diff --git "a/checkpoint-80000/trainer_state.json" "b/checkpoint-80000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-80000/trainer_state.json"
@@ -0,0 +1,9652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.901010548056354,
+  "global_step": 80000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 178.9465,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 164.9707,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 142.2782,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 121.5122,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 91.8622,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 82.2062,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-05,
+      "loss": 72.6893,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8e-05,
+      "loss": 71.8709,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9e-05,
+      "loss": 69.9995,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001,
+      "loss": 70.6458,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994977448744865e-05,
+      "loss": 73.9929,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.989954897489729e-05,
+      "loss": 66.52,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984932346234594e-05,
+      "loss": 65.8947,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979909794979458e-05,
+      "loss": 62.5809,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974887243724323e-05,
+      "loss": 61.212,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969864692469187e-05,
+      "loss": 68.2408,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.964842141214051e-05,
+      "loss": 61.5308,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959819589958916e-05,
+      "loss": 58.9116,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95479703870378e-05,
+      "loss": 60.0702,
+      "step": 950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949774487448646e-05,
+      "loss": 57.6135,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944751936193509e-05,
+      "loss": 50.9231,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939729384938373e-05,
+      "loss": 51.187,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934706833683238e-05,
+      "loss": 52.1127,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.929684282428102e-05,
+      "loss": 47.4608,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.924661731172968e-05,
+      "loss": 51.6108,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.919639179917831e-05,
+      "loss": 46.5874,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.914616628662697e-05,
+      "loss": 41.4706,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.90959407740756e-05,
+      "loss": 43.7544,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.904571526152426e-05,
+      "loss": 44.6039,
+      "step": 1450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.899548974897289e-05,
+      "loss": 41.4384,
+      "step": 1500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.894526423642154e-05,
+      "loss": 42.8289,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889503872387019e-05,
+      "loss": 39.9726,
+      "step": 1600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884481321131882e-05,
+      "loss": 43.9533,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879458769876748e-05,
+      "loss": 38.7605,
+      "step": 1700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87443621862161e-05,
+      "loss": 39.5425,
+      "step": 1750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869413667366476e-05,
+      "loss": 37.588,
+      "step": 1800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.86439111611134e-05,
+      "loss": 39.7744,
+      "step": 1850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.859368564856205e-05,
+      "loss": 38.2154,
+      "step": 1900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.85434601360107e-05,
+      "loss": 35.0806,
+      "step": 1950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.849323462345934e-05,
+      "loss": 39.061,
+      "step": 2000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.844300911090798e-05,
+      "loss": 35.1544,
+      "step": 2050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.839278359835663e-05,
+      "loss": 38.123,
+      "step": 2100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.834255808580527e-05,
+      "loss": 33.1144,
+      "step": 2150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.829233257325392e-05,
+      "loss": 34.3476,
+      "step": 2200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.824210706070256e-05,
+      "loss": 29.5665,
+      "step": 2250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.81918815481512e-05,
+      "loss": 35.8756,
+      "step": 2300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.814165603559985e-05,
+      "loss": 37.2579,
+      "step": 2350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.809143052304849e-05,
+      "loss": 33.6245,
+      "step": 2400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.804120501049714e-05,
+      "loss": 35.6543,
+      "step": 2450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.799097949794578e-05,
+      "loss": 36.7847,
+      "step": 2500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.794075398539442e-05,
+      "loss": 33.463,
+      "step": 2550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.789052847284307e-05,
+      "loss": 32.2215,
+      "step": 2600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.784030296029171e-05,
+      "loss": 33.4301,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.779007744774036e-05,
+      "loss": 29.9579,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.773985193518901e-05,
+      "loss": 31.9141,
+      "step": 2750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.768962642263764e-05,
+      "loss": 33.2049,
+      "step": 2800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.763940091008629e-05,
+      "loss": 32.8774,
+      "step": 2850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.758917539753493e-05,
+      "loss": 29.0858,
+      "step": 2900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.753894988498358e-05,
+      "loss": 30.1145,
+      "step": 2950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.748872437243222e-05,
+      "loss": 27.6986,
+      "step": 3000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.743849885988087e-05,
+      "loss": 31.7807,
+      "step": 3050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.738827334732952e-05,
+      "loss": 30.5108,
+      "step": 3100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.733804783477815e-05,
+      "loss": 31.0909,
+      "step": 3150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.728782232222681e-05,
+      "loss": 27.9057,
+      "step": 3200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.723759680967544e-05,
+      "loss": 29.7323,
+      "step": 3250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.71873712971241e-05,
+      "loss": 29.7527,
+      "step": 3300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.713714578457273e-05,
+      "loss": 29.1442,
+      "step": 3350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.708692027202137e-05,
+      "loss": 30.8906,
+      "step": 3400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.703669475947003e-05,
+      "loss": 26.8419,
+      "step": 3450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.698646924691866e-05,
+      "loss": 29.2181,
+      "step": 3500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.693624373436732e-05,
+      "loss": 27.6549,
+      "step": 3550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.688601822181595e-05,
+      "loss": 34.0701,
+      "step": 3600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.683579270926461e-05,
+      "loss": 24.7487,
+      "step": 3650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.678556719671325e-05,
+      "loss": 30.0266,
+      "step": 3700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.67353416841619e-05,
+      "loss": 25.5011,
+      "step": 3750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.668511617161054e-05,
+      "loss": 26.1437,
+      "step": 3800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.663489065905918e-05,
+      "loss": 23.2303,
+      "step": 3850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.658466514650783e-05,
+      "loss": 26.357,
+      "step": 3900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.653443963395646e-05,
+      "loss": 27.2201,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.648421412140512e-05,
+      "loss": 25.5695,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.643398860885376e-05,
+      "loss": 24.8346,
+      "step": 4050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.63837630963024e-05,
+      "loss": 22.3957,
+      "step": 4100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.633353758375105e-05,
+      "loss": 24.9532,
+      "step": 4150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.628331207119969e-05,
+      "loss": 23.1574,
+      "step": 4200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.623308655864834e-05,
+      "loss": 23.7018,
+      "step": 4250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.618286104609698e-05,
+      "loss": 25.1433,
+      "step": 4300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.613263553354562e-05,
+      "loss": 25.0571,
+      "step": 4350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.608241002099427e-05,
+      "loss": 24.2231,
+      "step": 4400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.603218450844291e-05,
+      "loss": 23.0983,
+      "step": 4450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.598195899589156e-05,
+      "loss": 25.0078,
+      "step": 4500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.59317334833402e-05,
+      "loss": 20.6933,
+      "step": 4550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.588150797078884e-05,
+      "loss": 23.6196,
+      "step": 4600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.583128245823749e-05,
+      "loss": 25.2331,
+      "step": 4650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.578105694568613e-05,
+      "loss": 24.7932,
+      "step": 4700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.573083143313478e-05,
+      "loss": 24.3586,
+      "step": 4750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.568060592058342e-05,
+      "loss": 22.7161,
+      "step": 4800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.563038040803208e-05,
+      "loss": 22.4188,
+      "step": 4850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.558015489548071e-05,
+      "loss": 21.6516,
+      "step": 4900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.552992938292937e-05,
+      "loss": 21.78,
+      "step": 4950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.5479703870378e-05,
+      "loss": 21.0172,
+      "step": 5000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.542947835782665e-05,
+      "loss": 22.4624,
+      "step": 5050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.537925284527528e-05,
+      "loss": 23.6615,
+      "step": 5100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.532902733272393e-05,
+      "loss": 21.8091,
+      "step": 5150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.527880182017259e-05,
+      "loss": 21.4173,
+      "step": 5200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.522857630762122e-05,
+      "loss": 20.5415,
+      "step": 5250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.517835079506987e-05,
+      "loss": 21.0639,
+      "step": 5300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.51281252825185e-05,
+      "loss": 21.6078,
+      "step": 5350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.507789976996716e-05,
+      "loss": 19.4142,
+      "step": 5400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.50276742574158e-05,
+      "loss": 20.2504,
+      "step": 5450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.497744874486445e-05,
+      "loss": 23.8683,
+      "step": 5500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.49272232323131e-05,
+      "loss": 19.7559,
+      "step": 5550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.487699771976174e-05,
+      "loss": 21.1743,
+      "step": 5600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.482677220721038e-05,
+      "loss": 21.1908,
+      "step": 5650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.477654669465901e-05,
+      "loss": 20.9591,
+      "step": 5700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.472632118210767e-05,
+      "loss": 20.9036,
+      "step": 5750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.46760956695563e-05,
+      "loss": 22.249,
+      "step": 5800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.462587015700496e-05,
+      "loss": 19.1093,
+      "step": 5850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.45756446444536e-05,
+      "loss": 21.2714,
+      "step": 5900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.452541913190225e-05,
+      "loss": 21.3794,
+      "step": 5950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.447519361935089e-05,
+      "loss": 20.0326,
+      "step": 6000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.442496810679954e-05,
+      "loss": 19.8004,
+      "step": 6050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.437474259424818e-05,
+      "loss": 19.0229,
+      "step": 6100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.432451708169682e-05,
+      "loss": 17.6587,
+      "step": 6150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.427429156914547e-05,
+      "loss": 21.9247,
+      "step": 6200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.422406605659411e-05,
+      "loss": 19.743,
+      "step": 6250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.417384054404276e-05,
+      "loss": 22.9746,
+      "step": 6300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.41236150314914e-05,
+      "loss": 19.6693,
+      "step": 6350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.407338951894004e-05,
+      "loss": 19.1141,
+      "step": 6400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.402316400638869e-05,
+      "loss": 18.3847,
+      "step": 6450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.397293849383733e-05,
+      "loss": 18.9357,
+      "step": 6500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.392271298128598e-05,
+      "loss": 18.9316,
+      "step": 6550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.387248746873462e-05,
+      "loss": 20.9141,
+      "step": 6600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.382226195618326e-05,
+      "loss": 18.7472,
+      "step": 6650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.377203644363192e-05,
+      "loss": 18.8577,
+      "step": 6700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.372181093108055e-05,
+      "loss": 17.8061,
+      "step": 6750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.36715854185292e-05,
+      "loss": 19.4687,
+      "step": 6800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.362135990597784e-05,
+      "loss": 19.5103,
+      "step": 6850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.357113439342648e-05,
+      "loss": 18.5319,
+      "step": 6900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.352090888087514e-05,
+      "loss": 20.16,
+      "step": 6950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.347068336832377e-05,
+      "loss": 18.1913,
+      "step": 7000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.342045785577243e-05,
+      "loss": 21.341,
+      "step": 7050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.337023234322106e-05,
+      "loss": 16.7701,
+      "step": 7100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.332000683066972e-05,
+      "loss": 18.045,
+      "step": 7150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.326978131811835e-05,
+      "loss": 16.0393,
+      "step": 7200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.3219555805567e-05,
+      "loss": 17.4833,
+      "step": 7250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316933029301565e-05,
+      "loss": 17.3978,
+      "step": 7300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31191047804643e-05,
+      "loss": 18.2649,
+      "step": 7350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.306887926791294e-05,
+      "loss": 16.3891,
+      "step": 7400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.301865375536157e-05,
+      "loss": 21.4399,
+      "step": 7450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.296842824281023e-05,
+      "loss": 16.3082,
+      "step": 7500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.291820273025886e-05,
+      "loss": 14.8713,
+      "step": 7550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.286797721770751e-05,
+      "loss": 16.3099,
+      "step": 7600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.281775170515616e-05,
+      "loss": 17.8771,
+      "step": 7650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.27675261926048e-05,
+      "loss": 17.1421,
+      "step": 7700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.271730068005345e-05,
+      "loss": 16.6478,
+      "step": 7750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.266707516750209e-05,
+      "loss": 15.3247,
+      "step": 7800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.261684965495073e-05,
+      "loss": 17.6577,
+      "step": 7850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.256662414239938e-05,
+      "loss": 18.8549,
+      "step": 7900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.251639862984802e-05,
+      "loss": 17.4187,
+      "step": 7950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.246617311729667e-05,
+      "loss": 15.6643,
+      "step": 8000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.241594760474531e-05,
+      "loss": 17.1987,
+      "step": 8050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.236572209219396e-05,
+      "loss": 18.1712,
+      "step": 8100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.23154965796426e-05,
+      "loss": 15.8015,
+      "step": 8150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.226527106709124e-05,
+      "loss": 19.064,
+      "step": 8200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.221504555453989e-05,
+      "loss": 18.2748,
+      "step": 8250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.216482004198853e-05,
+      "loss": 15.0679,
+      "step": 8300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.211459452943718e-05,
+      "loss": 17.995,
+      "step": 8350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.206436901688582e-05,
+      "loss": 17.467,
+      "step": 8400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.201414350433448e-05,
+      "loss": 18.6665,
+      "step": 8450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.196391799178311e-05,
+      "loss": 17.2848,
+      "step": 8500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.191369247923175e-05,
+      "loss": 14.4767,
+      "step": 8550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.18634669666804e-05,
+      "loss": 17.5444,
+      "step": 8600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.181324145412904e-05,
+      "loss": 14.4661,
+      "step": 8650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.176301594157768e-05,
+      "loss": 16.3339,
+      "step": 8700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.171279042902633e-05,
+      "loss": 17.5122,
+      "step": 8750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.166256491647499e-05,
+      "loss": 16.7631,
+      "step": 8800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.161233940392362e-05,
+      "loss": 16.5193,
+      "step": 8850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.156211389137227e-05,
+      "loss": 17.8364,
+      "step": 8900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.15118883788209e-05,
+      "loss": 16.2916,
+      "step": 8950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.146166286626956e-05,
+      "loss": 14.1719,
+      "step": 9000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.141143735371819e-05,
+      "loss": 18.2987,
+      "step": 9050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.136121184116684e-05,
+      "loss": 17.4248,
+      "step": 9100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.13109863286155e-05,
+      "loss": 16.1862,
+      "step": 9150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.126076081606412e-05,
+      "loss": 16.3134,
+      "step": 9200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.121053530351278e-05,
+      "loss": 14.9158,
+      "step": 9250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.116030979096141e-05,
+      "loss": 15.2504,
+      "step": 9300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.111008427841007e-05,
+      "loss": 14.1967,
+      "step": 9350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.105985876585871e-05,
+      "loss": 17.3165,
+      "step": 9400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.100963325330736e-05,
+      "loss": 14.5912,
+      "step": 9450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.0959407740756e-05,
+      "loss": 17.5593,
+      "step": 9500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.090918222820465e-05,
+      "loss": 16.3421,
+      "step": 9550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.085895671565329e-05,
+      "loss": 16.2821,
+      "step": 9600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.080873120310192e-05,
+      "loss": 16.4985,
+      "step": 9650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.075850569055058e-05,
+      "loss": 16.1138,
+      "step": 9700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.070828017799922e-05,
+      "loss": 16.3997,
+      "step": 9750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.065805466544787e-05,
+      "loss": 15.518,
+      "step": 9800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.060782915289651e-05,
+      "loss": 13.8424,
+      "step": 9850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.055760364034515e-05,
+      "loss": 15.0784,
+      "step": 9900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.05073781277938e-05,
+      "loss": 14.0163,
+      "step": 9950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.045715261524244e-05,
+      "loss": 16.7863,
+      "step": 10000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.040692710269109e-05,
+      "loss": 13.6715,
+      "step": 10050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.035670159013973e-05,
+      "loss": 15.1071,
+      "step": 10100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.030647607758837e-05,
+      "loss": 14.2658,
+      "step": 10150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.025625056503703e-05,
+      "loss": 15.1115,
+      "step": 10200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.020602505248566e-05,
+      "loss": 14.028,
+      "step": 10250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.015579953993431e-05,
+      "loss": 13.3066,
+      "step": 10300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.010557402738295e-05,
+      "loss": 14.1185,
+      "step": 10350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.00553485148316e-05,
