170964026e-06,
+      "loss": 0.0028,
+      "step": 9520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6332349341383405e-06,
+      "loss": 0.0031,
+      "step": 9540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.619433876873689e-06,
+      "loss": 0.0033,
+      "step": 9560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.605630960168199e-06,
+      "loss": 0.0018,
+      "step": 9580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.591826398916567e-06,
+      "loss": 0.0023,
+      "step": 9600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.578020408039097e-06,
+      "loss": 0.0031,
+      "step": 9620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.5642132024783495e-06,
+      "loss": 0.0027,
+      "step": 9640
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5504049971957948e-06,
+      "loss": 0.0034,
+      "step": 9660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.536596007168471e-06,
+      "loss": 0.0033,
+      "step": 9680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.522786447385629e-06,
+      "loss": 0.0043,
+      "step": 9700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5089765328453937e-06,
+      "loss": 0.0031,
+      "step": 9720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4951664785514114e-06,
+      "loss": 0.0023,
+      "step": 9740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4813564995095044e-06,
+      "loss": 0.0041,
+      "step": 9760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4675468107243254e-06,
+      "loss": 0.0024,
+      "step": 9780
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4537376271960063e-06,
+      "loss": 0.0033,
+      "step": 9800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.439929163916811e-06,
+      "loss": 0.0038,
+      "step": 9820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.426121635867794e-06,
+      "loss": 0.0016,
+      "step": 9840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4123152580154478e-06,
+      "loss": 0.0033,
+      "step": 9860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3985102453083547e-06,
+      "loss": 0.0041,
+      "step": 9880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3847068126738487e-06,
+      "loss": 0.003,
+      "step": 9900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.370905175014659e-06,
+      "loss": 0.0025,
+      "step": 9920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3571055472055757e-06,
+      "loss": 0.003,
+      "step": 9940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.343308144090088e-06,
+      "loss": 0.0033,
+      "step": 9960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3295131804770587e-06,
+      "loss": 0.0038,
+      "step": 9980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3157208711373653e-06,
+      "loss": 0.0028,
+      "step": 10000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.301931430800565e-06,
+      "loss": 0.0036,
+      "step": 10020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.288145074151544e-06,
+      "loss": 0.0023,
+      "step": 10040
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.274362015827182e-06,
+      "loss": 0.003,
+      "step": 10060
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.260582470413009e-06,
+      "loss": 0.0027,
+      "step": 10080
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2468066524398583e-06,
+      "loss": 0.0025,
+      "step": 10100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2330347763805356e-06,
+      "loss": 0.0048,
+      "step": 10120
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2192670566464756e-06,
+      "loss": 0.0035,
+      "step": 10140
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2055037075844037e-06,
+      "loss": 0.0035,
+      "step": 10160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1917449434729954e-06,
+      "loss": 0.0033,
+      "step": 10180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.177990978519551e-06,
+      "loss": 0.0026,
+      "step": 10200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1642420268566492e-06,
+      "loss": 0.0035,
+      "step": 10220
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.150498302538817e-06,
+      "loss": 0.0034,
+      "step": 10240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.136760019539201e-06,
+      "loss": 0.0034,
+      "step": 10260
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.123027391746231e-06,
+      "loss": 0.0022,
+      "step": 10280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.109300632960295e-06,
+      "loss": 0.0016,
+      "step": 10300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.095579956890403e-06,
+      "loss": 0.0034,
+      "step": 10320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.081865577150867e-06,
+      "loss": 0.0036,
+      "step": 10340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.068157707257976e-06,
+      "loss": 0.0022,
+      "step": 10360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.054456560626661e-06,
+      "loss": 0.003,
+      "step": 10380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.0407623505671857e-06,
+      "loss": 0.003,
+      "step": 10400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.02707529028182e-06,
+      "loss": 0.0036,
+      "step": 10420
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.013395592861518e-06,
+      "loss": 0.0035,
+      "step": 10440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.9997234712826023e-06,
+      "loss": 0.0032,
+      "step": 10460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.986059138403451e-06,
+      "loss": 0.0026,
+      "step": 10480
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.972402806961182e-06,
+      "loss": 0.0029,
+      "step": 10500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.9587546895683386e-06,
+      "loss": 0.0022,
+      "step": 10520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9451149987095806e-06,
+      "loss": 0.0031,
+      "step": 10540
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9314839467383804e-06,
