diff --git "a/checkpoint-40000/trainer_state.json" "b/checkpoint-40000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-40000/trainer_state.json"
@@ -0,0 +1,4834 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.950505274028177,
+  "global_step": 40000,
+  "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
+    }
+  ],
+  "max_steps": 100051,
+  "num_train_epochs": 8,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}