+      "loss": 14.061,
+      "step": 10400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.000512300228024e-05,
+      "loss": 15.2439,
+      "step": 10450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.995489748972888e-05,
+      "loss": 13.3617,
+      "step": 10500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.990467197717754e-05,
+      "loss": 14.5514,
+      "step": 10550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.985444646462617e-05,
+      "loss": 15.2426,
+      "step": 10600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.980422095207483e-05,
+      "loss": 16.6418,
+      "step": 10650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.975399543952346e-05,
+      "loss": 13.3146,
+      "step": 10700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.970376992697212e-05,
+      "loss": 14.9333,
+      "step": 10750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.965354441442075e-05,
+      "loss": 14.4502,
+      "step": 10800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.960331890186939e-05,
+      "loss": 14.7886,
+      "step": 10850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.955309338931805e-05,
+      "loss": 15.0266,
+      "step": 10900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.950286787676668e-05,
+      "loss": 14.543,
+      "step": 10950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.945264236421534e-05,
+      "loss": 15.8078,
+      "step": 11000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.940241685166397e-05,
+      "loss": 13.6052,
+      "step": 11050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.935219133911263e-05,
+      "loss": 14.2995,
+      "step": 11100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.930196582656126e-05,
+      "loss": 15.732,
+      "step": 11150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.925174031400991e-05,
+      "loss": 14.0573,
+      "step": 11200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.920151480145856e-05,
+      "loss": 17.5941,
+      "step": 11250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.91512892889072e-05,
+      "loss": 14.7829,
+      "step": 11300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.910106377635585e-05,
+      "loss": 14.6669,
+      "step": 11350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.905083826380448e-05,
+      "loss": 14.3315,
+      "step": 11400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.900061275125313e-05,
+      "loss": 14.2639,
+      "step": 11450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.895038723870176e-05,
+      "loss": 14.3226,
+      "step": 11500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.890016172615042e-05,
+      "loss": 14.4975,
+      "step": 11550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.884993621359907e-05,
+      "loss": 14.8436,
+      "step": 11600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.879971070104771e-05,
+      "loss": 13.8481,
+      "step": 11650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.874948518849635e-05,
+      "loss": 12.8151,
+      "step": 11700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.8699259675945e-05,
+      "loss": 13.1659,
+      "step": 11750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.864903416339364e-05,
+      "loss": 15.0919,
+      "step": 11800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.859880865084229e-05,
+      "loss": 14.4382,
+      "step": 11850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.854858313829093e-05,
+      "loss": 14.0989,
+      "step": 11900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.849835762573957e-05,
+      "loss": 14.5763,
+      "step": 11950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.844813211318822e-05,
+      "loss": 13.4144,
+      "step": 12000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.839790660063686e-05,
+      "loss": 15.6018,
+      "step": 12050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.83476810880855e-05,
+      "loss": 14.7849,
+      "step": 12100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.829745557553415e-05,
+      "loss": 14.441,
+      "step": 12150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.82472300629828e-05,
+      "loss": 14.2135,
+      "step": 12200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.819700455043144e-05,
+      "loss": 17.1245,
+      "step": 12250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.814677903788008e-05,
+      "loss": 14.6629,
+      "step": 12300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.809655352532873e-05,
+      "loss": 16.6715,
+      "step": 12350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.804632801277738e-05,
+      "loss": 13.0133,
+      "step": 12400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.799610250022601e-05,
+      "loss": 14.1551,
+      "step": 12450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.794587698767466e-05,
+      "loss": 14.019,
+      "step": 12500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.78956514751233e-05,
+      "loss": 14.4279,
+      "step": 12550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.784542596257195e-05,
+      "loss": 12.5293,
+      "step": 12600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.77952004500206e-05,
+      "loss": 15.0403,
+      "step": 12650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.774497493746924e-05,
+      "loss": 13.8193,
+      "step": 12700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.769474942491789e-05,
+      "loss": 13.1564,
+      "step": 12750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.764452391236652e-05,
+      "loss": 14.6415,
+      "step": 12800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.759429839981518e-05,
+      "loss": 12.2339,
+      "step": 12850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.754407288726381e-05,
+      "loss": 12.1604,
+      "step": 12900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.749384737471247e-05,
+      "loss": 15.4939,
+      "step": 12950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.744362186216111e-05,
+      "loss": 13.9713,
+      "step": 13000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.739339634960976e-05,
+      "loss": 14.0986,
+      "step": 13050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.73431708370584e-05,
+      "loss": 13.6334,
+      "step": 13100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.729294532450703e-05,
+      "loss": 13.5201,
+      "step": 13150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.724271981195569e-05,
+      "loss": 14.3793,
+      "step": 13200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.719249429940432e-05,
+      "loss": 13.1741,
+      "step": 13250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.714226878685298e-05,
+      "loss": 11.7782,
+      "step": 13300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.709204327430162e-05,
+      "loss": 12.2758,
+      "step": 13350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.704181776175027e-05,
+      "loss": 13.1723,
+      "step": 13400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.699159224919891e-05,
+      "loss": 14.0858,
+      "step": 13450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.694136673664755e-05,
+      "loss": 11.2836,
+      "step": 13500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.68911412240962e-05,
+      "loss": 15.7226,
+      "step": 13550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.684091571154484e-05,
+      "loss": 15.8889,
+      "step": 13600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.679069019899349e-05,
+      "loss": 12.2185,
+      "step": 13650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.674046468644213e-05,
+      "loss": 11.4647,
+      "step": 13700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.669023917389077e-05,
+      "loss": 13.1238,
+      "step": 13750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.664001366133942e-05,
+      "loss": 11.909,
+      "step": 13800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.658978814878806e-05,
+      "loss": 12.5478,
+      "step": 13850
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.65395626362367e-05,
+      "loss": 13.017,
+      "step": 13900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.648933712368535e-05,
+      "loss": 12.9134,
+      "step": 13950
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.6439111611134e-05,
+      "loss": 13.3485,
+      "step": 14000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.638888609858264e-05,
+      "loss": 11.4706,
+      "step": 14050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.633866058603128e-05,
+      "loss": 11.1063,
+      "step": 14100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.628843507347994e-05,
+      "loss": 12.7408,
+      "step": 14150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.623820956092857e-05,
+      "loss": 12.0689,
+      "step": 14200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.618798404837721e-05,
+      "loss": 11.0724,
+      "step": 14250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.613775853582586e-05,
+      "loss": 12.5685,
+      "step": 14300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.60875330232745e-05,
+      "loss": 12.7776,
+      "step": 14350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.603730751072315e-05,
+      "loss": 11.3066,
+      "step": 14400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.598708199817179e-05,
+      "loss": 13.06,
+      "step": 14450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.593685648562045e-05,
+      "loss": 15.6523,
+      "step": 14500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.588663097306908e-05,
+      "loss": 12.019,
+      "step": 14550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.583640546051774e-05,
+      "loss": 11.0941,
+      "step": 14600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.578617994796637e-05,
+      "loss": 12.4755,
+      "step": 14650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.573595443541502e-05,
+      "loss": 13.7012,
+      "step": 14700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.568572892286366e-05,
+      "loss": 12.2024,
+      "step": 14750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.56355034103123e-05,
+      "loss": 12.4744,
+      "step": 14800
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.558527789776096e-05,
+      "loss": 12.3234,
+      "step": 14850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.553505238520959e-05,
+      "loss": 12.5616,
+      "step": 14900
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.548482687265824e-05,
+      "loss": 11.9559,
+      "step": 14950
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.543460136010688e-05,
+      "loss": 12.0734,
+      "step": 15000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.538437584755553e-05,
+      "loss": 13.0341,
+      "step": 15050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.533415033500418e-05,
+      "loss": 12.7406,
+      "step": 15100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.528392482245282e-05,
+      "loss": 11.7258,
+      "step": 15150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.523369930990147e-05,
+      "loss": 11.8709,
+      "step": 15200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.518347379735011e-05,
+      "loss": 11.7021,
+      "step": 15250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.513324828479875e-05,
+      "loss": 13.2674,
+      "step": 15300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.508302277224738e-05,
+      "loss": 11.9099,
+      "step": 15350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.503279725969604e-05,
+      "loss": 11.7841,
+      "step": 15400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.498257174714469e-05,
+      "loss": 11.9573,
+      "step": 15450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.493234623459333e-05,
+      "loss": 11.7211,
+      "step": 15500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.488212072204197e-05,
+      "loss": 12.3513,
+      "step": 15550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.483189520949062e-05,
+      "loss": 11.0709,
+      "step": 15600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.478166969693926e-05,
+      "loss": 11.6544,
+      "step": 15650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.47314441843879e-05,
+      "loss": 11.8285,
+      "step": 15700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.468121867183655e-05,
+      "loss": 10.4208,
+      "step": 15750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.46309931592852e-05,
+      "loss": 10.7821,
+      "step": 15800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.458076764673384e-05,
+      "loss": 13.2724,
+      "step": 15850
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.45305421341825e-05,
+      "loss": 10.9219,
+      "step": 15900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.448031662163113e-05,
+      "loss": 12.2532,
+      "step": 15950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.443009110907977e-05,
+      "loss": 11.0132,
+      "step": 16000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.437986559652841e-05,
+      "loss": 12.319,
+      "step": 16050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.432964008397706e-05,
+      "loss": 12.9871,
+      "step": 16100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.42794145714257e-05,
+      "loss": 12.0625,
+      "step": 16150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.422918905887435e-05,
+      "loss": 13.4629,
+      "step": 16200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.4178963546323e-05,
+      "loss": 10.9291,
+      "step": 16250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.412873803377163e-05,
+      "loss": 13.7719,
+      "step": 16300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.407851252122029e-05,
+      "loss": 11.3634,
+      "step": 16350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.402828700866892e-05,
+      "loss": 12.7941,
+      "step": 16400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.397806149611758e-05,
+      "loss": 11.8863,
+      "step": 16450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.392783598356621e-05,
+      "loss": 9.5225,
+      "step": 16500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.387761047101485e-05,
+      "loss": 12.983,
+      "step": 16550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.382738495846351e-05,
+      "loss": 11.8489,
+      "step": 16600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.377715944591214e-05,
+      "loss": 11.8122,
+      "step": 16650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.37269339333608e-05,
+      "loss": 12.3387,
+      "step": 16700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.367670842080943e-05,
+      "loss": 13.4648,
+      "step": 16750
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.362648290825809e-05,
+      "loss": 10.2301,
+      "step": 16800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.357625739570672e-05,
+      "loss": 11.492,
+      "step": 16850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.352603188315538e-05,
+      "loss": 12.5997,
+      "step": 16900
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.347580637060402e-05,
+      "loss": 11.5588,
+      "step": 16950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.342558085805266e-05,
+      "loss": 11.8627,
+      "step": 17000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.337535534550131e-05,
+      "loss": 13.2469,
+      "step": 17050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.332512983294994e-05,
+      "loss": 10.4327,
+      "step": 17100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.32749043203986e-05,
+      "loss": 12.7566,
+      "step": 17150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.322467880784723e-05,
+      "loss": 11.0729,
+      "step": 17200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.317445329529588e-05,
+      "loss": 12.3484,
+      "step": 17250
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.312422778274453e-05,
+      "loss": 10.5193,
+      "step": 17300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.307400227019317e-05,
+      "loss": 12.2369,
+      "step": 17350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.302377675764182e-05,
+      "loss": 12.2976,
+      "step": 17400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.297355124509046e-05,
+      "loss": 12.3852,
+      "step": 17450
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.29233257325391e-05,
+      "loss": 11.2137,
+      "step": 17500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.287310021998775e-05,
+      "loss": 11.609,
+      "step": 17550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.282287470743639e-05,
+      "loss": 13.3339,
+      "step": 17600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.277264919488504e-05,
+      "loss": 11.4263,
+      "step": 17650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.272242368233368e-05,
+      "loss": 12.6949,
+      "step": 17700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.267219816978233e-05,
+      "loss": 11.4767,
+      "step": 17750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.262197265723097e-05,
+      "loss": 12.2225,
+      "step": 17800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.257174714467961e-05,
+      "loss": 11.0755,
+      "step": 17850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.252152163212826e-05,
+      "loss": 11.9677,
+      "step": 17900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.24712961195769e-05,
+      "loss": 11.098,
+      "step": 17950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.242107060702555e-05,
+      "loss": 11.1102,
+      "step": 18000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.237084509447419e-05,
+      "loss": 11.4985,
+      "step": 18050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.232061958192285e-05,
+      "loss": 11.7356,
+      "step": 18100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.227039406937148e-05,
+      "loss": 11.3336,
+      "step": 18150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.222016855682012e-05,
+      "loss": 11.0448,
+      "step": 18200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.216994304426877e-05,
+      "loss": 10.9986,
+      "step": 18250
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.211971753171741e-05,
+      "loss": 10.768,
+      "step": 18300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.206949201916607e-05,
+      "loss": 11.6844,
+      "step": 18350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.20192665066147e-05,
+      "loss": 11.5615,
+      "step": 18400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.196904099406336e-05,
+      "loss": 11.4019,
+      "step": 18450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.191881548151199e-05,
+      "loss": 12.1784,
+      "step": 18500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.186858996896064e-05,
+      "loss": 12.4565,
+      "step": 18550
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.181836445640927e-05,
+      "loss": 11.0557,
+      "step": 18600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.176813894385793e-05,
+      "loss": 12.1892,
+      "step": 18650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.171791343130658e-05,
+      "loss": 12.0531,
+      "step": 18700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.166768791875522e-05,
+      "loss": 10.1791,
+      "step": 18750
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.161746240620386e-05,
+      "loss": 11.2501,
+      "step": 18800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.15672368936525e-05,
+      "loss": 9.92,
+      "step": 18850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.151701138110115e-05,
+      "loss": 10.0603,
+      "step": 18900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.146678586854978e-05,
+      "loss": 10.9477,
+      "step": 18950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.141656035599844e-05,
+      "loss": 9.7579,
+      "step": 19000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.136633484344708e-05,
+      "loss": 11.243,
+      "step": 19050
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.131610933089573e-05,
+      "loss": 11.0069,
+      "step": 19100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.126588381834437e-05,
+      "loss": 9.7387,
+      "step": 19150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.121565830579302e-05,
+      "loss": 11.4624,
+      "step": 19200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.116543279324166e-05,
+      "loss": 12.1299,
+      "step": 19250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.11152072806903e-05,
+      "loss": 12.2796,
+      "step": 19300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.106498176813895e-05,
+      "loss": 10.3295,
+      "step": 19350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.101475625558759e-05,
+      "loss": 10.0709,
+      "step": 19400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.096453074303624e-05,
+      "loss": 11.0725,
+      "step": 19450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.091430523048488e-05,
+      "loss": 10.7882,
+      "step": 19500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.086407971793352e-05,
+      "loss": 11.4124,
+      "step": 19550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.081385420538217e-05,
+      "loss": 10.4941,
+      "step": 19600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.076362869283081e-05,
+      "loss": 11.8687,
+      "step": 19650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.071340318027946e-05,
+      "loss": 11.3221,
+      "step": 19700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.06631776677281e-05,
+      "loss": 10.2167,
+      "step": 19750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.061295215517675e-05,
+      "loss": 10.5425,
+      "step": 19800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.05627266426254e-05,
+      "loss": 11.2982,
+      "step": 19850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.051250113007403e-05,
+      "loss": 12.0685,
+      "step": 19900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.046227561752268e-05,
+      "loss": 10.6613,
+      "step": 19950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.041205010497132e-05,
+      "loss": 10.8245,
+      "step": 20000
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 10.409339904785156,
+      "eval_runtime": 890.9956,
+      "eval_samples_per_second": 14.7,
+      "eval_steps_per_second": 3.676,
+      "eval_wer": 0.2624627273109067,
+      "step": 20000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.036182459241997e-05,
+      "loss": 10.671,
+      "step": 20050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.031159907986861e-05,
+      "loss": 11.0263,
+      "step": 20100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.026137356731725e-05,
+      "loss": 11.0571,
+      "step": 20150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.021114805476591e-05,
+      "loss": 13.0778,
+      "step": 20200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.016092254221454e-05,
+      "loss": 11.0495,
+      "step": 20250
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.01106970296632e-05,
+      "loss": 10.6039,