+      "loss": 0.0022,
+      "step": 10560
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.9178617458737108e-06,
+      "loss": 0.0038,
+      "step": 10580
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9042486081967423e-06,
+      "loss": 0.0026,
+      "step": 10600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.890644745647547e-06,
+      "loss": 0.0018,
+      "step": 10620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.8770503700217888e-06,
+      "loss": 0.0021,
+      "step": 10640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8634656929674385e-06,
+      "loss": 0.0021,
+      "step": 10660
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.849890925981465e-06,
+      "loss": 0.0027,
+      "step": 10680
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.836326280406553e-06,
+      "loss": 0.0031,
+      "step": 10700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.8227719674278105e-06,
+      "loss": 0.0041,
+      "step": 10720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.809228198069475e-06,
+      "loss": 0.0018,
+      "step": 10740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7956951831916363e-06,
+      "loss": 0.0026,
+      "step": 10760
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.782173133486949e-06,
+      "loss": 0.0023,
+      "step": 10780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.768662259477354e-06,
+      "loss": 0.0035,
+      "step": 10800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.755162771510797e-06,
+      "loss": 0.0021,
+      "step": 10820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7416748797579606e-06,
+      "loss": 0.002,
+      "step": 10840
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7281987942089876e-06,
+      "loss": 0.0027,
+      "step": 10860
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.714734724670209e-06,
+      "loss": 0.0025,
+      "step": 10880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7012828807608863e-06,
+      "loss": 0.0022,
+      "step": 10900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6878434719099384e-06,
+      "loss": 0.0041,
+      "step": 10920
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6744167073526904e-06,
+      "loss": 0.0031,
+      "step": 10940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6610027961276042e-06,
+      "loss": 0.0022,
+      "step": 10960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6476019470730365e-06,
+      "loss": 0.0024,
+      "step": 10980
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6342143688239786e-06,
+      "loss": 0.0044,
+      "step": 11000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6208402698088155e-06,
+      "loss": 0.0026,
+      "step": 11020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.607479858246072e-06,
+      "loss": 0.0031,
+      "step": 11040
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.594133342141177e-06,
+      "loss": 0.0025,
+      "step": 11060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.580800929283228e-06,
+      "loss": 0.0032,
+      "step": 11080
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.5674828272417462e-06,
+      "loss": 0.0028,
+      "step": 11100
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.554179243363453e-06,
+      "loss": 0.0029,
+      "step": 11120
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.540890384769043e-06,
+      "loss": 0.0022,
+      "step": 11140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.5276164583499516e-06,
+      "loss": 0.0021,
+      "step": 11160
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.514357670765138e-06,
+      "loss": 0.0026,
+      "step": 11180
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.501114228437872e-06,
+      "loss": 0.0031,
+      "step": 11200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.487886337552514e-06,
+      "loss": 0.002,
+      "step": 11220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.474674204051304e-06,
+      "loss": 0.0018,
+      "step": 11240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4614780336311665e-06,
+      "loss": 0.0025,
+      "step": 11260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4482980317404924e-06,
+      "loss": 0.0031,
+      "step": 11280
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.4351344035759546e-06,
+      "loss": 0.0018,
+      "step": 11300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.421987354079301e-06,
+      "loss": 0.0041,
+      "step": 11320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.408857087934172e-06,
+      "loss": 0.0041,
+      "step": 11340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3957438095629154e-06,
+      "loss": 0.0029,
+      "step": 11360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3826477231233922e-06,
+      "loss": 0.0025,
+      "step": 11380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.369569032505813e-06,
+      "loss": 0.003,
+      "step": 11400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3565079413295523e-06,
+      "loss": 0.0023,
+      "step": 11420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.343464652939986e-06,
+      "loss": 0.0033,
+      "step": 11440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3304393704053184e-06,
+      "loss": 0.0033,
+      "step": 11460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.317432296513425e-06,
+      "loss": 0.0029,
+      "step": 11480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3044436337686986e-06,
+      "loss": 0.0031,
+      "step": 11500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.29147358438889e-06,
+      "loss": 0.0022,
+      "step": 11520
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2785223503019595e-06,
+      "loss": 0.0039,
+      "step": 11540
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2655901331429415e-06,
+      "loss": 0.0032,
+      "step": 11560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2526771342507983e-06,
+      "loss": 0.0037,
+      "step": 11580
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.240427769335911e-06,
+      "loss": 0.0025,
+      "step": 11600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.227552824029434e-06,
+      "loss": 0.0025,
+      "step": 11620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2146976891846745e-06,
+      "loss": 0.0026,
+      "step": 11640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2018625649405147e-06,
+      "loss": 0.0028,
+      "step": 11660
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1890476511242986e-06,
+      "loss": 0.0034,
+      "step": 11680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.176253147248715e-06,
+      "loss": 0.0018,
+      "step": 11700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.163479252508695e-06,
+      "loss": 0.003,
+      "step": 11720
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.150726165778316e-06,
+      "loss": 0.0025,
+      "step": 11740
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1379940856076917e-06,
+      "loss": 0.002,
+      "step": 11760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.125283210219894e-06,
+      "loss": 0.0042,
+      "step": 11780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.1125937375078623e-06,
+      "loss": 0.0023,
+      "step": 11800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0999258650313204e-06,
+      "loss": 0.0024,
+      "step": 11820
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.087279790013702e-06,
+      "loss": 0.0022,
+      "step": 11840
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0746557093390825e-06,
+      "loss": 0.0014,
+      "step": 11860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.0620538195491133e-06,
+      "loss": 0.0012,
+      "step": 11880
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0494743168399564e-06,
+      "loss": 0.0024,
+      "step": 11900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.036917397059236e-06,
+      "loss": 0.0018,
+      "step": 11920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.02438325570299e-06,
+      "loss": 0.0021,
+      "step": 11940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.0118720879126206e-06,
+      "loss": 0.0014,
+      "step": 11960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.99938408847186e-06,
+      "loss": 0.0017,
+      "step": 11980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9869194518037385e-06,
+      "loss": 0.0012,
+      "step": 12000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9744783719675555e-06,
+      "loss": 0.0009,
+      "step": 12020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.962061042655859e-06,
+      "loss": 0.0017,
+      "step": 12040
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9496676571914295e-06,
+      "loss": 0.0016,
+      "step": 12060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9372984085242725e-06,
+      "loss": 0.0014,
+      "step": 12080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9249534892286107e-06,
+      "loss": 0.0024,
+      "step": 12100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.91263309149989e-06,
+      "loss": 0.0016,
+      "step": 12120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9003374071517838e-06,
+      "loss": 0.0008,
+      "step": 12140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8880666276132063e-06,
+      "loss": 0.0008,
+      "step": 12160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8758209439253417e-06,
+      "loss": 0.0023,
+      "step": 12180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8636005467386516e-06,
+      "loss": 0.0009,
+      "step": 12200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8514056263099228e-06,
+      "loss": 0.001,
+      "step": 12220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8392363724993038e-06,
+      "loss": 0.0012,
+      "step": 12240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8270929747673356e-06,
+      "loss": 0.0011,
+      "step": 12260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8149756221720191e-06,
+      "loss": 0.0013,
+      "step": 12280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8028845033658587e-06,
+      "loss": 0.0009,
+      "step": 12300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7908198065929378e-06,
+      "loss": 0.0019,
+      "step": 12320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.7787817196859714e-06,
+      "loss": 0.0012,
+      "step": 12340
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7667704300633958e-06,
+      "loss": 0.002,
+      "step": 12360
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.7547861247264506e-06,
+      "loss": 0.001,
+      "step": 12380
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.742828990256253e-06,
+      "loss": 0.0014,
+      "step": 12400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7308992128109082e-06,
+      "loss": 0.0007,
+      "step": 12420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7189969781226078e-06,
+      "loss": 0.0014,
+      "step": 12440
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.7071224714947338e-06,
+      "loss": 0.001,
+      "step": 12460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.695275877798971e-06,
+      "loss": 0.0012,
+      "step": 12480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.683457381472442e-06,
+      "loss": 0.0013,
+      "step": 12500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6716671665148244e-06,
+      "loss": 0.0012,
+      "step": 12520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6599054164854832e-06,
+      "loss": 0.0013,
+      "step": 12540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.648172314500629e-06,
+      "loss": 0.0023,
+      "step": 12560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6364680432304499e-06,
+      "loss": 0.0006,
+      "step": 12580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6247927848962769e-06,
+      "loss": 0.0024,
+      "step": 12600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6131467212677448e-06,