+      "step": 20300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.006047151711183e-05,
+      "loss": 11.4221,
+      "step": 20350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.001024600456049e-05,
+      "loss": 10.7975,
+      "step": 20400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.996002049200912e-05,
+      "loss": 10.1123,
+      "step": 20450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.990979497945776e-05,
+      "loss": 10.2241,
+      "step": 20500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.985956946690642e-05,
+      "loss": 10.0191,
+      "step": 20550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.980934395435505e-05,
+      "loss": 10.649,
+      "step": 20600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.975911844180371e-05,
+      "loss": 9.6091,
+      "step": 20650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.970889292925234e-05,
+      "loss": 9.9386,
+      "step": 20700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.9658667416701e-05,
+      "loss": 11.2646,
+      "step": 20750
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.960844190414964e-05,
+      "loss": 10.0181,
+      "step": 20800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.955821639159828e-05,
+      "loss": 11.9437,
+      "step": 20850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.950799087904693e-05,
+      "loss": 10.9254,
+      "step": 20900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.945776536649557e-05,
+      "loss": 11.7954,
+      "step": 20950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.940753985394422e-05,
+      "loss": 9.6569,
+      "step": 21000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.935731434139286e-05,
+      "loss": 10.6546,
+      "step": 21050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.93070888288415e-05,
+      "loss": 10.2795,
+      "step": 21100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.925686331629015e-05,
+      "loss": 10.4595,
+      "step": 21150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.920663780373879e-05,
+      "loss": 9.2921,
+      "step": 21200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.915641229118744e-05,
+      "loss": 10.1245,
+      "step": 21250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.910618677863608e-05,
+      "loss": 11.2896,
+      "step": 21300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.905596126608472e-05,
+      "loss": 11.3328,
+      "step": 21350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.900573575353337e-05,
+      "loss": 10.0718,
+      "step": 21400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.895551024098201e-05,
+      "loss": 10.8954,
+      "step": 21450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.890528472843066e-05,
+      "loss": 10.2921,
+      "step": 21500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.88550592158793e-05,
+      "loss": 9.4609,
+      "step": 21550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.880483370332796e-05,
+      "loss": 11.4751,
+      "step": 21600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.875460819077659e-05,
+      "loss": 10.1189,
+      "step": 21650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.870438267822523e-05,
+      "loss": 11.6478,
+      "step": 21700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.865415716567388e-05,
+      "loss": 11.2943,
+      "step": 21750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.860393165312252e-05,
+      "loss": 11.5788,
+      "step": 21800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.855370614057116e-05,
+      "loss": 10.638,
+      "step": 21850
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.850348062801981e-05,
+      "loss": 9.2895,
+      "step": 21900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.845325511546847e-05,
+      "loss": 11.4984,
+      "step": 21950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.84030296029171e-05,
+      "loss": 10.3685,
+      "step": 22000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.835280409036575e-05,
+      "loss": 10.0115,
+      "step": 22050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.830257857781439e-05,
+      "loss": 10.2941,
+      "step": 22100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.825235306526304e-05,
+      "loss": 10.8751,
+      "step": 22150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.820212755271167e-05,
+      "loss": 10.7477,
+      "step": 22200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.815190204016032e-05,
+      "loss": 12.2573,
+      "step": 22250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.810167652760897e-05,
+      "loss": 10.1055,
+      "step": 22300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.80514510150576e-05,
+      "loss": 10.7913,
+      "step": 22350
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.800122550250626e-05,
+      "loss": 9.4701,
+      "step": 22400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.79509999899549e-05,
+      "loss": 9.9434,
+      "step": 22450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.790077447740355e-05,
+      "loss": 10.9016,
+      "step": 22500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.785054896485218e-05,
+      "loss": 10.1733,
+      "step": 22550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.780032345230084e-05,
+      "loss": 11.0693,
+      "step": 22600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.775009793974948e-05,
+      "loss": 10.4538,
+      "step": 22650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.769987242719813e-05,
+      "loss": 10.5127,
+      "step": 22700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.764964691464677e-05,
+      "loss": 10.1074,
+      "step": 22750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.75994214020954e-05,
+      "loss": 11.2803,
+      "step": 22800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.754919588954406e-05,
+      "loss": 10.9954,
+      "step": 22850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.749897037699269e-05,
+      "loss": 10.1006,
+      "step": 22900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.744874486444135e-05,
+      "loss": 10.9978,
+      "step": 22950
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.739851935188999e-05,
+      "loss": 10.5885,
+      "step": 23000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.734829383933864e-05,
+      "loss": 10.5676,
+      "step": 23050
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.729806832678728e-05,
+      "loss": 11.3204,
+      "step": 23100
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.724784281423592e-05,
+      "loss": 10.5388,
+      "step": 23150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.719761730168457e-05,
+      "loss": 10.7915,
+      "step": 23200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.714739178913321e-05,
+      "loss": 11.9486,
+      "step": 23250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.709716627658186e-05,
+      "loss": 11.6693,
+      "step": 23300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.70469407640305e-05,
+      "loss": 9.2664,
+      "step": 23350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.699671525147914e-05,
+      "loss": 12.1429,
+      "step": 23400
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.694648973892779e-05,
+      "loss": 10.1155,
+      "step": 23450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.689626422637643e-05,
+      "loss": 10.1562,
+      "step": 23500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.684603871382508e-05,
+      "loss": 11.3484,
+      "step": 23550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.679581320127372e-05,
+      "loss": 9.5912,
+      "step": 23600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.674558768872236e-05,
+      "loss": 11.1067,
+      "step": 23650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.669536217617101e-05,
+      "loss": 11.7182,
+      "step": 23700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.664513666361965e-05,
+      "loss": 10.1444,
+      "step": 23750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.659491115106831e-05,
+      "loss": 11.2671,
+      "step": 23800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.654468563851694e-05,
+      "loss": 10.9027,
+      "step": 23850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.64944601259656e-05,
+      "loss": 10.9078,
+      "step": 23900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.644423461341423e-05,
+      "loss": 10.5441,
+      "step": 23950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.639400910086287e-05,
+      "loss": 9.8617,
+      "step": 24000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.634378358831153e-05,
+      "loss": 10.8022,
+      "step": 24050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.629355807576016e-05,
+      "loss": 10.3082,
+      "step": 24100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.624333256320882e-05,
+      "loss": 9.8398,
+      "step": 24150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.619310705065745e-05,
+      "loss": 10.3631,
+      "step": 24200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.61428815381061e-05,
+      "loss": 10.6078,
+      "step": 24250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.609265602555474e-05,
+      "loss": 11.366,
+      "step": 24300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.60424305130034e-05,
+      "loss": 12.1154,
+      "step": 24350
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.599220500045204e-05,
+      "loss": 11.3429,
+      "step": 24400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.594197948790068e-05,
+      "loss": 9.135,
+      "step": 24450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.589175397534933e-05,
+      "loss": 10.3796,
+      "step": 24500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.584152846279796e-05,
+      "loss": 10.6452,
+      "step": 24550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.579130295024661e-05,
+      "loss": 9.6237,
+      "step": 24600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.574107743769525e-05,
+      "loss": 10.7158,
+      "step": 24650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.56908519251439e-05,
+      "loss": 9.8296,
+      "step": 24700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.564062641259255e-05,
+      "loss": 10.1654,
+      "step": 24750
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.559040090004119e-05,
+      "loss": 10.395,
+      "step": 24800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.554017538748984e-05,
+      "loss": 10.3067,
+      "step": 24850
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.548994987493848e-05,
+      "loss": 10.7243,
+      "step": 24900
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.543972436238712e-05,
+      "loss": 10.4022,
+      "step": 24950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.538949884983577e-05,
+      "loss": 10.5045,
+      "step": 25000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.533927333728441e-05,
+      "loss": 11.2205,
+      "step": 25050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.528904782473306e-05,
+      "loss": 10.5375,
+      "step": 25100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.52388223121817e-05,
+      "loss": 10.4876,
+      "step": 25150
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.518859679963034e-05,
+      "loss": 9.2096,
+      "step": 25200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.513837128707899e-05,
+      "loss": 10.0442,
+      "step": 25250
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.508814577452763e-05,
+      "loss": 9.8174,
+      "step": 25300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.503792026197628e-05,
+      "loss": 10.8789,
+      "step": 25350
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.498769474942492e-05,
+      "loss": 9.8789,
+      "step": 25400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.493746923687356e-05,
+      "loss": 11.1431,
+      "step": 25450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.488724372432221e-05,
+      "loss": 10.4659,
+      "step": 25500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.483701821177087e-05,
+      "loss": 10.7342,
+      "step": 25550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.47867926992195e-05,
+      "loss": 10.7841,
+      "step": 25600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.473656718666814e-05,
+      "loss": 9.6162,
+      "step": 25650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.468634167411678e-05,
+      "loss": 10.3568,
+      "step": 25700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.463611616156543e-05,
+      "loss": 9.6701,
+      "step": 25750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.458589064901407e-05,
+      "loss": 9.4003,
+      "step": 25800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.453566513646272e-05,
+      "loss": 9.6621,
+      "step": 25850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.448543962391137e-05,
+      "loss": 10.1086,
+      "step": 25900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.443521411136e-05,
+      "loss": 11.5655,
+      "step": 25950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.438498859880866e-05,
+      "loss": 8.9418,
+      "step": 26000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.433476308625729e-05,
+      "loss": 9.2415,
+      "step": 26050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.428453757370595e-05,
+      "loss": 9.4192,
+      "step": 26100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.423431206115458e-05,
+      "loss": 9.1755,
+      "step": 26150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.418408654860322e-05,
+      "loss": 9.6327,
+      "step": 26200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.413386103605188e-05,
+      "loss": 10.3333,
+      "step": 26250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.408363552350051e-05,
+      "loss": 10.298,
+      "step": 26300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.403341001094917e-05,
+      "loss": 10.7038,
+      "step": 26350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.39831844983978e-05,
+      "loss": 10.5099,
+      "step": 26400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.393295898584646e-05,
+      "loss": 9.8063,
+      "step": 26450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.38827334732951e-05,
+      "loss": 9.5784,
+      "step": 26500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.383250796074375e-05,
+      "loss": 10.1958,
+      "step": 26550
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.378228244819239e-05,
+      "loss": 9.6869,
+      "step": 26600
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.373205693564103e-05,
+      "loss": 10.3761,
+      "step": 26650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.368183142308968e-05,
+      "loss": 11.6806,
+      "step": 26700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.363160591053832e-05,
+      "loss": 10.3183,
+      "step": 26750
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.358138039798697e-05,
+      "loss": 11.041,
+      "step": 26800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.353115488543561e-05,
+      "loss": 9.6997,
+      "step": 26850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.348092937288425e-05,
+      "loss": 9.6029,
+      "step": 26900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.34307038603329e-05,
+      "loss": 10.3322,
+      "step": 26950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.338047834778154e-05,
+      "loss": 9.9009,
+      "step": 27000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.333025283523019e-05,
+      "loss": 10.4815,
+      "step": 27050
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.328002732267883e-05,
+      "loss": 11.7049,
+      "step": 27100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.322980181012748e-05,
+      "loss": 10.7831,
+      "step": 27150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.317957629757612e-05,
+      "loss": 8.735,
+      "step": 27200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.312935078502476e-05,
+      "loss": 9.4056,
+      "step": 27250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.307912527247342e-05,
+      "loss": 10.7689,
+      "step": 27300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.302889975992205e-05,
+      "loss": 9.5266,
+      "step": 27350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.29786742473707e-05,
+      "loss": 8.2467,
+      "step": 27400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.292844873481934e-05,
+      "loss": 8.6572,
+      "step": 27450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.287822322226798e-05,
+      "loss": 8.4693,
+      "step": 27500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.282799770971663e-05,
+      "loss": 10.4867,
+      "step": 27550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.277777219716527e-05,
+      "loss": 8.9364,
+      "step": 27600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.272754668461393e-05,
+      "loss": 10.0109,
+      "step": 27650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.267732117206256e-05,
+      "loss": 9.5535,
+      "step": 27700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.262709565951122e-05,
+      "loss": 9.3029,
+      "step": 27750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.257687014695985e-05,
+      "loss": 9.854,
+      "step": 27800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.25266446344085e-05,
+      "loss": 9.5327,
+      "step": 27850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.247641912185714e-05,
+      "loss": 9.8255,
+      "step": 27900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.242619360930578e-05,
+      "loss": 9.9737,
+      "step": 27950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.237596809675444e-05,
+      "loss": 9.0471,
+      "step": 28000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.232574258420307e-05,
+      "loss": 10.0566,
+      "step": 28050
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.227551707165173e-05,
+      "loss": 9.4781,
+      "step": 28100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.222529155910036e-05,
+      "loss": 8.7599,
+      "step": 28150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.217506604654901e-05,
+      "loss": 8.7605,
+      "step": 28200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.212484053399764e-05,
+      "loss": 10.061,
+      "step": 28250
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.20746150214463e-05,
+      "loss": 9.6124,
+      "step": 28300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.202438950889495e-05,
+      "loss": 10.4776,
+      "step": 28350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.197416399634359e-05,
+      "loss": 9.2169,
+      "step": 28400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.192393848379223e-05,
+      "loss": 9.3654,
+      "step": 28450
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.187371297124086e-05,
+      "loss": 9.4445,
+      "step": 28500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.182348745868952e-05,
+      "loss": 8.3614,
+      "step": 28550
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.177326194613815e-05,
+      "loss": 9.1661,
+      "step": 28600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.172303643358681e-05,
+      "loss": 9.4976,
+      "step": 28650
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.167281092103545e-05,
+      "loss": 9.125,
+      "step": 28700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.16225854084841e-05,
+      "loss": 8.9051,
+      "step": 28750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.157235989593274e-05,
+      "loss": 8.9753,
+      "step": 28800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.152213438338139e-05,
+      "loss": 9.133,
+      "step": 28850
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.147190887083003e-05,
+      "loss": 9.9677,
+      "step": 28900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.142168335827867e-05,
+      "loss": 8.725,
+      "step": 28950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.137145784572732e-05,
+      "loss": 8.831,
+      "step": 29000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.132123233317596e-05,
+      "loss": 7.8207,
+      "step": 29050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.127100682062461e-05,
+      "loss": 9.3707,
+      "step": 29100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.122078130807325e-05,
+      "loss": 10.4259,
+      "step": 29150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.11705557955219e-05,
+      "loss": 8.1836,
+      "step": 29200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.112033028297054e-05,
+      "loss": 9.0874,
+      "step": 29250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.107010477041918e-05,
+      "loss": 9.5957,
+      "step": 29300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.101987925786783e-05,
+      "loss": 8.7545,
+      "step": 29350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.096965374531647e-05,
+      "loss": 8.4478,
+      "step": 29400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.091942823276512e-05,
+      "loss": 8.601,
+      "step": 29450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.086920272021377e-05,
+      "loss": 9.6172,
+      "step": 29500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.08189772076624e-05,
+      "loss": 9.0805,
+      "step": 29550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.076875169511106e-05,
+      "loss": 9.6039,
+      "step": 29600
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.071852618255969e-05,
+      "loss": 9.3622,
+      "step": 29650
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.066830067000834e-05,
+      "loss": 8.8765,
+      "step": 29700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.061807515745699e-05,
+      "loss": 8.992,
+      "step": 29750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.056784964490562e-05,
+      "loss": 10.3564,
+      "step": 29800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.051762413235428e-05,
+      "loss": 8.8092,
+      "step": 29850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.046739861980291e-05,
+      "loss": 9.8373,
+      "step": 29900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.041717310725157e-05,
+      "loss": 8.004,
+      "step": 29950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.03669475947002e-05,
+      "loss": 9.4461,
+      "step": 30000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.031672208214886e-05,
+      "loss": 8.4964,
+      "step": 30050
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.02664965695975e-05,
+      "loss": 10.3181,
+      "step": 30100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.021627105704615e-05,
+      "loss": 8.6637,
+      "step": 30150
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.016604554449479e-05,
+      "loss": 10.1703,
+      "step": 30200
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.011582003194342e-05,
+      "loss": 9.2846,
+      "step": 30250
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.006559451939208e-05,
+      "loss": 8.5913,
+      "step": 30300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.001536900684071e-05,
+      "loss": 9.1308,
+      "step": 30350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.996514349428937e-05,
+      "loss": 11.2229,
+      "step": 30400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.991491798173801e-05,
+      "loss": 8.5923,
+      "step": 30450
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.986469246918665e-05,
+      "loss": 9.9826,
+      "step": 30500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.98144669566353e-05,
+      "loss": 8.4765,
+      "step": 30550
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.976424144408394e-05,
+      "loss": 8.7624,
+      "step": 30600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.971401593153259e-05,
+      "loss": 9.238,
+      "step": 30650
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.966379041898123e-05,
+      "loss": 8.4976,
+      "step": 30700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.961356490642987e-05,
+      "loss": 9.1886,
+      "step": 30750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.956333939387852e-05,
+      "loss": 8.4443,
+      "step": 30800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.951311388132716e-05,
+      "loss": 8.3648,
+      "step": 30850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.94628883687758e-05,
+      "loss": 9.2509,
+      "step": 30900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.941266285622445e-05,
+      "loss": 8.3765,
+      "step": 30950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.93624373436731e-05,
+      "loss": 9.6616,
+      "step": 31000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.931221183112174e-05,
+      "loss": 9.658,
+      "step": 31050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.926198631857038e-05,
+      "loss": 8.7527,
+      "step": 31100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.921176080601903e-05,
+      "loss": 8.7148,
+      "step": 31150
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.916153529346767e-05,
+      "loss": 8.5962,
+      "step": 31200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.911130978091633e-05,
+      "loss": 9.2625,
+      "step": 31250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.906108426836496e-05,
+      "loss": 8.8352,
+      "step": 31300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.90108587558136e-05,
+      "loss": 7.3991,
+      "step": 31350
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.896063324326225e-05,
+      "loss": 9.9391,
+      "step": 31400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.891040773071089e-05,
+      "loss": 8.9575,
+      "step": 31450
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.886018221815954e-05,
+      "loss": 7.9103,
+      "step": 31500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.880995670560818e-05,
+      "loss": 8.5276,
+      "step": 31550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.875973119305684e-05,
+      "loss": 8.5427,
+      "step": 31600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.870950568050547e-05,
+      "loss": 8.4672,
+      "step": 31650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.865928016795412e-05,
+      "loss": 8.9638,
+      "step": 31700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.860905465540276e-05,
+      "loss": 8.3136,
+      "step": 31750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.855882914285141e-05,
+      "loss": 8.8076,
+      "step": 31800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.850860363030004e-05,
+      "loss": 8.6041,
+      "step": 31850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.845837811774869e-05,
+      "loss": 9.1751,
+      "step": 31900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.840815260519735e-05,
+      "loss": 8.5955,
+      "step": 31950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.835792709264598e-05,
+      "loss": 9.0927,
+      "step": 32000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.830770158009463e-05,
+      "loss": 7.9647,
+      "step": 32050
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.825747606754326e-05,
+      "loss": 10.2647,
+      "step": 32100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.820725055499192e-05,
+      "loss": 8.3442,
+      "step": 32150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.815702504244057e-05,
+      "loss": 9.2019,
+      "step": 32200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.810679952988921e-05,
+      "loss": 8.345,
+      "step": 32250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.805657401733785e-05,
+      "loss": 9.1835,
+      "step": 32300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.80063485047865e-05,
+      "loss": 9.1846,
+      "step": 32350
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.795612299223514e-05,
+      "loss": 9.0015,
+      "step": 32400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.790589747968379e-05,
+      "loss": 8.2404,
+      "step": 32450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.785567196713243e-05,
+      "loss": 8.8715,
+      "step": 32500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.780544645458107e-05,
+      "loss": 8.817,
+      "step": 32550
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.775522094202972e-05,
+      "loss": 9.2154,
+      "step": 32600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.770499542947836e-05,
+      "loss": 9.1914,
+      "step": 32650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.7654769916927e-05,
+      "loss": 9.2804,
+      "step": 32700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.760454440437565e-05,
+      "loss": 9.177,
+      "step": 32750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.75543188918243e-05,
+      "loss": 8.8259,
+      "step": 32800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.750409337927294e-05,
+      "loss": 8.6121,
+      "step": 32850
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.745386786672158e-05,
+      "loss": 8.644,
+      "step": 32900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.740364235417023e-05,
+      "loss": 8.5743,
+      "step": 32950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.735341684161888e-05,
+      "loss": 8.7636,
+      "step": 33000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.730319132906751e-05,
+      "loss": 8.3064,
+      "step": 33050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.725296581651616e-05,
+      "loss": 8.8806,
+      "step": 33100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.72027403039648e-05,
+      "loss": 8.8212,
+      "step": 33150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.715251479141345e-05,
+      "loss": 9.5261,
+      "step": 33200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.710228927886209e-05,
+      "loss": 9.0764,
+      "step": 33250
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.705206376631073e-05,
+      "loss": 7.399,
+      "step": 33300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.700183825375939e-05,
+      "loss": 9.4119,
+      "step": 33350
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.695161274120802e-05,
+      "loss": 8.4576,
+      "step": 33400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.690138722865668e-05,
+      "loss": 8.024,
+      "step": 33450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.685116171610531e-05,
+      "loss": 9.1605,
+      "step": 33500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.680093620355397e-05,
+      "loss": 8.3661,
+      "step": 33550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.67507106910026e-05,
+      "loss": 8.4145,
+      "step": 33600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.670048517845124e-05,
+      "loss": 7.824,
+      "step": 33650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.66502596658999e-05,
+      "loss": 9.129,
+      "step": 33700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.660003415334853e-05,
+      "loss": 9.0876,
+      "step": 33750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.654980864079719e-05,
+      "loss": 8.6961,
+      "step": 33800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.649958312824582e-05,
+      "loss": 8.1584,
+      "step": 33850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.644935761569448e-05,
+      "loss": 8.6587,
+      "step": 33900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.639913210314311e-05,
+      "loss": 8.1059,
+      "step": 33950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.634890659059176e-05,
+      "loss": 9.2588,
+      "step": 34000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.629868107804041e-05,
+      "loss": 8.6443,
+      "step": 34050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.624845556548905e-05,
+      "loss": 8.8006,
+      "step": 34100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.61982300529377e-05,
+      "loss": 9.2288,
+      "step": 34150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.614800454038633e-05,
+      "loss": 9.0328,
+      "step": 34200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.609777902783499e-05,
+      "loss": 7.8269,
+      "step": 34250
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.604755351528362e-05,
+      "loss": 8.5883,
+      "step": 34300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.599732800273227e-05,
+      "loss": 9.9388,
+      "step": 34350
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.594710249018092e-05,
+      "loss": 8.6776,
+      "step": 34400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.589687697762956e-05,
+      "loss": 7.2287,
+      "step": 34450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.58466514650782e-05,
+      "loss": 7.7042,
+      "step": 34500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.579642595252685e-05,
+      "loss": 9.0004,
+      "step": 34550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.57462004399755e-05,
+      "loss": 9.3279,
+      "step": 34600
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.569597492742414e-05,
+      "loss": 8.9144,
+      "step": 34650
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.564574941487278e-05,
+      "loss": 9.3319,
+      "step": 34700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.559552390232143e-05,
+      "loss": 9.4986,
+      "step": 34750
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.554529838977007e-05,
+      "loss": 9.002,
+      "step": 34800
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.549507287721871e-05,
+      "loss": 8.6061,
+      "step": 34850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.544484736466736e-05,
+      "loss": 7.4598,
+      "step": 34900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.5394621852116e-05,
+      "loss": 8.6618,
+      "step": 34950
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.534439633956465e-05,
+      "loss": 9.0226,
+      "step": 35000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.529417082701329e-05,
+      "loss": 7.9738,
+      "step": 35050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.524394531446193e-05,
+      "loss": 8.7871,
+      "step": 35100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.519371980191058e-05,
+      "loss": 8.8744,
+      "step": 35150
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.514349428935924e-05,
+      "loss": 8.3771,
+      "step": 35200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.509326877680787e-05,
+      "loss": 8.058,
+      "step": 35250
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.504304326425652e-05,
+      "loss": 8.2627,
+      "step": 35300
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.499281775170515e-05,
+      "loss": 8.1643,
+      "step": 35350
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.49425922391538e-05,
+      "loss": 7.888,
+      "step": 35400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.489236672660246e-05,
+      "loss": 7.9235,
+      "step": 35450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.484214121405109e-05,
+      "loss": 8.1139,
+      "step": 35500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.479191570149974e-05,
+      "loss": 8.7467,
+      "step": 35550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.474169018894837e-05,
+      "loss": 7.4693,
+      "step": 35600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.469146467639703e-05,
+      "loss": 8.5167,
+      "step": 35650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.464123916384566e-05,
+      "loss": 9.5274,
+      "step": 35700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.459101365129432e-05,
+      "loss": 8.9735,
+      "step": 35750
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.454078813874296e-05,
+      "loss": 8.1756,
+      "step": 35800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.449056262619161e-05,
+      "loss": 7.8084,
+      "step": 35850
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.444033711364025e-05,
+      "loss": 8.2671,
+      "step": 35900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.439011160108888e-05,
+      "loss": 8.6628,
+      "step": 35950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.433988608853754e-05,
+      "loss": 9.8654,
+      "step": 36000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.428966057598617e-05,
+      "loss": 9.104,
+      "step": 36050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.423943506343483e-05,
+      "loss": 9.4156,
+      "step": 36100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.418920955088347e-05,
+      "loss": 8.9803,
+      "step": 36150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.413898403833212e-05,
+      "loss": 8.9584,
+      "step": 36200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.408875852578076e-05,
+      "loss": 7.3683,
+      "step": 36250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.40385330132294e-05,
+      "loss": 8.3277,
+      "step": 36300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.398830750067805e-05,
+      "loss": 9.3236,
+      "step": 36350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.393808198812669e-05,
+      "loss": 8.6918,
+      "step": 36400
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.388785647557534e-05,
+      "loss": 8.9422,
+      "step": 36450
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.383763096302398e-05,
+      "loss": 8.8438,
+      "step": 36500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.378740545047263e-05,
+      "loss": 8.7752,
+      "step": 36550
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.373717993792127e-05,
+      "loss": 8.6483,
+      "step": 36600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.368695442536991e-05,
+      "loss": 8.5753,
+      "step": 36650
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.363672891281856e-05,
+      "loss": 8.1893,
+      "step": 36700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.35865034002672e-05,
+      "loss": 8.189,
+      "step": 36750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.353627788771585e-05,
+      "loss": 8.2979,
+      "step": 36800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.348605237516449e-05,
+      "loss": 8.3904,
+      "step": 36850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.343582686261313e-05,
+      "loss": 9.3274,
+      "step": 36900
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.338560135006179e-05,
+      "loss": 7.7663,
+      "step": 36950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.333537583751042e-05,
+      "loss": 8.2105,
+      "step": 37000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.328515032495907e-05,
+      "loss": 8.035,
+      "step": 37050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.323492481240771e-05,
+      "loss": 9.5032,
+      "step": 37100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.318469929985635e-05,
+      "loss": 8.3856,
+      "step": 37150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.3134473787305e-05,
+      "loss": 8.9941,
+      "step": 37200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.308424827475364e-05,
+      "loss": 8.3987,
+      "step": 37250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.30340227622023e-05,
+      "loss": 9.1753,
+      "step": 37300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.298379724965093e-05,
+      "loss": 7.9557,
+      "step": 37350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.293357173709959e-05,
+      "loss": 7.725,
+      "step": 37400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.288334622454822e-05,
+      "loss": 8.0807,
+      "step": 37450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.283312071199688e-05,
+      "loss": 8.6492,
+      "step": 37500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.27828951994455e-05,
+      "loss": 8.4716,
+      "step": 37550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.273266968689416e-05,
+      "loss": 8.7209,
+      "step": 37600
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.268244417434281e-05,
+      "loss": 8.4902,
+      "step": 37650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.263221866179144e-05,
+      "loss": 7.9589,
+      "step": 37700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.25819931492401e-05,
+      "loss": 9.3285,
+      "step": 37750
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.253176763668873e-05,
+      "loss": 9.0506,
+      "step": 37800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.248154212413738e-05,
+      "loss": 7.9992,
+      "step": 37850
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.243131661158603e-05,
+      "loss": 8.029,
+      "step": 37900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.238109109903467e-05,
+      "loss": 8.6681,
+      "step": 37950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.233086558648332e-05,
+      "loss": 8.5906,
+      "step": 38000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.228064007393196e-05,
+      "loss": 10.4719,
+      "step": 38050
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.22304145613806e-05,
+      "loss": 8.2759,
+      "step": 38100
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.218018904882925e-05,
+      "loss": 8.2633,
+      "step": 38150
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.212996353627789e-05,
+      "loss": 8.5218,
+      "step": 38200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.207973802372654e-05,
+      "loss": 8.0609,
+      "step": 38250
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.202951251117518e-05,
+      "loss": 9.3672,
+      "step": 38300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.197928699862382e-05,
+      "loss": 10.1768,
+      "step": 38350
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.192906148607247e-05,
+      "loss": 9.4389,
+      "step": 38400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.187883597352111e-05,
+      "loss": 7.6737,
+      "step": 38450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.182861046096976e-05,
+      "loss": 9.2337,
+      "step": 38500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.17783849484184e-05,
+      "loss": 8.7846,
+      "step": 38550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.172815943586704e-05,
+      "loss": 7.8709,
+      "step": 38600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.167793392331569e-05,
+      "loss": 8.8688,
+      "step": 38650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.162770841076435e-05,
+      "loss": 8.4087,
+      "step": 38700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.157748289821298e-05,
+      "loss": 7.7129,
+      "step": 38750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.152725738566162e-05,
+      "loss": 9.3196,
+      "step": 38800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.147703187311027e-05,
+      "loss": 8.8242,
+      "step": 38850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.142680636055891e-05,
+      "loss": 8.4237,
+      "step": 38900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.137658084800755e-05,
+      "loss": 8.9383,
+      "step": 38950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.13263553354562e-05,
+      "loss": 8.3749,
+      "step": 39000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.127612982290485e-05,
+      "loss": 8.8894,
+      "step": 39050
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.122590431035349e-05,
+      "loss": 8.2975,
+      "step": 39100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.117567879780214e-05,
+      "loss": 8.0517,
+      "step": 39150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.112545328525077e-05,
+      "loss": 8.0154,
+      "step": 39200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.107522777269943e-05,
+      "loss": 8.4887,
+      "step": 39250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.102500226014807e-05,
+      "loss": 8.7064,
+      "step": 39300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.0974776747596706e-05,
+      "loss": 9.7375,
+      "step": 39350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.0924551235045357e-05,
+      "loss": 8.8614,
+      "step": 39400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.0874325722493994e-05,
+      "loss": 8.302,
+      "step": 39450
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.0824100209942645e-05,
+      "loss": 7.8469,
+      "step": 39500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.077387469739129e-05,
+      "loss": 9.0706,
+      "step": 39550
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.072364918483994e-05,
+      "loss": 9.1398,
+      "step": 39600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.067342367228858e-05,
+      "loss": 8.1838,
+      "step": 39650
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.062319815973723e-05,
+      "loss": 9.2303,
+      "step": 39700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.0572972647185865e-05,
+      "loss": 8.3715,
+      "step": 39750
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.0522747134634516e-05,
+      "loss": 8.409,
+      "step": 39800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.047252162208315e-05,
+      "loss": 8.6441,
+      "step": 39850
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.04222961095318e-05,
+      "loss": 9.0975,
+      "step": 39900
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.037207059698045e-05,
+      "loss": 8.0691,
+      "step": 39950
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.0321845084429085e-05,
+      "loss": 8.6646,
+      "step": 40000
+    },
+    {
+      "epoch": 2.95,
+      "eval_loss": 8.163222312927246,
+      "eval_runtime": 957.6189,
+      "eval_samples_per_second": 13.678,
+      "eval_steps_per_second": 3.42,
+      "eval_wer": 0.22493805384066187,
+      "step": 40000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.0271619571877736e-05,
+      "loss": 8.4278,
+      "step": 40050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.022139405932637e-05,
+      "loss": 8.1656,
+      "step": 40100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.0171168546775024e-05,
+      "loss": 7.7975,
+      "step": 40150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.012094303422366e-05,
+      "loss": 7.5465,
+      "step": 40200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.007071752167231e-05,
+      "loss": 8.3986,
+      "step": 40250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.0020492009120956e-05,