+      "loss": 0.0016,
+      "step": 12620
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.6015300336599622e-06,
+      "loss": 0.0016,
+      "step": 12640
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5899429029306885e-06,
+      "loss": 0.0015,
+      "step": 12660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5783855094775178e-06,
+      "loss": 0.0011,
+      "step": 12680
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.5668580332350714e-06,
+      "loss": 0.0012,
+      "step": 12700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5553606536721957e-06,
+      "loss": 0.0013,
+      "step": 12720
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5438935497891685e-06,
+      "loss": 0.0008,
+      "step": 12740
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.5324569001149115e-06,
+      "loss": 0.0009,
+      "step": 12760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.5210508827042116e-06,
+      "loss": 0.0011,
+      "step": 12780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.509675675134948e-06,
+      "loss": 0.0013,
+      "step": 12800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.4983314545053285e-06,
+      "loss": 0.0019,
+      "step": 12820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4870183974311313e-06,
+      "loss": 0.0019,
+      "step": 12840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.475736680042956e-06,
+      "loss": 0.0009,
+      "step": 12860
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.4644864779834814e-06,
+      "loss": 0.0015,
+      "step": 12880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.453267966404731e-06,
+      "loss": 0.0015,
+      "step": 12900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4420813199653446e-06,
+      "loss": 0.001,
+      "step": 12920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.4309267128278613e-06,
+      "loss": 0.0012,
+      "step": 12940
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4198043186560094e-06,
+      "loss": 0.0014,
+      "step": 12960
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.4087143106119948e-06,
+      "loss": 0.0015,
+      "step": 12980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3976568613538146e-06,
+      "loss": 0.0022,
+      "step": 13000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.386632143032567e-06,
+      "loss": 0.0016,
+      "step": 13020
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3756403272897624e-06,
+      "loss": 0.0016,
+      "step": 13040
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.364681585254663e-06,
+      "loss": 0.0019,
+      "step": 13060
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3537560875416142e-06,
+      "loss": 0.002,
+      "step": 13080
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3428640042473855e-06,
+      "loss": 0.0019,
+      "step": 13100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3320055049485208e-06,
+      "loss": 0.0013,
+      "step": 13120
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3211807586987092e-06,
+      "loss": 0.0009,
+      "step": 13140
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.3103899340261412e-06,
+      "loss": 0.0008,
+      "step": 13160
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2996331989308904e-06,
+      "loss": 0.0011,
+      "step": 13180
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2889107208822962e-06,
+      "loss": 0.0012,
+      "step": 13200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2782226668163582e-06,
+      "loss": 0.0019,
+      "step": 13220
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2675692031331374e-06,
+      "loss": 0.0017,
+      "step": 13240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2569504956941595e-06,
+      "loss": 0.0012,
+      "step": 13260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2463667098198443e-06,
+      "loss": 0.0016,
+      "step": 13280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2358180102869205e-06,
+      "loss": 0.0012,
+      "step": 13300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.225304561325867e-06,
+      "loss": 0.0016,
+      "step": 13320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.214826526618353e-06,
+      "loss": 0.0012,
+      "step": 13340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.204384069294691e-06,
+      "loss": 0.0011,
+      "step": 13360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1939773519312967e-06,
+      "loss": 0.0009,
+      "step": 13380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1836065365481582e-06,
+      "loss": 0.002,
+      "step": 13400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.173271784606312e-06,
+      "loss": 0.0009,
+      "step": 13420
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1629732570053327e-06,
+      "loss": 0.0013,
+      "step": 13440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1527111140808235e-06,
+      "loss": 0.0012,
+      "step": 13460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.142485515601924e-06,
+      "loss": 0.002,
+      "step": 13480
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1322966207688196e-06,
+      "loss": 0.0013,
+      "step": 13500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1221445882102655e-06,
+      "loss": 0.0011,
+      "step": 13520
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1120295759811154e-06,
+      "loss": 0.001,
+      "step": 13540
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1019517415598616e-06,
+      "loss": 0.0013,
+      "step": 13560
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.091911241846181e-06,
+      "loss": 0.0011,
+      "step": 13580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0819082331585003e-06,
+      "loss": 0.0012,
+      "step": 13600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0719428712315477e-06,
+      "loss": 0.0018,
+      "step": 13620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0620153112139418e-06,
+      "loss": 0.0021,