+      "loss": 8.3762,
+      "step": 40300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.997026649656961e-05,
+      "loss": 8.6175,
+      "step": 40350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.9920040984018244e-05,
+      "loss": 8.5622,
+      "step": 40400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.9869815471466895e-05,
+      "loss": 8.1824,
+      "step": 40450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.981958995891553e-05,
+      "loss": 7.2886,
+      "step": 40500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.9769364446364177e-05,
+      "loss": 8.3469,
+      "step": 40550
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.971913893381282e-05,
+      "loss": 8.6257,
+      "step": 40600
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.9668913421261465e-05,
+      "loss": 7.7071,
+      "step": 40650
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.9618687908710116e-05,
+      "loss": 7.8413,
+      "step": 40700
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.956846239615875e-05,
+      "loss": 7.6704,
+      "step": 40750
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.9518236883607404e-05,
+      "loss": 7.3902,
+      "step": 40800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.946801137105604e-05,
+      "loss": 8.3296,
+      "step": 40850
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.941778585850469e-05,
+      "loss": 7.0884,
+      "step": 40900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.9367560345953336e-05,
+      "loss": 7.043,
+      "step": 40950
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.931733483340198e-05,
+      "loss": 7.5367,
+      "step": 41000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.9267109320850624e-05,
+      "loss": 8.3064,
+      "step": 41050
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.921688380829926e-05,
+      "loss": 7.6769,
+      "step": 41100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.916665829574791e-05,
+      "loss": 9.067,
+      "step": 41150
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.911643278319655e-05,
+      "loss": 8.3565,
+      "step": 41200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.90662072706452e-05,
+      "loss": 7.8335,
+      "step": 41250
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.9015981758093844e-05,
+      "loss": 7.9617,
+      "step": 41300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.8965756245542495e-05,
+      "loss": 8.6728,
+      "step": 41350
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.891553073299113e-05,
+      "loss": 7.9142,
+      "step": 41400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.886530522043978e-05,
+      "loss": 7.7702,
+      "step": 41450
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.881507970788842e-05,
+      "loss": 8.2997,
+      "step": 41500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.876485419533707e-05,
+      "loss": 8.1519,
+      "step": 41550
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.871462868278571e-05,
+      "loss": 7.3762,
+      "step": 41600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.866440317023435e-05,
+      "loss": 7.5129,
+      "step": 41650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.8614177657683e-05,
+      "loss": 8.2537,
+      "step": 41700
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.856395214513164e-05,
+      "loss": 8.4148,
+      "step": 41750
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.851372663258029e-05,
+      "loss": 7.1737,
+      "step": 41800
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.846350112002893e-05,
+      "loss": 7.2628,
+      "step": 41850
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.841327560747758e-05,
+      "loss": 7.2933,
+      "step": 41900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.836305009492622e-05,
+      "loss": 7.7675,
+      "step": 41950
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.831282458237487e-05,
+      "loss": 8.2344,
+      "step": 42000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.826259906982351e-05,
+      "loss": 7.1329,
+      "step": 42050
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.821237355727216e-05,
+      "loss": 7.3924,
+      "step": 42100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.81621480447208e-05,
+      "loss": 6.6189,
+      "step": 42150
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.811192253216944e-05,
+      "loss": 7.3457,
+      "step": 42200
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.806169701961809e-05,
+      "loss": 8.9924,
+      "step": 42250
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.8011471507066725e-05,
+      "loss": 7.6315,
+      "step": 42300
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.7961245994515376e-05,
+      "loss": 8.4726,
+      "step": 42350
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.791102048196402e-05,
+      "loss": 7.1755,
+      "step": 42400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.786079496941267e-05,
+      "loss": 7.5716,
+      "step": 42450
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.781056945686131e-05,
+      "loss": 7.938,
+      "step": 42500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.776034394430996e-05,
+      "loss": 7.3833,
+      "step": 42550
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.7710118431758596e-05,
+      "loss": 6.4276,
+      "step": 42600
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.765989291920725e-05,
+      "loss": 6.8907,
+      "step": 42650
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.7609667406655884e-05,
+      "loss": 7.7592,
+      "step": 42700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.755944189410453e-05,
+      "loss": 7.4997,
+      "step": 42750
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.750921638155318e-05,
+      "loss": 7.2821,
+      "step": 42800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.745899086900182e-05,
+      "loss": 7.4861,
+      "step": 42850
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.740876535645047e-05,
+      "loss": 7.9266,
+      "step": 42900
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.7358539843899105e-05,
+      "loss": 7.6244,
+      "step": 42950
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.7308314331347756e-05,
+      "loss": 7.382,
+      "step": 43000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.725808881879639e-05,
+      "loss": 8.1925,
+      "step": 43050
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.7207863306245044e-05,
+      "loss": 8.3185,
+      "step": 43100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.715763779369369e-05,
+      "loss": 7.091,
+      "step": 43150
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.710741228114234e-05,
+      "loss": 7.8352,
+      "step": 43200
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.7057186768590976e-05,
+      "loss": 6.6085,
+      "step": 43250
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.700696125603963e-05,
+      "loss": 7.8052,
+      "step": 43300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6956735743488264e-05,
+      "loss": 8.1999,
+      "step": 43350
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.690651023093691e-05,
+      "loss": 7.2801,
+      "step": 43400
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.685628471838555e-05,
+      "loss": 7.6289,
+      "step": 43450
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6806059205834196e-05,
+      "loss": 6.8215,
+      "step": 43500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.675583369328285e-05,
+      "loss": 7.1678,
+      "step": 43550
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6705608180731484e-05,
+      "loss": 7.6612,
+      "step": 43600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6655382668180135e-05,
+      "loss": 7.8899,
+      "step": 43650
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.660515715562877e-05,
+      "loss": 7.8546,
+      "step": 43700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.655493164307742e-05,
+      "loss": 7.319,
+      "step": 43750
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.650470613052607e-05,
+      "loss": 7.3317,
+      "step": 43800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.645448061797471e-05,
+      "loss": 7.8875,
+      "step": 43850
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.6404255105423355e-05,
+      "loss": 7.8145,
+      "step": 43900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.635402959287199e-05,
+      "loss": 7.0667,
+      "step": 43950
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.6303804080320643e-05,
+      "loss": 7.7603,
+      "step": 44000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.625357856776928e-05,
+      "loss": 7.6111,
+      "step": 44050
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.620335305521793e-05,
+      "loss": 7.9858,
+      "step": 44100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6153127542666576e-05,
+      "loss": 8.9896,
+      "step": 44150
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6102902030115226e-05,
+      "loss": 8.4081,
+      "step": 44200
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6052676517563864e-05,
+      "loss": 7.4748,
+      "step": 44250
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6002451005012515e-05,
+      "loss": 8.2133,
+      "step": 44300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.595222549246115e-05,
+      "loss": 7.3073,
+      "step": 44350
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.59019999799098e-05,
+      "loss": 7.9638,
+      "step": 44400
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.585177446735844e-05,
+      "loss": 7.9653,
+      "step": 44450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5801548954807084e-05,
+      "loss": 7.8583,
+      "step": 44500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5751323442255735e-05,
+      "loss": 8.0561,
+      "step": 44550
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.570109792970437e-05,
+      "loss": 8.1276,
+      "step": 44600
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.565087241715302e-05,
+      "loss": 7.7357,
+      "step": 44650
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.560064690460166e-05,
+      "loss": 7.7529,
+      "step": 44700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.555042139205031e-05,
+      "loss": 7.2583,
+      "step": 44750
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.550019587949895e-05,
+      "loss": 6.4675,
+      "step": 44800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.54499703669476e-05,
+      "loss": 7.3658,
+      "step": 44850
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.539974485439624e-05,
+      "loss": 8.278,
+      "step": 44900
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.5349519341844894e-05,
+      "loss": 7.3867,
+      "step": 44950
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.529929382929353e-05,
+      "loss": 7.4187,
+      "step": 45000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.524906831674217e-05,
+      "loss": 7.5281,
+      "step": 45050
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.519884280419082e-05,
+      "loss": 7.8815,
+      "step": 45100
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.514861729163946e-05,
+      "loss": 7.2487,
+      "step": 45150
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.509839177908811e-05,
+      "loss": 8.3441,
+      "step": 45200
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.504816626653675e-05,
+      "loss": 7.4892,
+      "step": 45250
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.49979407539854e-05,
+      "loss": 7.7789,
+      "step": 45300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.494771524143404e-05,
+      "loss": 7.3951,
+      "step": 45350
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.489748972888269e-05,
+      "loss": 7.8756,
+      "step": 45400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.484726421633133e-05,
+      "loss": 7.9274,
+      "step": 45450
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.479703870377998e-05,
+      "loss": 8.1525,
+      "step": 45500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.4746813191228616e-05,
+      "loss": 7.5597,
+      "step": 45550
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.469658767867726e-05,
+      "loss": 7.8939,
+      "step": 45600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.464636216612591e-05,
+      "loss": 6.1451,
+      "step": 45650
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.459613665357455e-05,
+      "loss": 7.224,
+      "step": 45700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.45459111410232e-05,
+      "loss": 7.2489,
+      "step": 45750
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.4495685628471836e-05,
+      "loss": 7.4162,
+      "step": 45800
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.444546011592049e-05,
+      "loss": 6.8503,
+      "step": 45850
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.4395234603369124e-05,
+      "loss": 6.7087,
+      "step": 45900
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.4345009090817775e-05,
+      "loss": 6.9697,
+      "step": 45950
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.429478357826642e-05,
+      "loss": 7.8369,
+      "step": 46000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.424455806571507e-05,
+      "loss": 7.7567,
+      "step": 46050
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.419433255316371e-05,
+      "loss": 6.6241,
+      "step": 46100
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.414410704061236e-05,
+      "loss": 7.5218,
+      "step": 46150
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.4093881528060995e-05,
+      "loss": 7.2338,
+      "step": 46200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.404365601550964e-05,
+      "loss": 7.0707,
+      "step": 46250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.3993430502958283e-05,
+      "loss": 7.6922,
+      "step": 46300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.394320499040693e-05,
+      "loss": 8.6056,
+      "step": 46350
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.389297947785558e-05,
+      "loss": 7.4641,
+      "step": 46400
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.3842753965304216e-05,
+      "loss": 7.1716,
+      "step": 46450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.3792528452752866e-05,
+      "loss": 7.6382,
+      "step": 46500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.3742302940201504e-05,
+      "loss": 7.0739,
+      "step": 46550
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.3692077427650155e-05,
+      "loss": 7.8667,
+      "step": 46600
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.36418519150988e-05,
+      "loss": 7.559,
+      "step": 46650
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.359162640254744e-05,
+      "loss": 7.6078,
+      "step": 46700
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.354140088999609e-05,
+      "loss": 7.7994,
+      "step": 46750
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.3491175377444724e-05,
+      "loss": 7.0418,
+      "step": 46800
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.3440949864893375e-05,
+      "loss": 7.3261,
+      "step": 46850
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.339072435234201e-05,
+      "loss": 7.9914,
+      "step": 46900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.334049883979066e-05,
+      "loss": 7.1998,
+      "step": 46950
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.329027332723931e-05,
+      "loss": 7.3343,
+      "step": 47000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.324004781468796e-05,
+      "loss": 8.1604,
+      "step": 47050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.3189822302136595e-05,
+      "loss": 7.5405,
+      "step": 47100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.3139596789585246e-05,
+      "loss": 7.9409,
+      "step": 47150
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.308937127703388e-05,
+      "loss": 8.0573,
+      "step": 47200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.3039145764482534e-05,
+      "loss": 7.2927,
+      "step": 47250
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.298892025193117e-05,
+      "loss": 6.9476,
+      "step": 47300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.2938694739379815e-05,
+      "loss": 7.1999,
+      "step": 47350
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.2888469226828466e-05,
+      "loss": 7.8224,
+      "step": 47400
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.2838243714277103e-05,
+      "loss": 8.1369,
+      "step": 47450
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.2788018201725754e-05,
+      "loss": 6.7302,
+      "step": 47500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.273779268917439e-05,
+      "loss": 8.0819,
+      "step": 47550
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.268756717662304e-05,
+      "loss": 7.8832,
+      "step": 47600
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.263734166407168e-05,
+      "loss": 8.4479,
+      "step": 47650
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.258711615152033e-05,
+      "loss": 7.7838,
+      "step": 47700
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.2536890638968975e-05,
+      "loss": 8.3843,
+      "step": 47750
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.2486665126417625e-05,
+      "loss": 6.9055,
+      "step": 47800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.243643961386626e-05,
+      "loss": 6.6339,
+      "step": 47850
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.23862141013149e-05,
+      "loss": 7.0316,
+      "step": 47900
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.233598858876355e-05,
+      "loss": 7.4569,
+      "step": 47950
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.228576307621219e-05,
+      "loss": 7.6204,
+      "step": 48000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.223553756366084e-05,
+      "loss": 7.1085,
+      "step": 48050
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.218531205110948e-05,
+      "loss": 7.7254,
+      "step": 48100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.2135086538558134e-05,
+      "loss": 7.1486,
+      "step": 48150
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.208486102600677e-05,
+      "loss": 6.9297,
+      "step": 48200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.203463551345542e-05,
+      "loss": 7.5314,
+      "step": 48250
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.198441000090406e-05,
+      "loss": 7.68,
+      "step": 48300
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.193418448835271e-05,
+      "loss": 7.9467,
+      "step": 48350
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.188395897580135e-05,
+      "loss": 6.7188,
+      "step": 48400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.183373346325e-05,
+      "loss": 7.7619,
+      "step": 48450
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.178350795069864e-05,
+      "loss": 7.6537,
+      "step": 48500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.173328243814728e-05,
+      "loss": 6.9593,
+      "step": 48550
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.168305692559593e-05,
+      "loss": 7.4834,
+      "step": 48600
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.163283141304457e-05,
+      "loss": 8.2864,
+      "step": 48650
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.158260590049322e-05,
+      "loss": 7.234,
+      "step": 48700
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.1532380387941856e-05,
+      "loss": 7.2513,
+      "step": 48750
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.1482154875390506e-05,
+      "loss": 7.508,
+      "step": 48800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.143192936283915e-05,
+      "loss": 7.1513,
+      "step": 48850
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.13817038502878e-05,
+      "loss": 7.8882,
+      "step": 48900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.133147833773644e-05,
+      "loss": 8.0859,
+      "step": 48950
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.128125282518509e-05,
+      "loss": 7.5506,
+      "step": 49000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.123102731263373e-05,
+      "loss": 7.9777,
+      "step": 49050
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.118080180008237e-05,
+      "loss": 8.3599,
+      "step": 49100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.1130576287531015e-05,
+      "loss": 7.105,
+      "step": 49150
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.108035077497966e-05,
+      "loss": 8.2692,
+      "step": 49200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.103012526242831e-05,
+      "loss": 7.9098,
+      "step": 49250
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.097989974987695e-05,
+      "loss": 7.1698,
+      "step": 49300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.09296742373256e-05,
+      "loss": 7.406,
+      "step": 49350
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.0879448724774235e-05,
+      "loss": 8.3276,
+      "step": 49400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.0829223212222886e-05,
+      "loss": 7.5714,
+      "step": 49450
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.077899769967153e-05,
+      "loss": 7.0839,
+      "step": 49500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.0728772187120174e-05,
+      "loss": 7.0589,
+      "step": 49550
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.067854667456882e-05,
+      "loss": 7.4998,
+      "step": 49600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0628321162017455e-05,
+      "loss": 7.3495,
+      "step": 49650
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0578095649466106e-05,
+      "loss": 7.5101,
+      "step": 49700
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0527870136914743e-05,
+      "loss": 6.7707,
+      "step": 49750
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0477644624363394e-05,
+      "loss": 7.5822,
+      "step": 49800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.042741911181204e-05,
+      "loss": 6.5937,
+      "step": 49850
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.037719359926069e-05,
+      "loss": 7.2497,
+      "step": 49900
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.0326968086709326e-05,
+      "loss": 7.35,
+      "step": 49950
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.027674257415798e-05,
+      "loss": 7.767,
+      "step": 50000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.0226517061606615e-05,
+      "loss": 8.3228,
+      "step": 50050
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.0176291549055265e-05,
+      "loss": 7.6905,
+      "step": 50100
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.01260660365039e-05,
+      "loss": 7.8275,
+      "step": 50150
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.007584052395255e-05,
+      "loss": 8.0724,
+      "step": 50200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.00256150114012e-05,
+      "loss": 7.0501,
+      "step": 50250
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.997538949884984e-05,
+      "loss": 7.4269,