+      "step": 13640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.052125707665773e-06,
+      "loss": 0.0025,
+      "step": 13660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0422742145561938e-06,
+      "loss": 0.0012,
+      "step": 13680
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0324609852610178e-06,
+      "loss": 0.0021,
+      "step": 13700
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.022686172560347e-06,
+      "loss": 0.0012,
+      "step": 13720
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0129499286361782e-06,
+      "loss": 0.0007,
+      "step": 13740
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0032524050700372e-06,
+      "loss": 0.0012,
+      "step": 13760
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.935937528406236e-07,
+      "loss": 0.0011,
+      "step": 13780
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.83974122321459e-07,
+      "loss": 0.0008,
+      "step": 13800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.74393663278542e-07,
+      "loss": 0.0013,
+      "step": 13820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.648525248680157e-07,
+      "loss": 0.0009,
+      "step": 13840
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.553508556338546e-07,
+      "loss": 0.0011,
+      "step": 13860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.458888035055434e-07,
+      "loss": 0.001,
+      "step": 13880
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.364665157957729e-07,
+      "loss": 0.0014,
+      "step": 13900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.270841391981565e-07,
+      "loss": 0.0023,
+      "step": 13920
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.177418197849354e-07,
+      "loss": 0.0013,
+      "step": 13940
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.084397030047103e-07,
+      "loss": 0.002,
+      "step": 13960
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.991779336801755e-07,
+      "loss": 0.001,
+      "step": 13980
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 8.899566560058645e-07,
+      "loss": 0.0011,
+      "step": 14000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.807760135459038e-07,
+      "loss": 0.0008,
+      "step": 14020
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.716361492317806e-07,
+      "loss": 0.0011,
+      "step": 14040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.62537205360114e-07,
+      "loss": 0.0014,
+      "step": 14060
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.534793235904423e-07,
+      "loss": 0.0012,
+      "step": 14080
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.444626449430168e-07,
+      "loss": 0.0012,
+      "step": 14100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.354873097966053e-07,
+      "loss": 0.0014,
+      "step": 14120
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.26553457886308e-07,
+      "loss": 0.0012,
+      "step": 14140
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.176612283013801e-07,
+      "loss": 0.0015,
+      "step": 14160
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 8.088107594830711e-07,
+      "loss": 0.001,
+      "step": 14180
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.00002189222461e-07,
+      "loss": 0.0013,
+      "step": 14200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.912356546583222e-07,
+      "loss": 0.0014,
+      "step": 14220
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.825112922749856e-07,
+      "loss": 0.0013,
+      "step": 14240
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.738292379002057e-07,
+      "loss": 0.0015,
+      "step": 14260
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.651896267030573e-07,
+      "loss": 0.0009,
+      "step": 14280
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.565925931918236e-07,
+      "loss": 0.001,
+      "step": 14300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.480382712119103e-07,
+      "loss": 0.0015,
+      "step": 14320
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.395267939437475e-07,
+      "loss": 0.0015,
+      "step": 14340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.310582939007311e-07,
+      "loss": 0.0014,
+      "step": 14360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.226329029271541e-07,
+      "loss": 0.0013,
+      "step": 14380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.142507521961495e-07,
+      "loss": 0.0008,
+      "step": 14400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.059119722076539e-07,
+      "loss": 0.0018,
+      "step": 14420
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.976166927863763e-07,
+      "loss": 0.002,
+      "step": 14440
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.893650430797719e-07,
+      "loss": 0.0009,
+      "step": 14460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.811571515560327e-07,
+      "loss": 0.0013,
+      "step": 14480
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.72993146002093e-07,
+      "loss": 0.0005,
+      "step": 14500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.648731535216334e-07,
+      "loss": 0.0012,
+      "step": 14520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.567973005331017e-07,
+      "loss": 0.001,
+      "step": 14540
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.487657127677525e-07,
+      "loss": 0.001,
+      "step": 14560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.407785152676812e-07,
+      "loss": 0.0013,
+      "step": 14580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.328358323838806e-07,
+      "loss": 0.0009,
+      "step": 14600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.24937787774306e-07,
+      "loss": 0.0009,
+      "step": 14620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.170845044019476e-07,
+      "loss": 0.0012,
+      "step": 14640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.092761045329183e-07,
+      "loss": 0.0011,
+      "step": 14660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.015127097345485e-07,
+      "loss": 0.0017,
+      "step": 14680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.937944408734949e-07,
+      "loss": 0.0007,
+      "step": 14700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.861214181138573e-07,
+      "loss": 0.0012,
+      "step": 14720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.784937609153085e-07,
+      "loss": 0.0011,
+      "step": 14740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.709115880312354e-07,
+      "loss": 0.0011,
+      "step": 14760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.63375017506888e-07,
+      "loss": 0.0008,
+      "step": 14780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.558841666775436e-07,
+      "loss": 0.0019,
+      "step": 14800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.484391521666787e-07,
+      "loss": 0.0012,
+      "step": 14820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.410400898841539e-07,
+      "loss": 0.0016,
+      "step": 14840
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.336870950244099e-07,
+      "loss": 0.0008,
+      "step": 14860
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.26380282064672e-07,
+      "loss": 0.0012,
+      "step": 14880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.191197647631709e-07,
+      "loss": 0.0018,
+      "step": 14900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.119056561573692e-07,
+      "loss": 0.001,
+      "step": 14920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.047380685622007e-07,
+      "loss": 0.0012,
+      "step": 14940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.976171135683283e-07,
+      "loss": 0.0012,
+      "step": 14960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.905429020403957e-07,
+      "loss": 0.0016,
+      "step": 14980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.8351554411531e-07,
+      "loss": 0.0011,
+      "step": 15000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.7653514920052715e-07,
+      "loss": 0.001,
+      "step": 15020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6960182597234096e-07,
+      "loss": 0.0013,
+      "step": 15040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.6271568237419857e-07,
+      "loss": 0.0013,
+      "step": 15060
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.5587682561501914e-07,
+      "loss": 0.0014,
+      "step": 15080
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.490853621675206e-07,
+      "loss": 0.0009,
+      "step": 15100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.423413977665633e-07,
+      "loss": 0.0018,
+      "step": 15120
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.356450374075089e-07,
+      "loss": 0.0012,
+      "step": 15140
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.289963853445783e-07,
+      "loss": 0.0013,
+      "step": 15160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.2239554508923403e-07,
+      "loss": 0.0008,
+      "step": 15180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.1584261940856414e-07,
+      "loss": 0.0012,
+      "step": 15200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.0933771032368855e-07,
+      "loss": 0.0016,
+      "step": 15220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.0288091910816475e-07,
+      "loss": 0.0017,
+      "step": 15240
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.964723462864111e-07,
+      "loss": 0.0017,
+      "step": 15260
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.9011209163214874e-07,
+      "loss": 0.0011,
+      "step": 15280
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.8380025416683987e-07,
+      "loss": 0.0013,
+      "step": 15300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.775369321581513e-07,
+      "loss": 0.0006,
+      "step": 15320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.713222231184218e-07,
+      "loss": 0.0012,
+      "step": 15340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.651562238031462e-07,
+      "loss": 0.0014,
+      "step": 15360
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.590390302094672e-07,
+      "loss": 0.0008,
+      "step": 15380
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.529707375746818e-07,
+      "loss": 0.001,
+      "step": 15400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.469514403747584e-07,
+      "loss": 0.0013,
+      "step": 15420
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.40981232322865e-07,
+      "loss": 0.0014,
+      "step": 15440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.350602063679118e-07,
+      "loss": 0.0009,
+      "step": 15460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.2918845469310335e-07,
+      "loss": 0.0009,
+      "step": 15480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.2336606871450347e-07,
+      "loss": 0.0004,
+      "step": 15500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.1759313907961054e-07,
+      "loss": 0.001,
+      "step": 15520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.118697556659491e-07,
+      "loss": 0.0008,
+      "step": 15540
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.0619600757966826e-07,
+      "loss": 0.0014,
+      "step": 15560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.0057198315415475e-07,
+      "loss": 0.0013,
+      "step": 15580
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.9499776994866074e-07,
+      "loss": 0.0013,
+      "step": 15600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.8947345474693413e-07,
+      "loss": 0.0013,
+      "step": 15620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.83999123555873e-07,
+      "loss": 0.0013,
+      "step": 15640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.7857486160418626e-07,
+      "loss": 0.0015,
+      "step": 15660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7320075334106434e-07,
+      "loss": 0.0008,
+      "step": 15680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.678768824348644e-07,
+      "loss": 0.0012,