+      "step": 50300
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.9925163986298486e-05,
+      "loss": 7.5186,
+      "step": 50350
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.987493847374713e-05,
+      "loss": 8.2606,
+      "step": 50400
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.9824712961195774e-05,
+      "loss": 8.2097,
+      "step": 50450
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.977448744864441e-05,
+      "loss": 7.468,
+      "step": 50500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.9724261936093055e-05,
+      "loss": 8.2075,
+      "step": 50550
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.9674036423541706e-05,
+      "loss": 7.3928,
+      "step": 50600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.962381091099035e-05,
+      "loss": 7.2907,
+      "step": 50650
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.9573585398438994e-05,
+      "loss": 7.706,
+      "step": 50700
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.952335988588764e-05,
+      "loss": 7.301,
+      "step": 50750
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.947313437333628e-05,
+      "loss": 6.9109,
+      "step": 50800
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.9422908860784926e-05,
+      "loss": 6.6967,
+      "step": 50850
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.937268334823357e-05,
+      "loss": 5.9484,
+      "step": 50900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9322457835682214e-05,
+      "loss": 7.8288,
+      "step": 50950
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9272232323130865e-05,
+      "loss": 7.3987,
+      "step": 51000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.92220068105795e-05,
+      "loss": 7.3714,
+      "step": 51050
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9171781298028147e-05,
+      "loss": 7.258,
+      "step": 51100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.912155578547679e-05,
+      "loss": 6.8541,
+      "step": 51150
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.9071330272925435e-05,
+      "loss": 7.085,
+      "step": 51200
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.902110476037408e-05,
+      "loss": 6.7827,
+      "step": 51250
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.897087924782273e-05,
+      "loss": 6.6806,
+      "step": 51300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8920653735271374e-05,
+      "loss": 7.2918,
+      "step": 51350
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.887042822272002e-05,
+      "loss": 7.9022,
+      "step": 51400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.882020271016866e-05,
+      "loss": 7.6094,
+      "step": 51450
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8769977197617306e-05,
+      "loss": 8.1048,
+      "step": 51500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.871975168506595e-05,
+      "loss": 6.9056,
+      "step": 51550
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.866952617251459e-05,
+      "loss": 6.4347,
+      "step": 51600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.861930065996324e-05,
+      "loss": 7.307,
+      "step": 51650
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.856907514741188e-05,
+      "loss": 7.649,
+      "step": 51700
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.8518849634860526e-05,
+      "loss": 6.7706,
+      "step": 51750
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.846862412230917e-05,
+      "loss": 6.7943,
+      "step": 51800
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.8418398609757814e-05,
+      "loss": 7.654,
+      "step": 51850
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.836817309720646e-05,
+      "loss": 7.6245,
+      "step": 51900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.83179475846551e-05,
+      "loss": 7.8284,
+      "step": 51950
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.8267722072103746e-05,
+      "loss": 6.9516,
+      "step": 52000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.82174965595524e-05,
+      "loss": 7.1367,
+      "step": 52050
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.816727104700104e-05,
+      "loss": 7.4153,
+      "step": 52100
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.8117045534449685e-05,
+      "loss": 6.5358,
+      "step": 52150
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.806682002189832e-05,
+      "loss": 7.5443,
+      "step": 52200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.8016594509346967e-05,
+      "loss": 7.8184,
+      "step": 52250
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.796636899679561e-05,
+      "loss": 6.9702,
+      "step": 52300
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.791614348424426e-05,
+      "loss": 8.3988,
+      "step": 52350
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.7865917971692905e-05,
+      "loss": 8.1025,
+      "step": 52400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.781569245914155e-05,
+      "loss": 7.81,
+      "step": 52450
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.7765466946590194e-05,
+      "loss": 6.6809,
+      "step": 52500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.771524143403884e-05,
+      "loss": 6.81,
+      "step": 52550
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.766501592148748e-05,
+      "loss": 7.1717,
+      "step": 52600
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.7614790408936126e-05,
+      "loss": 7.3114,
+      "step": 52650
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.756456489638477e-05,
+      "loss": 7.2819,
+      "step": 52700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.7514339383833414e-05,
+      "loss": 6.6964,
+      "step": 52750
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.746411387128206e-05,
+      "loss": 8.1118,
+      "step": 52800
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.74138883587307e-05,
+      "loss": 8.1901,
+      "step": 52850
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.7363662846179346e-05,
+      "loss": 6.8883,
+      "step": 52900
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.731343733362799e-05,
+      "loss": 7.2554,
+      "step": 52950
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.7263211821076634e-05,
+      "loss": 7.402,
+      "step": 53000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.721298630852528e-05,
+      "loss": 8.8808,
+      "step": 53050
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.716276079597393e-05,
+      "loss": 7.1652,
+      "step": 53100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.711253528342257e-05,
+      "loss": 6.884,
+      "step": 53150
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.706230977087122e-05,
+      "loss": 7.4472,
+      "step": 53200
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.701208425831986e-05,
+      "loss": 6.8787,
+      "step": 53250
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.6961858745768505e-05,
+      "loss": 6.9316,
+      "step": 53300
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.691163323321714e-05,
+      "loss": 7.1614,
+      "step": 53350
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.6861407720665787e-05,
+      "loss": 7.193,
+      "step": 53400
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.681118220811444e-05,
+      "loss": 7.5875,
+      "step": 53450
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.676095669556308e-05,
+      "loss": 7.0836,
+      "step": 53500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.6710731183011725e-05,
+      "loss": 7.2054,
+      "step": 53550
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.666050567046037e-05,
+      "loss": 6.95,
+      "step": 53600
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.6610280157909014e-05,
+      "loss": 6.6366,
+      "step": 53650
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.656005464535766e-05,
+      "loss": 6.7976,
+      "step": 53700
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.65098291328063e-05,
+      "loss": 7.1371,
+      "step": 53750
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.6459603620254946e-05,
+      "loss": 6.7457,
+      "step": 53800
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.6409378107703597e-05,
+      "loss": 6.6139,
+      "step": 53850
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.6359152595152234e-05,
+      "loss": 7.9291,
+      "step": 53900
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.630892708260088e-05,
+      "loss": 7.4235,
+      "step": 53950
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.625870157004952e-05,
+      "loss": 6.8609,
+      "step": 54000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.6208476057498166e-05,
+      "loss": 6.6128,
+      "step": 54050
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.615825054494681e-05,
+      "loss": 7.0313,
+      "step": 54100
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.610802503239546e-05,
+      "loss": 7.3654,
+      "step": 54150
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.6057799519844105e-05,
+      "loss": 8.266,
+      "step": 54200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.600757400729275e-05,
+      "loss": 7.9471,
+      "step": 54250
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.595734849474139e-05,
+      "loss": 6.0877,
+      "step": 54300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.590712298219004e-05,
+      "loss": 6.7453,
+      "step": 54350
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.585689746963868e-05,
+      "loss": 5.8985,
+      "step": 54400
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.580667195708732e-05,
+      "loss": 7.4527,
+      "step": 54450
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.575644644453597e-05,
+      "loss": 7.0419,
+      "step": 54500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.570622093198461e-05,
+      "loss": 6.281,
+      "step": 54550
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.565599541943326e-05,
+      "loss": 6.6096,
+      "step": 54600
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.56057699068819e-05,
+      "loss": 7.0341,
+      "step": 54650
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.5555544394330545e-05,
+      "loss": 6.621,
+      "step": 54700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.550531888177919e-05,
+      "loss": 7.4405,
+      "step": 54750
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.5455093369227834e-05,
+      "loss": 7.2506,
+      "step": 54800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.540486785667648e-05,
+      "loss": 7.1534,
+      "step": 54850
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.535464234412513e-05,
+      "loss": 6.5829,
+      "step": 54900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.530441683157377e-05,
+      "loss": 7.0338,
+      "step": 54950
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.5254191319022417e-05,
+      "loss": 6.6234,
+      "step": 55000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.5203965806471054e-05,
+      "loss": 6.2412,
+      "step": 55050
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.51537402939197e-05,
+      "loss": 6.3439,
+      "step": 55100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.510351478136834e-05,
+      "loss": 6.8272,
+      "step": 55150
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.5053289268816986e-05,
+      "loss": 6.4758,
+      "step": 55200
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.500306375626564e-05,
+      "loss": 6.434,
+      "step": 55250
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.495283824371428e-05,
+      "loss": 6.5471,
+      "step": 55300
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.4902612731162925e-05,
+      "loss": 6.5088,
+      "step": 55350
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.485238721861157e-05,
+      "loss": 6.6941,
+      "step": 55400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.480216170606021e-05,
+      "loss": 6.3248,
+      "step": 55450
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.475193619350886e-05,
+      "loss": 7.2989,
+      "step": 55500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.47017106809575e-05,
+      "loss": 7.0947,
+      "step": 55550
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.4651485168406145e-05,
+      "loss": 6.4896,
+      "step": 55600
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.460125965585479e-05,
+      "loss": 5.9249,
+      "step": 55650
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.455103414330343e-05,
+      "loss": 6.7801,
+      "step": 55700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.450080863075208e-05,
+      "loss": 6.2216,
+      "step": 55750
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.445058311820072e-05,
+      "loss": 6.8346,
+      "step": 55800
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.4400357605649366e-05,
+      "loss": 7.2863,
+      "step": 55850
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.435013209309801e-05,
+      "loss": 7.406,
+      "step": 55900
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.429990658054666e-05,
+      "loss": 6.0165,
+      "step": 55950
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.4249681067995304e-05,
+      "loss": 6.8052,
+      "step": 56000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.419945555544395e-05,
+      "loss": 6.679,
+      "step": 56050
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.414923004289259e-05,
+      "loss": 6.2087,
+      "step": 56100
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.4099004530341237e-05,
+      "loss": 6.5904,
+      "step": 56150
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.4048779017789874e-05,
+      "loss": 6.4147,
+      "step": 56200
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.399855350523852e-05,
+      "loss": 6.6465,
+      "step": 56250
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.394832799268717e-05,
+      "loss": 7.11,
+      "step": 56300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.389810248013581e-05,
+      "loss": 7.0558,
+      "step": 56350
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.384787696758446e-05,
+      "loss": 6.922,
+      "step": 56400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.37976514550331e-05,
+      "loss": 7.2125,
+      "step": 56450
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.3747425942481745e-05,
+      "loss": 6.4719,
+      "step": 56500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.369720042993039e-05,
+      "loss": 7.204,
+      "step": 56550
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.364697491737903e-05,
+      "loss": 7.2371,
+      "step": 56600
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.359674940482768e-05,
+      "loss": 6.6131,
+      "step": 56650
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.354652389227633e-05,
+      "loss": 6.6349,
+      "step": 56700
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3496298379724965e-05,
+      "loss": 5.9137,
+      "step": 56750
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.344607286717361e-05,
+      "loss": 6.5402,
+      "step": 56800
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.339584735462225e-05,
+      "loss": 7.3351,
+      "step": 56850
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.33456218420709e-05,
+      "loss": 8.1387,
+      "step": 56900
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.329539632951954e-05,
+      "loss": 7.0783,
+      "step": 56950
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.324517081696819e-05,
+      "loss": 6.947,
+      "step": 57000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3194945304416836e-05,
+      "loss": 6.1526,
+      "step": 57050
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.314471979186548e-05,
+      "loss": 7.273,
+      "step": 57100
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.3094494279314124e-05,
+      "loss": 7.0958,
+      "step": 57150
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.304426876676277e-05,
+      "loss": 6.4413,
+      "step": 57200
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.299404325421141e-05,
+      "loss": 6.597,
+      "step": 57250
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.294381774166006e-05,
+      "loss": 6.6893,
+      "step": 57300
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.28935922291087e-05,
+      "loss": 6.4746,
+      "step": 57350
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2843366716557345e-05,
+      "loss": 7.376,
+      "step": 57400
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.279314120400599e-05,
+      "loss": 7.2823,
+      "step": 57450
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.274291569145463e-05,
+      "loss": 6.3184,
+      "step": 57500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.269269017890328e-05,
+      "loss": 6.4526,
+      "step": 57550
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.264246466635192e-05,
+      "loss": 6.7892,
+      "step": 57600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.2592239153800565e-05,
+      "loss": 6.2082,
+      "step": 57650
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.254201364124921e-05,
+      "loss": 7.1488,
+      "step": 57700
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.249178812869786e-05,
+      "loss": 6.9399,
+      "step": 57750
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.2441562616146504e-05,
+      "loss": 6.8596,
+      "step": 57800
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.239133710359515e-05,
+      "loss": 6.8899,
+      "step": 57850
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.2341111591043785e-05,
+      "loss": 6.8196,
+      "step": 57900
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.229088607849243e-05,
+      "loss": 7.5114,
+      "step": 57950
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.224066056594107e-05,
+      "loss": 6.4122,
+      "step": 58000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.219043505338972e-05,
+      "loss": 6.9228,
+      "step": 58050
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.214020954083837e-05,
+      "loss": 6.3687,
+      "step": 58100
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.208998402828701e-05,
+      "loss": 6.8616,
+      "step": 58150
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.2039758515735656e-05,
+      "loss": 6.002,
+      "step": 58200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.19895330031843e-05,
+      "loss": 6.0985,
+      "step": 58250
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.1939307490632944e-05,
+      "loss": 6.5857,
+      "step": 58300
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.188908197808159e-05,
+      "loss": 6.257,
+      "step": 58350
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.183885646553023e-05,
+      "loss": 6.9222,
+      "step": 58400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.178863095297888e-05,
+      "loss": 6.7801,
+      "step": 58450
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.173840544042752e-05,
+      "loss": 6.3861,
+      "step": 58500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.1688179927876165e-05,
+      "loss": 6.8685,
+      "step": 58550
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.163795441532481e-05,
+      "loss": 6.9948,
+      "step": 58600
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.158772890277345e-05,
+      "loss": 6.0965,
+      "step": 58650
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.15375033902221e-05,
+      "loss": 7.282,
+      "step": 58700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.148727787767074e-05,
+      "loss": 7.6165,
+      "step": 58750
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.143705236511939e-05,
+      "loss": 6.734,
+      "step": 58800
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.1386826852568036e-05,
+      "loss": 6.0334,
+      "step": 58850
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.133660134001668e-05,
+      "loss": 6.5306,
+      "step": 58900
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.1286375827465324e-05,
+      "loss": 7.4324,
+      "step": 58950
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.123615031491397e-05,
+      "loss": 7.234,
+      "step": 59000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.1185924802362605e-05,
+      "loss": 6.7196,
+      "step": 59050
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.113569928981125e-05,
+      "loss": 6.0641,
+      "step": 59100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.10854737772599e-05,
+      "loss": 5.9373,
+      "step": 59150
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.1035248264708544e-05,
+      "loss": 6.4428,
+      "step": 59200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.098502275215719e-05,
+      "loss": 6.7303,
+      "step": 59250
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.093479723960583e-05,
+      "loss": 6.5585,
+      "step": 59300
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.0884571727054476e-05,
+      "loss": 5.837,
+      "step": 59350
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.083434621450312e-05,
+      "loss": 6.574,
+      "step": 59400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.0784120701951764e-05,
+      "loss": 7.4319,
+      "step": 59450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.073389518940041e-05,
+      "loss": 6.1092,
+      "step": 59500
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.068366967684906e-05,
+      "loss": 6.4683,
+      "step": 59550
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.06334441642977e-05,
+      "loss": 7.1323,
+      "step": 59600
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.058321865174634e-05,
+      "loss": 6.7011,
+      "step": 59650
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.0532993139194985e-05,
+      "loss": 6.7281,
+      "step": 59700
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.048276762664363e-05,
+      "loss": 6.0361,
+      "step": 59750
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.043254211409227e-05,
+      "loss": 6.5359,
+      "step": 59800
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.0382316601540924e-05,
+      "loss": 7.3576,
+      "step": 59850
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.033209108898957e-05,
+      "loss": 6.7324,
+      "step": 59900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.028186557643821e-05,
+      "loss": 7.1445,
+      "step": 59950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.0231640063886856e-05,
+      "loss": 5.426,
+      "step": 60000
+    },
+    {
+      "epoch": 4.43,
+      "eval_loss": 7.799332618713379,
+      "eval_runtime": 963.4222,
+      "eval_samples_per_second": 13.595,
+      "eval_steps_per_second": 3.399,
+      "eval_wer": 0.20775061946159337,
+      "step": 60000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.01814145513355e-05,
+      "loss": 6.577,
+      "step": 60050
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.0131189038784144e-05,
+      "loss": 7.1169,
+      "step": 60100
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.008096352623279e-05,
+      "loss": 7.2535,
+      "step": 60150
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.003073801368143e-05,
+      "loss": 6.2427,