+      "step": 15700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.626033317718093e-07,
+      "loss": 0.0013,
+      "step": 15720
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.573801834546994e-07,
+      "loss": 0.001,
+      "step": 15740
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.52207518801627e-07,
+      "loss": 0.0012,
+      "step": 15760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.470854183447172e-07,
+      "loss": 0.001,
+      "step": 15780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.4201396182887346e-07,
+      "loss": 0.0012,
+      "step": 15800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.369932282105317e-07,
+      "loss": 0.0007,
+      "step": 15820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.320232956564338e-07,
+      "loss": 0.0008,
+      "step": 15840
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2710424154241305e-07,
+      "loss": 0.0011,
+      "step": 15860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.2223614245218549e-07,
+      "loss": 0.0012,
+      "step": 15880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1741907417615712e-07,
+      "loss": 0.001,
+      "step": 15900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.1265311171025004e-07,
+      "loss": 0.0029,
+      "step": 15920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0793832925472832e-07,
+      "loss": 0.0008,
+      "step": 15940
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.03274800213045e-07,
+      "loss": 0.0012,
+      "step": 15960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.986625971907006e-07,
+      "loss": 0.0009,
+      "step": 15980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9410179199411136e-07,
+      "loss": 0.0014,
+      "step": 16000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8959245562949117e-07,
+      "loss": 0.0014,
+      "step": 16020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.851346583017469e-07,
+      "loss": 0.0018,
+      "step": 16040
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8072846941338539e-07,
+      "loss": 0.0017,
+      "step": 16060
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7637395756343148e-07,
+      "loss": 0.0016,
+      "step": 16080
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7207119054636199e-07,
+      "loss": 0.0016,
+      "step": 16100
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6782023535104972e-07,
+      "loss": 0.001,
+      "step": 16120
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6362115815971873e-07,
+      "loss": 0.0011,
+      "step": 16140
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5947402434691605e-07,
+      "loss": 0.0013,
+      "step": 16160
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5537889847849522e-07,
+      "loss": 0.0012,
+      "step": 16180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5133584431060488e-07,
+      "loss": 0.0014,
+      "step": 16200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4734492478870392e-07,
+      "loss": 0.0015,
+      "step": 16220
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.4340620204657667e-07,
+      "loss": 0.0008,
+      "step": 16240
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3951973740536593e-07,
+      "loss": 0.0009,
+      "step": 16260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3568559137262108e-07,
+      "loss": 0.0012,
+      "step": 16280
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.3190382364135223e-07,
+      "loss": 0.001,
+      "step": 16300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2817449308910457e-07,
+      "loss": 0.0007,
+      "step": 16320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.244976577770379e-07,
+      "loss": 0.0012,
+      "step": 16340
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.2087337494902562e-07,
+      "loss": 0.001,
+      "step": 16360
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.173017010307642e-07,
+      "loss": 0.0013,
+      "step": 16380
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.1378269162889026e-07,
+      "loss": 0.0011,
+      "step": 16400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.103164015301193e-07,
+      "loss": 0.001,
+      "step": 16420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0690288470039084e-07,
+      "loss": 0.0012,
+      "step": 16440
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.0354219428402877e-07,
+      "loss": 0.0008,
+      "step": 16460
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.002343826029129e-07,
+      "loss": 0.0013,
+      "step": 16480
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.714098734934384e-08,
+      "loss": 0.0012,
+      "step": 16500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.393643657431666e-08,
+      "loss": 0.0021,
+      "step": 16520
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.078491408455208e-08,
+      "loss": 0.0023,
+      "step": 16540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.768646894543824e-08,
+      "loss": 0.001,
+      "step": 16560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.464114939601475e-08,
+      "loss": 0.0009,
+      "step": 16580
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.164900284821919e-08,
+      "loss": 0.0009,
+      "step": 16600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.871007588615043e-08,
+      "loss": 0.001,
+      "step": 16620
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.582441426534154e-08,
+      "loss": 0.0008,
+      "step": 16640
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.299206291205068e-08,
+      "loss": 0.0011,
+      "step": 16660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.021306592255894e-08,
+      "loss": 0.0013,
+      "step": 16680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.748746656248289e-08,
+      "loss": 0.001,
+      "step": 16700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.481530726610552e-08,
+      "loss": 0.0008,
+      "step": 16720
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.219662963571093e-08,
+      "loss": 0.0014,
+      "step": 16740
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.963147444093897e-08,
+      "loss": 0.001,
+      "step": 16760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.711988161814946e-08,
+      "loss": 0.0011,
+      "step": 16780
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.4661890269801283e-08,
+      "loss": 0.0018,
+      "step": 16800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.225753866384425e-08,
+      "loss": 0.0012,
+      "step": 16820
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.9906864233120296e-08,
+      "loss": 0.001,
+      "step": 16840
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.760990357478451e-08,
+      "loss": 0.0013,
+      "step": 16860
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.536669244973274e-08,
+      "loss": 0.0013,
+      "step": 16880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.3177265782046335e-08,
+      "loss": 0.0013,
+      "step": 16900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.1041657658446585e-08,
+      "loss": 0.0009,
+      "step": 16920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.895990132776622e-08,
+      "loss": 0.0016,
+      "step": 16940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.693202920043032e-08,
+      "loss": 0.0013,
+      "step": 16960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.505548968809458e-08,
+      "loss": 0.0015,
+      "step": 16980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.313278180031337e-08,
+      "loss": 0.0009,
+      "step": 17000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.126404883707201e-08,
+      "loss": 0.0014,
+      "step": 17020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.9449319892279502e-08,
+      "loss": 0.0011,
+      "step": 17040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.7688623219068008e-08,
+      "loss": 0.0009,
+      "step": 17060
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.598198622935338e-08,
+      "loss": 0.001,
+      "step": 17080
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.432943549340616e-08,
+      "loss": 0.001,
+      "step": 17100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.2730996739440846e-08,
+      "loss": 0.0017,
+      "step": 17120
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.118669485321295e-08,
+      "loss": 0.001,
+      "step": 17140
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9696553877633138e-08,
+      "loss": 0.0006,
+      "step": 17160
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8260597012391464e-08,
+      "loss": 0.0008,
+      "step": 17180
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6878846613597952e-08,
+      "loss": 0.0011,
+      "step": 17200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5551324193432856e-08,
+      "loss": 0.001,
+      "step": 17220
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4278050419812893e-08,
+      "loss": 0.0019,
+      "step": 17240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.305904511606909e-08,
+      "loss": 0.0008,
+      "step": 17260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1894327260637882e-08,
+      "loss": 0.0013,
+      "step": 17280
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.0783914986765397e-08,
+      "loss": 0.001,
+      "step": 17300
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.727825582226901e-09,
+      "loss": 0.001,
+      "step": 17320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.726075489055185e-09,
+      "loss": 0.0009,
+      "step": 17340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.778680303286045e-09,
+      "loss": 0.0009,
+      "step": 17360
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.885654774715422e-09,
+      "loss": 0.0006,
+      "step": 17380
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.047012806669361e-09,
+      "loss": 0.0014,
+      "step": 17400
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.262767455788353e-09,
+      "loss": 0.0012,
+      "step": 17420
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.532930931822165e-09,
+      "loss": 0.0015,
+      "step": 17440
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.857514597441769e-09,
+      "loss": 0.0013,
+      "step": 17460
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.2365289680621466e-09,
+      "loss": 0.0005,
+      "step": 17480
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.6699837116771485e-09,
+      "loss": 0.0012,
+      "step": 17500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1578876487106657e-09,
+      "loss": 0.001,
+      "step": 17520
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7002487518779087e-09,
+      "loss": 0.0009,
+      "step": 17540
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2970741460630041e-09,
+      "loss": 0.0009,
+      "step": 17560
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.483701082055317e-10,
+      "loss": 0.0015,
+      "step": 17580
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.541420672057096e-10,
+      "loss": 0.0016,
+      "step": 17600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.143946038373536e-10,
+      "loss": 0.0011,
+      "step": 17620
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.291314506787101e-10,
+      "loss": 0.0012,
+      "step": 17640
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.835549205261528e-11,
+      "loss": 0.0019,
+      "step": 17660
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.2068763981808458e-11,
+      "loss": 0.0009,
+      "step": 17680
+    },
+    {
+      "epoch": 3.0,
+      "step": 17694,
+      "total_flos": 1.176514057372631e+18,
+      "train_loss": 0.0032725768327810704,
+      "train_runtime": 79905.6476,
+      "train_samples_per_second": 3.543,
+      "train_steps_per_second": 0.221
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 17694,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 3000,
+  "total_flos": 1.176514057372631e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}