+      "step": 60200
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.9980512501130076e-05,
+      "loss": 6.3184,
+      "step": 60250
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.993028698857872e-05,
+      "loss": 6.9348,
+      "step": 60300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.9880061476027364e-05,
+      "loss": 6.5074,
+      "step": 60350
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.982983596347601e-05,
+      "loss": 7.2028,
+      "step": 60400
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.977961045092465e-05,
+      "loss": 7.1185,
+      "step": 60450
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.9729384938373296e-05,
+      "loss": 6.5089,
+      "step": 60500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.967915942582194e-05,
+      "loss": 5.8853,
+      "step": 60550
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.962893391327059e-05,
+      "loss": 6.0402,
+      "step": 60600
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.9578708400719235e-05,
+      "loss": 6.6078,
+      "step": 60650
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.952848288816788e-05,
+      "loss": 6.4986,
+      "step": 60700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.947825737561652e-05,
+      "loss": 6.9922,
+      "step": 60750
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.942803186306516e-05,
+      "loss": 6.8327,
+      "step": 60800
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.9377806350513805e-05,
+      "loss": 6.7131,
+      "step": 60850
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.932758083796245e-05,
+      "loss": 6.3372,
+      "step": 60900
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.92773553254111e-05,
+      "loss": 6.5337,
+      "step": 60950
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.9227129812859744e-05,
+      "loss": 7.5148,
+      "step": 61000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.917690430030839e-05,
+      "loss": 6.2416,
+      "step": 61050
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.912667878775703e-05,
+      "loss": 6.3845,
+      "step": 61100
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.9076453275205676e-05,
+      "loss": 7.0156,
+      "step": 61150
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.902622776265432e-05,
+      "loss": 6.4239,
+      "step": 61200
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.8976002250102964e-05,
+      "loss": 7.2111,
+      "step": 61250
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.892577673755161e-05,
+      "loss": 6.5958,
+      "step": 61300
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.887555122500025e-05,
+      "loss": 7.3125,
+      "step": 61350
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.8825325712448896e-05,
+      "loss": 6.0597,
+      "step": 61400
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.877510019989754e-05,
+      "loss": 6.3881,
+      "step": 61450
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.8724874687346184e-05,
+      "loss": 6.2309,
+      "step": 61500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.867464917479483e-05,
+      "loss": 6.7603,
+      "step": 61550
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.862442366224347e-05,
+      "loss": 7.386,
+      "step": 61600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.857419814969212e-05,
+      "loss": 7.1897,
+      "step": 61650
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.852397263714077e-05,
+      "loss": 6.8813,
+      "step": 61700
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.847374712458941e-05,
+      "loss": 7.9826,
+      "step": 61750
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.8423521612038055e-05,
+      "loss": 6.4649,
+      "step": 61800
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.83732960994867e-05,
+      "loss": 6.0005,
+      "step": 61850
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.832307058693534e-05,
+      "loss": 6.7249,
+      "step": 61900
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.827284507438398e-05,
+      "loss": 7.3703,
+      "step": 61950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.822261956183263e-05,
+      "loss": 6.3157,
+      "step": 62000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.8172394049281276e-05,
+      "loss": 6.3982,
+      "step": 62050
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.812216853672992e-05,
+      "loss": 6.3995,
+      "step": 62100
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.8071943024178564e-05,
+      "loss": 6.9792,
+      "step": 62150
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.802171751162721e-05,
+      "loss": 7.7098,
+      "step": 62200
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.797149199907585e-05,
+      "loss": 7.3092,
+      "step": 62250
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.7921266486524496e-05,
+      "loss": 6.7666,
+      "step": 62300
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.787104097397314e-05,
+      "loss": 6.1829,
+      "step": 62350
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.782081546142179e-05,
+      "loss": 8.2604,
+      "step": 62400
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.777058994887043e-05,
+      "loss": 6.7275,
+      "step": 62450
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.772036443631907e-05,
+      "loss": 6.8682,
+      "step": 62500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7670138923767716e-05,
+      "loss": 7.4531,
+      "step": 62550
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.761991341121636e-05,
+      "loss": 7.4792,
+      "step": 62600
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7569687898665004e-05,
+      "loss": 6.3364,
+      "step": 62650
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7519462386113655e-05,
+      "loss": 6.395,
+      "step": 62700
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.74692368735623e-05,
+      "loss": 6.4644,
+      "step": 62750
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.741901136101094e-05,
+      "loss": 7.6636,
+      "step": 62800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.736878584845959e-05,
+      "loss": 6.5346,
+      "step": 62850
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.731856033590823e-05,
+      "loss": 7.7544,
+      "step": 62900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.7268334823356875e-05,
+      "loss": 7.1518,
+      "step": 62950
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.721810931080552e-05,
+      "loss": 5.2845,
+      "step": 63000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7167883798254163e-05,
+      "loss": 6.4635,
+      "step": 63050
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.711765828570281e-05,
+      "loss": 6.7313,
+      "step": 63100
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.706743277315145e-05,
+      "loss": 6.2767,
+      "step": 63150
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.7017207260600096e-05,
+      "loss": 6.2349,
+      "step": 63200
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.696698174804874e-05,
+      "loss": 7.2152,
+      "step": 63250
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.6916756235497384e-05,
+      "loss": 6.4904,
+      "step": 63300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.686653072294603e-05,
+      "loss": 6.4779,
+      "step": 63350
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.681630521039467e-05,
+      "loss": 7.0359,
+      "step": 63400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.676607969784332e-05,
+      "loss": 6.3846,
+      "step": 63450
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.671585418529197e-05,
+      "loss": 6.3923,
+      "step": 63500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.666562867274061e-05,
+      "loss": 6.7271,
+      "step": 63550
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.661540316018925e-05,
+      "loss": 6.6546,
+      "step": 63600
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.656517764763789e-05,
+      "loss": 7.1482,
+      "step": 63650
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.6514952135086536e-05,
+      "loss": 6.4931,
+      "step": 63700
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.646472662253518e-05,
+      "loss": 5.1346,
+      "step": 63750
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.641450110998383e-05,
+      "loss": 6.25,
+      "step": 63800
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.6364275597432475e-05,
+      "loss": 6.5812,
+      "step": 63850
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.631405008488112e-05,
+      "loss": 6.4797,
+      "step": 63900
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.626382457232976e-05,
+      "loss": 6.2886,
+      "step": 63950
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.621359905977841e-05,
+      "loss": 6.4452,
+      "step": 64000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.616337354722705e-05,
+      "loss": 6.5534,
+      "step": 64050
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.6113148034675695e-05,
+      "loss": 6.9353,
+      "step": 64100
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.606292252212434e-05,
+      "loss": 6.1219,
+      "step": 64150
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6012697009572984e-05,
+      "loss": 6.8233,
+      "step": 64200
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.596247149702163e-05,
+      "loss": 7.1924,
+      "step": 64250
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.591224598447027e-05,
+      "loss": 7.1116,
+      "step": 64300
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.5862020471918916e-05,
+      "loss": 7.2818,
+      "step": 64350
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.581179495936756e-05,
+      "loss": 6.3182,
+      "step": 64400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.5761569446816204e-05,
+      "loss": 6.7712,
+      "step": 64450
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.5711343934264855e-05,
+      "loss": 6.7902,
+      "step": 64500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.56611184217135e-05,
+      "loss": 6.5504,
+      "step": 64550
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.561089290916214e-05,
+      "loss": 6.3599,
+      "step": 64600
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.556066739661079e-05,
+      "loss": 6.4758,
+      "step": 64650
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.551044188405943e-05,
+      "loss": 7.2899,
+      "step": 64700
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.546021637150807e-05,
+      "loss": 6.6164,
+      "step": 64750
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.540999085895671e-05,
+      "loss": 6.0466,
+      "step": 64800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.535976534640536e-05,
+      "loss": 6.2209,
+      "step": 64850
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.530953983385401e-05,
+      "loss": 6.6098,
+      "step": 64900
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.525931432130265e-05,
+      "loss": 5.959,
+      "step": 64950
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.5209088808751295e-05,
+      "loss": 6.6942,
+      "step": 65000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.515886329619994e-05,
+      "loss": 7.7404,
+      "step": 65050
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.510863778364858e-05,
+      "loss": 6.5342,
+      "step": 65100
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.505841227109723e-05,
+      "loss": 6.1434,
+      "step": 65150
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.500818675854587e-05,
+      "loss": 5.8523,
+      "step": 65200
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.495796124599452e-05,
+      "loss": 5.7111,
+      "step": 65250
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.490773573344316e-05,
+      "loss": 6.4092,
+      "step": 65300
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.4857510220891804e-05,
+      "loss": 6.5057,
+      "step": 65350
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.480728470834045e-05,
+      "loss": 5.9863,
+      "step": 65400
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.475705919578909e-05,
+      "loss": 6.0978,
+      "step": 65450
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.4706833683237736e-05,
+      "loss": 6.6325,
+      "step": 65500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.4656608170686387e-05,
+      "loss": 5.9716,
+      "step": 65550
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.460638265813503e-05,
+      "loss": 5.9812,
+      "step": 65600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.4556157145583675e-05,
+      "loss": 5.6784,
+      "step": 65650
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.450593163303232e-05,
+      "loss": 6.7303,
+      "step": 65700
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.445570612048096e-05,
+      "loss": 7.527,
+      "step": 65750
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.440548060792961e-05,
+      "loss": 6.3649,
+      "step": 65800
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.435525509537825e-05,
+      "loss": 6.6456,
+      "step": 65850
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.4305029582826895e-05,
+      "loss": 6.2518,
+      "step": 65900
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.425480407027554e-05,
+      "loss": 6.0104,
+      "step": 65950
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.420457855772418e-05,
+      "loss": 7.0261,
+      "step": 66000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.415435304517283e-05,
+      "loss": 6.4373,
+      "step": 66050
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.410412753262147e-05,
+      "loss": 6.3931,
+      "step": 66100
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.4053902020070115e-05,
+      "loss": 7.0678,
+      "step": 66150
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.400367650751876e-05,
+      "loss": 6.9086,
+      "step": 66200
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.39534509949674e-05,
+      "loss": 6.3431,
+      "step": 66250
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3903225482416054e-05,
+      "loss": 7.298,
+      "step": 66300
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.38529999698647e-05,
+      "loss": 6.4188,
+      "step": 66350
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.380277445731334e-05,
+      "loss": 6.1998,
+      "step": 66400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.375254894476198e-05,
+      "loss": 6.7306,
+      "step": 66450
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.3702323432210624e-05,
+      "loss": 6.0333,
+      "step": 66500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.365209791965927e-05,
+      "loss": 6.9375,
+      "step": 66550
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.360187240710791e-05,
+      "loss": 6.4168,
+      "step": 66600
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.355164689455656e-05,
+      "loss": 6.6486,
+      "step": 66650
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3501421382005207e-05,
+      "loss": 5.6693,
+      "step": 66700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.345119586945385e-05,
+      "loss": 6.7548,
+      "step": 66750
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.3400970356902495e-05,
+      "loss": 8.7782,
+      "step": 66800
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.335074484435114e-05,
+      "loss": 6.9741,
+      "step": 66850
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.330051933179978e-05,
+      "loss": 6.423,
+      "step": 66900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.325029381924843e-05,
+      "loss": 6.0077,
+      "step": 66950
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.320006830669707e-05,
+      "loss": 6.4085,
+      "step": 67000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3149842794145715e-05,
+      "loss": 6.383,
+      "step": 67050
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.309961728159436e-05,
+      "loss": 5.8758,
+      "step": 67100
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3049391769043e-05,
+      "loss": 7.1169,
+      "step": 67150
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.299916625649165e-05,
+      "loss": 7.5655,
+      "step": 67200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.294894074394029e-05,
+      "loss": 6.1727,
+      "step": 67250
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.2898715231388935e-05,
+      "loss": 6.0568,
+      "step": 67300
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.2848489718837586e-05,
+      "loss": 6.9697,
+      "step": 67350
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.279826420628623e-05,
+      "loss": 6.7938,
+      "step": 67400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.2748038693734874e-05,
+      "loss": 6.3006,
+      "step": 67450
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.269781318118352e-05,
+      "loss": 6.3299,
+      "step": 67500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.264758766863216e-05,
+      "loss": 6.3615,
+      "step": 67550
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.25973621560808e-05,
+      "loss": 6.1106,
+      "step": 67600
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.2547136643529444e-05,
+      "loss": 5.571,
+      "step": 67650
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.2496911130978094e-05,
+      "loss": 6.5922,
+      "step": 67700
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.244668561842674e-05,
+      "loss": 5.663,
+      "step": 67750
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.239646010587538e-05,
+      "loss": 7.3669,
+      "step": 67800
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.2346234593324027e-05,
+      "loss": 6.015,
+      "step": 67850
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.229600908077267e-05,
+      "loss": 5.8678,
+      "step": 67900
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.2245783568221315e-05,
+      "loss": 5.5537,
+      "step": 67950
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.219555805566996e-05,
+      "loss": 6.175,
+      "step": 68000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.21453325431186e-05,
+      "loss": 5.9018,
+      "step": 68050
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.2095107030567254e-05,
+      "loss": 6.9064,
+      "step": 68100
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.204488151801589e-05,
+      "loss": 6.8775,
+      "step": 68150
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.1994656005464535e-05,
+      "loss": 5.6397,
+      "step": 68200
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.194443049291318e-05,
+      "loss": 5.815,
+      "step": 68250
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.189420498036182e-05,
+      "loss": 6.0795,
+      "step": 68300
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.184397946781047e-05,
+      "loss": 6.8721,
+      "step": 68350
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.179375395525912e-05,
+      "loss": 6.4936,
+      "step": 68400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.174352844270776e-05,
+      "loss": 5.7195,
+      "step": 68450
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.1693302930156406e-05,
+      "loss": 5.6897,
+      "step": 68500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.164307741760505e-05,
+      "loss": 6.2271,
+      "step": 68550
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.1592851905053694e-05,
+      "loss": 6.3731,
+      "step": 68600
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.154262639250234e-05,
+      "loss": 5.8502,
+      "step": 68650
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.149240087995098e-05,
+      "loss": 5.8768,
+      "step": 68700
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.1442175367399626e-05,
+      "loss": 6.4265,
+      "step": 68750
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.139194985484827e-05,
+      "loss": 5.8952,
+      "step": 68800
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.1341724342296914e-05,
+      "loss": 5.3739,
+      "step": 68850
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.129149882974556e-05,
+      "loss": 6.4317,
+      "step": 68900
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.12412733171942e-05,
+      "loss": 5.6296,
+      "step": 68950
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.1191047804642847e-05,
+      "loss": 6.2448,
+      "step": 69000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.114082229209149e-05,
+      "loss": 6.116,
+      "step": 69050
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.1090596779540135e-05,
+      "loss": 6.4299,
+      "step": 69100
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.1040371266988785e-05,
+      "loss": 6.7337,
+      "step": 69150
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.099014575443743e-05,
+      "loss": 6.6103,
+      "step": 69200
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.0939920241886074e-05,
+      "loss": 5.6155,
+      "step": 69250
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.088969472933471e-05,
+      "loss": 5.7131,
+      "step": 69300
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.0839469216783355e-05,
+      "loss": 5.6799,
+      "step": 69350
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.0789243704232e-05,
+      "loss": 5.9907,
+      "step": 69400
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.073901819168064e-05,
+      "loss": 5.7125,
+      "step": 69450
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.0688792679129294e-05,
+      "loss": 6.3093,
+      "step": 69500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.063856716657794e-05,
+      "loss": 6.1113,
+      "step": 69550
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.058834165402658e-05,
+      "loss": 5.5845,
+      "step": 69600
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.0538116141475226e-05,
+      "loss": 5.9267,
+      "step": 69650
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.048789062892387e-05,
+      "loss": 6.0062,
+      "step": 69700
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.0437665116372514e-05,
+      "loss": 6.005,
+      "step": 69750
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.038743960382116e-05,
+      "loss": 5.854,
+      "step": 69800
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.0337214091269806e-05,
+      "loss": 6.3468,
+      "step": 69850
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.0286988578718446e-05,
+      "loss": 5.9127,
+      "step": 69900
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.023676306616709e-05,
+      "loss": 6.5192,
+      "step": 69950
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.0186537553615734e-05,
+      "loss": 5.8418,
+      "step": 70000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.013631204106438e-05,
+      "loss": 6.0775,
+      "step": 70050
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.0086086528513026e-05,
+      "loss": 5.926,
+      "step": 70100
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.003586101596167e-05,
+      "loss": 5.8467,
+      "step": 70150
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.9985635503410314e-05,
+      "loss": 6.441,
+      "step": 70200
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.9935409990858958e-05,
+      "loss": 5.6337,
+      "step": 70250
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.9885184478307606e-05,
+      "loss": 6.3408,
+      "step": 70300
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.983495896575625e-05,
+      "loss": 6.0077,
+      "step": 70350
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.9784733453204894e-05,
+      "loss": 5.7263,
+      "step": 70400
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.9734507940653534e-05,
+      "loss": 6.8161,
+      "step": 70450
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.968428242810218e-05,
+      "loss": 6.4292,
+      "step": 70500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.9634056915550822e-05,
+      "loss": 6.0751,
+      "step": 70550
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.9583831402999466e-05,
+      "loss": 6.2439,
+      "step": 70600
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.9533605890448114e-05,
+      "loss": 5.6406,
+      "step": 70650
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.9483380377896758e-05,
+      "loss": 5.5623,
+      "step": 70700
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.9433154865345402e-05,
+      "loss": 6.3392,
+      "step": 70750
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.9382929352794046e-05,
+      "loss": 7.3752,
+      "step": 70800
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.9332703840242693e-05,
+      "loss": 6.2126,
+      "step": 70850
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.9282478327691338e-05,
+      "loss": 5.3583,
+      "step": 70900
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.923225281513998e-05,
+      "loss": 5.4659,
+      "step": 70950
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.9182027302588626e-05,
+      "loss": 6.1876,
+      "step": 71000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.9131801790037266e-05,
+      "loss": 5.8878,
+      "step": 71050
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.908157627748591e-05,
+      "loss": 6.2974,
+      "step": 71100
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.9031350764934558e-05,
+      "loss": 6.348,
+      "step": 71150
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.8981125252383202e-05,
+      "loss": 5.9929,
+      "step": 71200
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.8930899739831846e-05,
+      "loss": 5.9609,
+      "step": 71250
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.888067422728049e-05,
+      "loss": 5.4301,
+      "step": 71300
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.8830448714729137e-05,
+      "loss": 5.4559,
+      "step": 71350
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.878022320217778e-05,
+      "loss": 7.1806,
+      "step": 71400
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.8729997689626426e-05,
+      "loss": 5.6962,
+      "step": 71450
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.867977217707507e-05,
+      "loss": 5.2751,
+      "step": 71500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.8629546664523717e-05,
+      "loss": 5.8732,
+      "step": 71550
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.8579321151972354e-05,
+      "loss": 5.3111,
+      "step": 71600
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.8529095639421e-05,
+      "loss": 6.269,
+      "step": 71650
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.8478870126869646e-05,
+      "loss": 4.7494,
+      "step": 71700
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.842864461431829e-05,
+      "loss": 6.2853,
+      "step": 71750
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.8378419101766934e-05,
+      "loss": 6.7802,
+      "step": 71800
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.8328193589215578e-05,
+      "loss": 7.7665,
+      "step": 71850
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.8277968076664225e-05,
+      "loss": 5.7143,
+      "step": 71900
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.822774256411287e-05,
+      "loss": 6.1485,
+      "step": 71950
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.8177517051561514e-05,
+      "loss": 5.9611,
+      "step": 72000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.8127291539010158e-05,
+      "loss": 6.9089,
+      "step": 72050
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.8077066026458805e-05,
+      "loss": 5.3067,
+      "step": 72100
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.8026840513907442e-05,
+      "loss": 5.7255,
+      "step": 72150
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.7976615001356086e-05,
+      "loss": 6.6141,
+      "step": 72200
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.7926389488804734e-05,
+      "loss": 5.1587,
+      "step": 72250
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.7876163976253378e-05,
+      "loss": 6.2063,
+      "step": 72300
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.7825938463702022e-05,
+      "loss": 5.9984,
+      "step": 72350
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.7775712951150666e-05,
+      "loss": 6.2301,
+      "step": 72400
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.7725487438599313e-05,
+      "loss": 6.0619,
+      "step": 72450
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.7675261926047957e-05,
+      "loss": 7.1879,
+      "step": 72500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.76250364134966e-05,
+      "loss": 6.8024,
+      "step": 72550
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.7574810900945246e-05,
+      "loss": 7.412,
+      "step": 72600
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.7524585388393893e-05,
+      "loss": 6.172,
+      "step": 72650
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.7474359875842537e-05,
+      "loss": 5.9536,
+      "step": 72700
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.7424134363291178e-05,
+      "loss": 6.4215,
+      "step": 72750
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.7373908850739822e-05,
+      "loss": 5.6326,
+      "step": 72800
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.7323683338188466e-05,
+      "loss": 5.8943,
+      "step": 72850
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.727345782563711e-05,
+      "loss": 6.8689,
+      "step": 72900
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.7223232313085757e-05,
+      "loss": 6.2079,
+      "step": 72950
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.71730068005344e-05,
+      "loss": 6.4607,
+      "step": 73000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.7122781287983045e-05,
+      "loss": 6.0781,
+      "step": 73050
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.707255577543169e-05,
+      "loss": 5.7624,
+      "step": 73100
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.7022330262880337e-05,
+      "loss": 6.0385,
+      "step": 73150
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.697210475032898e-05,
+      "loss": 5.9751,
+      "step": 73200
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.6921879237777625e-05,
+      "loss": 6.3938,
+      "step": 73250
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.6871653725226266e-05,
+      "loss": 5.9229,
+      "step": 73300
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.682142821267491e-05,
+      "loss": 6.0674,
+      "step": 73350
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.6771202700123554e-05,
+      "loss": 6.7223,
+      "step": 73400
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.6720977187572198e-05,
+      "loss": 5.7889,
+      "step": 73450
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.6670751675020845e-05,
+      "loss": 7.1486,
+      "step": 73500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.662052616246949e-05,
+      "loss": 6.1844,
+      "step": 73550
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.6570300649918133e-05,
+      "loss": 6.198,
+      "step": 73600
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.6520075137366777e-05,
+      "loss": 6.6778,
+      "step": 73650
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.6469849624815425e-05,
+      "loss": 5.9788,
+      "step": 73700
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.641962411226407e-05,
+      "loss": 6.3568,
+      "step": 73750
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.6369398599712713e-05,
+      "loss": 5.9383,
+      "step": 73800
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.6319173087161357e-05,
+      "loss": 6.4832,
+      "step": 73850
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.6268947574609998e-05,
+      "loss": 5.883,
+      "step": 73900
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.6218722062058642e-05,
+      "loss": 6.213,
+      "step": 73950
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.616849654950729e-05,
+      "loss": 6.5404,
+      "step": 74000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.6118271036955933e-05,
+      "loss": 6.1246,
+      "step": 74050
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.6068045524404577e-05,
+      "loss": 6.0739,
+      "step": 74100
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.601782001185322e-05,
+      "loss": 6.2085,
+      "step": 74150
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.596759449930187e-05,
+      "loss": 6.7059,
+      "step": 74200
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.5917368986750513e-05,
+      "loss": 6.231,
+      "step": 74250
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.5867143474199157e-05,
+      "loss": 6.1287,
+      "step": 74300
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.58169179616478e-05,
+      "loss": 6.0583,
+      "step": 74350
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.576669244909645e-05,
+      "loss": 6.1552,
+      "step": 74400
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.5716466936545086e-05,
+      "loss": 6.4191,
+      "step": 74450
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.566624142399373e-05,
+      "loss": 7.2899,
+      "step": 74500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.5616015911442377e-05,
+      "loss": 6.3234,
+      "step": 74550
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.556579039889102e-05,
+      "loss": 5.6938,
+      "step": 74600
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.5515564886339665e-05,
+      "loss": 5.1418,
+      "step": 74650
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.546533937378831e-05,
+      "loss": 5.7747,
+      "step": 74700
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.5415113861236957e-05,
+      "loss": 5.7191,
+      "step": 74750
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.53648883486856e-05,
+      "loss": 6.0435,
+      "step": 74800
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.5314662836134245e-05,
+      "loss": 6.1592,
+      "step": 74850
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.526443732358289e-05,
+      "loss": 6.1625,
+      "step": 74900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.5214211811031536e-05,
+      "loss": 4.7958,
+      "step": 74950
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.516398629848018e-05,
+      "loss": 6.0618,
+      "step": 75000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.5113760785928818e-05,
+      "loss": 5.754,
+      "step": 75050
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.5063535273377465e-05,
+      "loss": 5.6346,
+      "step": 75100
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.501330976082611e-05,
+      "loss": 5.6922,
+      "step": 75150
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.4963084248274753e-05,
+      "loss": 6.6754,
+      "step": 75200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.4912858735723397e-05,
+      "loss": 5.2828,
+      "step": 75250
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.4862633223172045e-05,
+      "loss": 5.799,
+      "step": 75300
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.481240771062069e-05,
+      "loss": 5.8229,
+      "step": 75350
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.4762182198069333e-05,
+      "loss": 5.1759,
+      "step": 75400
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.4711956685517977e-05,
+      "loss": 5.9411,
+      "step": 75450
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.466173117296662e-05,
+      "loss": 5.4522,
+      "step": 75500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.4611505660415265e-05,
+      "loss": 6.0731,
+      "step": 75550
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.456128014786391e-05,
+      "loss": 5.9288,
+      "step": 75600
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.4511054635312557e-05,
+      "loss": 5.7434,
+      "step": 75650
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.44608291227612e-05,
+      "loss": 5.5638,
+      "step": 75700
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.441060361020984e-05,
+      "loss": 6.423,
+      "step": 75750
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.436037809765849e-05,
+      "loss": 5.4612,
+      "step": 75800
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.4310152585107133e-05,
+      "loss": 7.1213,
+      "step": 75850
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.4259927072555777e-05,
+      "loss": 6.366,
+      "step": 75900
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.420970156000442e-05,
+      "loss": 5.8278,
+      "step": 75950
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.4159476047453065e-05,
+      "loss": 6.1465,
+      "step": 76000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.410925053490171e-05,
+      "loss": 5.7868,
+      "step": 76050
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.4059025022350353e-05,
+      "loss": 5.8116,
+      "step": 76100
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.4008799509798997e-05,
+      "loss": 5.7459,
+      "step": 76150
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.3958573997247645e-05,
+      "loss": 6.2053,
+      "step": 76200
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.390834848469629e-05,
+      "loss": 5.768,
+      "step": 76250
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.385812297214493e-05,
+      "loss": 5.9021,
+      "step": 76300
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.3807897459593577e-05,
+      "loss": 6.2206,
+      "step": 76350
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.375767194704222e-05,
+      "loss": 6.3088,
+      "step": 76400
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.3707446434490865e-05,
+      "loss": 6.0105,
+      "step": 76450
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.365722092193951e-05,
+      "loss": 5.738,
+      "step": 76500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.3606995409388156e-05,
+      "loss": 6.2838,
+      "step": 76550
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.3556769896836797e-05,
+      "loss": 5.7041,
+      "step": 76600
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.350654438428544e-05,
+      "loss": 6.7796,
+      "step": 76650
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.345631887173409e-05,
+      "loss": 6.4432,
+      "step": 76700
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.3406093359182733e-05,
+      "loss": 5.5805,
+      "step": 76750
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.3355867846631377e-05,
+      "loss": 5.4049,
+      "step": 76800
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.330564233408002e-05,
+      "loss": 5.0643,
+      "step": 76850
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.3255416821528665e-05,
+      "loss": 5.4007,
+      "step": 76900
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.320519130897731e-05,
+      "loss": 5.3506,
+      "step": 76950
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.3154965796425953e-05,
+      "loss": 6.5889,
+      "step": 77000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.31047402838746e-05,
+      "loss": 6.206,
+      "step": 77050
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.3054514771323244e-05,
+      "loss": 6.03,
+      "step": 77100
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.3004289258771885e-05,
+      "loss": 5.6658,
+      "step": 77150
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.295406374622053e-05,
+      "loss": 6.5901,
+      "step": 77200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.2903838233669176e-05,
+      "loss": 6.3019,
+      "step": 77250
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.285361272111782e-05,
+      "loss": 5.5744,
+      "step": 77300
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.2803387208566465e-05,
+      "loss": 5.8269,
+      "step": 77350
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.275316169601511e-05,
+      "loss": 6.1005,
+      "step": 77400
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.2702936183463753e-05,
+      "loss": 6.0196,
+      "step": 77450
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.2652710670912397e-05,
+      "loss": 5.8475,
+      "step": 77500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.260248515836104e-05,
+      "loss": 6.4338,
+      "step": 77550
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.2552259645809688e-05,
+      "loss": 4.958,
+      "step": 77600
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.2502034133258332e-05,
+      "loss": 6.4737,
+      "step": 77650
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.2451808620706976e-05,
+      "loss": 6.3223,
+      "step": 77700
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.240158310815562e-05,
+      "loss": 7.171,
+      "step": 77750
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.2351357595604264e-05,
+      "loss": 6.5725,
+      "step": 77800
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.230113208305291e-05,
+      "loss": 5.7644,
+      "step": 77850
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.2250906570501553e-05,
+      "loss": 5.6257,
+      "step": 77900
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.22006810579502e-05,
+      "loss": 6.2325,
+      "step": 77950
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.215045554539884e-05,
+      "loss": 6.7106,
+      "step": 78000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.2100230032847485e-05,
+      "loss": 5.0113,
+      "step": 78050
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.205000452029613e-05,
+      "loss": 6.1309,
+      "step": 78100
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1999779007744776e-05,
+      "loss": 5.5098,
+      "step": 78150
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.194955349519342e-05,
+      "loss": 6.6709,
+      "step": 78200
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1899327982642064e-05,
+      "loss": 5.8039,
+      "step": 78250
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.184910247009071e-05,
+      "loss": 5.0271,
+      "step": 78300
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1798876957539352e-05,
+      "loss": 6.1648,
+      "step": 78350
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1748651444987996e-05,
+      "loss": 5.1838,
+      "step": 78400
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.169842593243664e-05,
+      "loss": 7.1149,
+      "step": 78450
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1648200419885288e-05,
+      "loss": 5.9544,
+      "step": 78500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1597974907333932e-05,
+      "loss": 6.4747,
+      "step": 78550
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1547749394782573e-05,
+      "loss": 5.6367,
+      "step": 78600
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.149752388223122e-05,
+      "loss": 5.8395,
+      "step": 78650
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.1447298369679864e-05,
+      "loss": 6.8058,
+      "step": 78700
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.1397072857128508e-05,
+      "loss": 6.4977,
+      "step": 78750
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.1346847344577152e-05,
+      "loss": 7.0943,
+      "step": 78800
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.12966218320258e-05,
+      "loss": 6.0009,
+      "step": 78850
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.124639631947444e-05,
+      "loss": 5.8074,
+      "step": 78900
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.1196170806923084e-05,
+      "loss": 6.277,
+      "step": 78950
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.114594529437173e-05,
+      "loss": 5.61,
+      "step": 79000
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.1095719781820376e-05,
+      "loss": 5.6585,
+      "step": 79050
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.104549426926902e-05,
+      "loss": 4.9836,
+      "step": 79100
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.099526875671766e-05,
+      "loss": 6.1327,
+      "step": 79150
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.0945043244166308e-05,
+      "loss": 6.2281,
+      "step": 79200
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0894817731614952e-05,
+      "loss": 5.9593,
+      "step": 79250
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0844592219063596e-05,
+      "loss": 5.1415,
+      "step": 79300
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.079436670651224e-05,
+      "loss": 5.719,
+      "step": 79350
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0744141193960888e-05,
+      "loss": 6.1617,
+      "step": 79400
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.069391568140953e-05,
+      "loss": 6.3103,
+      "step": 79450
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0643690168858172e-05,
+      "loss": 5.2091,
+      "step": 79500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.059346465630682e-05,
+      "loss": 6.1573,
+      "step": 79550
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0543239143755464e-05,
+      "loss": 6.3729,
+      "step": 79600
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0493013631204108e-05,
+      "loss": 6.1918,
+      "step": 79650
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0442788118652752e-05,
+      "loss": 6.7536,
+      "step": 79700
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0392562606101396e-05,
+      "loss": 5.7906,
+      "step": 79750
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.034233709355004e-05,
+      "loss": 6.3394,
+      "step": 79800
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0292111580998684e-05,
+      "loss": 5.8182,
+      "step": 79850
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0241886068447328e-05,
+      "loss": 5.8381,
+      "step": 79900
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0191660555895976e-05,
+      "loss": 6.4559,
+      "step": 79950
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0141435043344616e-05,
+      "loss": 5.5812,
+      "step": 80000
+    },
+    {
+      "epoch": 5.9,
+      "eval_loss": 7.74003791809082,
+      "eval_runtime": 967.3759,
+      "eval_samples_per_second": 13.54,
+      "eval_steps_per_second": 3.385,
+      "eval_wer": 0.19946663306874973,
+      "step": 80000
+    }
+  ],
+  "max_steps": 100051,
+  "num_train_epochs": 8,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}